diff --git "a/checkpoint-21604/trainer_state.json" "b/checkpoint-21604/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/checkpoint-21604/trainer_state.json"
@@ -0,0 +1,129681 @@
+{
+  "best_metric": 0.828673779964447,
+  "best_model_checkpoint": "./bert_clf_results/checkpoint-10803",
+  "epoch": 3.999629732481718,
+  "eval_steps": 50,
+  "global_step": 21604,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.3137436372049978e-08,
+      "loss": 1.6168,
+      "step": 1
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.6274872744099956e-08,
+      "loss": 1.6196,
+      "step": 2
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.941230911614993e-08,
+      "loss": 1.5701,
+      "step": 3
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.254974548819991e-08,
+      "loss": 1.5737,
+      "step": 4
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.1568718186024988e-07,
+      "loss": 1.6343,
+      "step": 5
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.3882461823229986e-07,
+      "loss": 1.589,
+      "step": 6
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.6196205460434984e-07,
+      "loss": 1.6121,
+      "step": 7
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.8509949097639983e-07,
+      "loss": 1.5986,
+      "step": 8
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.0823692734844978e-07,
+      "loss": 1.6146,
+      "step": 9
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.3137436372049976e-07,
+      "loss": 1.6132,
+      "step": 10
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.5451180009254977e-07,
+      "loss": 1.5957,
+      "step": 11
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.776492364645997e-07,
+      "loss": 1.5981,
+      "step": 12
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.0078667283664973e-07,
+      "loss": 1.5962,
+      "step": 13
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.239241092086997e-07,
+      "loss": 1.5841,
+      "step": 14
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.470615455807497e-07,
+      "loss": 1.6099,
+      "step": 15
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.7019898195279965e-07,
+      "loss": 1.6007,
+      "step": 16
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.9333641832484966e-07,
+      "loss": 1.606,
+      "step": 17
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.1647385469689956e-07,
+      "loss": 1.6054,
+      "step": 18
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.396112910689496e-07,
+      "loss": 1.5857,
+      "step": 19
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.627487274409995e-07,
+      "loss": 1.5752,
+      "step": 20
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.858861638130495e-07,
+      "loss": 1.6227,
+      "step": 21
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.090236001850995e-07,
+      "loss": 1.5897,
+      "step": 22
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.321610365571495e-07,
+      "loss": 1.5651,
+      "step": 23
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.552984729291994e-07,
+      "loss": 1.592,
+      "step": 24
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.784359093012495e-07,
+      "loss": 1.5798,
+      "step": 25
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.015733456732995e-07,
+      "loss": 1.573,
+      "step": 26
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.247107820453494e-07,
+      "loss": 1.5848,
+      "step": 27
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.478482184173994e-07,
+      "loss": 1.5586,
+      "step": 28
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.709856547894493e-07,
+      "loss": 1.5904,
+      "step": 29
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.941230911614994e-07,
+      "loss": 1.6061,
+      "step": 30
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.172605275335493e-07,
+      "loss": 1.5773,
+      "step": 31
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.403979639055993e-07,
+      "loss": 1.59,
+      "step": 32
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.635354002776492e-07,
+      "loss": 1.5862,
+      "step": 33
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.866728366496993e-07,
+      "loss": 1.5767,
+      "step": 34
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.098102730217492e-07,
+      "loss": 1.583,
+      "step": 35
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.329477093937991e-07,
+      "loss": 1.5734,
+      "step": 36
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.560851457658492e-07,
+      "loss": 1.5486,
+      "step": 37
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.792225821378992e-07,
+      "loss": 1.6086,
+      "step": 38
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.023600185099491e-07,
+      "loss": 1.5826,
+      "step": 39
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.25497454881999e-07,
+      "loss": 1.5789,
+      "step": 40
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.486348912540491e-07,
+      "loss": 1.5839,
+      "step": 41
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.71772327626099e-07,
+      "loss": 1.5762,
+      "step": 42
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.94909763998149e-07,
+      "loss": 1.5435,
+      "step": 43
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.018047200370199e-06,
+      "loss": 1.5782,
+      "step": 44
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.0411846367422489e-06,
+      "loss": 1.5961,
+      "step": 45
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.064322073114299e-06,
+      "loss": 1.574,
+      "step": 46
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.0874595094863489e-06,
+      "loss": 1.5487,
+      "step": 47
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.110596945858399e-06,
+      "loss": 1.5678,
+      "step": 48
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.133734382230449e-06,
+      "loss": 1.5507,
+      "step": 49
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.156871818602499e-06,
+      "loss": 1.5607,
+      "step": 50
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.180009254974549e-06,
+      "loss": 1.563,
+      "step": 51
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.203146691346599e-06,
+      "loss": 1.5827,
+      "step": 52
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.2262841277186487e-06,
+      "loss": 1.5422,
+      "step": 53
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.2494215640906987e-06,
+      "loss": 1.5712,
+      "step": 54
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.2725590004627487e-06,
+      "loss": 1.5183,
+      "step": 55
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.2956964368347988e-06,
+      "loss": 1.5499,
+      "step": 56
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.3188338732068488e-06,
+      "loss": 1.5274,
+      "step": 57
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.3419713095788986e-06,
+      "loss": 1.5098,
+      "step": 58
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.3651087459509488e-06,
+      "loss": 1.5415,
+      "step": 59
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.3882461823229988e-06,
+      "loss": 1.5304,
+      "step": 60
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.4113836186950486e-06,
+      "loss": 1.5043,
+      "step": 61
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.4345210550670986e-06,
+      "loss": 1.4905,
+      "step": 62
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.4576584914391486e-06,
+      "loss": 1.5124,
+      "step": 63
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.4807959278111986e-06,
+      "loss": 1.5448,
+      "step": 64
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.5039333641832486e-06,
+      "loss": 1.5475,
+      "step": 65
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.5270708005552984e-06,
+      "loss": 1.5293,
+      "step": 66
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.5502082369273484e-06,
+      "loss": 1.4874,
+      "step": 67
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.5733456732993986e-06,
+      "loss": 1.5267,
+      "step": 68
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.5964831096714482e-06,
+      "loss": 1.566,
+      "step": 69
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.6196205460434984e-06,
+      "loss": 1.4971,
+      "step": 70
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.6427579824155484e-06,
+      "loss": 1.5233,
+      "step": 71
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.6658954187875982e-06,
+      "loss": 1.486,
+      "step": 72
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.6890328551596485e-06,
+      "loss": 1.4889,
+      "step": 73
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.7121702915316985e-06,
+      "loss": 1.4971,
+      "step": 74
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.7353077279037483e-06,
+      "loss": 1.496,
+      "step": 75
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.7584451642757985e-06,
+      "loss": 1.4555,
+      "step": 76
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.781582600647848e-06,
+      "loss": 1.5348,
+      "step": 77
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.8047200370198983e-06,
+      "loss": 1.4634,
+      "step": 78
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.8278574733919483e-06,
+      "loss": 1.4585,
+      "step": 79
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.850994909763998e-06,
+      "loss": 1.5008,
+      "step": 80
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.8741323461360483e-06,
+      "loss": 1.5305,
+      "step": 81
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.8972697825080981e-06,
+      "loss": 1.4243,
+      "step": 82
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.920407218880148e-06,
+      "loss": 1.4903,
+      "step": 83
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.943544655252198e-06,
+      "loss": 1.4399,
+      "step": 84
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.966682091624248e-06,
+      "loss": 1.462,
+      "step": 85
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.989819527996298e-06,
+      "loss": 1.4689,
+      "step": 86
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.012956964368348e-06,
+      "loss": 1.4162,
+      "step": 87
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.036094400740398e-06,
+      "loss": 1.4556,
+      "step": 88
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.059231837112448e-06,
+      "loss": 1.4297,
+      "step": 89
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.0823692734844978e-06,
+      "loss": 1.4418,
+      "step": 90
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.105506709856548e-06,
+      "loss": 1.4154,
+      "step": 91
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.128644146228598e-06,
+      "loss": 1.4469,
+      "step": 92
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.1517815826006478e-06,
+      "loss": 1.332,
+      "step": 93
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.1749190189726978e-06,
+      "loss": 1.4486,
+      "step": 94
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.198056455344748e-06,
+      "loss": 1.4416,
+      "step": 95
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.221193891716798e-06,
+      "loss": 1.3474,
+      "step": 96
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.244331328088848e-06,
+      "loss": 1.4622,
+      "step": 97
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.267468764460898e-06,
+      "loss": 1.4207,
+      "step": 98
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.290606200832948e-06,
+      "loss": 1.4108,
+      "step": 99
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.313743637204998e-06,
+      "loss": 1.3578,
+      "step": 100
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.336881073577048e-06,
+      "loss": 1.3763,
+      "step": 101
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.360018509949098e-06,
+      "loss": 1.3235,
+      "step": 102
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.3831559463211474e-06,
+      "loss": 1.4353,
+      "step": 103
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.406293382693198e-06,
+      "loss": 1.4328,
+      "step": 104
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.429430819065248e-06,
+      "loss": 1.4679,
+      "step": 105
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.4525682554372975e-06,
+      "loss": 1.4324,
+      "step": 106
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.475705691809348e-06,
+      "loss": 1.4037,
+      "step": 107
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.4988431281813975e-06,
+      "loss": 1.3405,
+      "step": 108
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.5219805645534475e-06,
+      "loss": 1.3909,
+      "step": 109
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.5451180009254975e-06,
+      "loss": 1.3437,
+      "step": 110
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.5682554372975475e-06,
+      "loss": 1.4072,
+      "step": 111
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.5913928736695975e-06,
+      "loss": 1.3131,
+      "step": 112
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.6145303100416475e-06,
+      "loss": 1.2467,
+      "step": 113
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.6376677464136975e-06,
+      "loss": 1.4106,
+      "step": 114
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.6608051827857475e-06,
+      "loss": 1.2921,
+      "step": 115
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.683942619157797e-06,
+      "loss": 1.3257,
+      "step": 116
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.7070800555298475e-06,
+      "loss": 1.2759,
+      "step": 117
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.7302174919018976e-06,
+      "loss": 1.2046,
+      "step": 118
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.753354928273947e-06,
+      "loss": 1.291,
+      "step": 119
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.7764923646459976e-06,
+      "loss": 1.299,
+      "step": 120
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.799629801018047e-06,
+      "loss": 1.4075,
+      "step": 121
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.822767237390097e-06,
+      "loss": 1.2993,
+      "step": 122
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.8459046737621476e-06,
+      "loss": 1.2675,
+      "step": 123
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.869042110134197e-06,
+      "loss": 1.1282,
+      "step": 124
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.892179546506247e-06,
+      "loss": 1.1416,
+      "step": 125
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.915316982878297e-06,
+      "loss": 1.2935,
+      "step": 126
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.938454419250347e-06,
+      "loss": 1.5334,
+      "step": 127
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.961591855622397e-06,
+      "loss": 1.2523,
+      "step": 128
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.9847292919944472e-06,
+      "loss": 1.3841,
+      "step": 129
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.0078667283664972e-06,
+      "loss": 1.1407,
+      "step": 130
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.0310041647385472e-06,
+      "loss": 1.073,
+      "step": 131
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.054141601110597e-06,
+      "loss": 1.4068,
+      "step": 132
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.0772790374826472e-06,
+      "loss": 1.3205,
+      "step": 133
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.100416473854697e-06,
+      "loss": 1.0647,
+      "step": 134
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.123553910226747e-06,
+      "loss": 1.2132,
+      "step": 135
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.1466913465987973e-06,
+      "loss": 1.1658,
+      "step": 136
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.169828782970847e-06,
+      "loss": 1.4911,
+      "step": 137
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.1929662193428964e-06,
+      "loss": 1.1361,
+      "step": 138
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.2161036557149473e-06,
+      "loss": 1.1047,
+      "step": 139
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.239241092086997e-06,
+      "loss": 1.1923,
+      "step": 140
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.2623785284590465e-06,
+      "loss": 1.0872,
+      "step": 141
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.285515964831097e-06,
+      "loss": 1.3475,
+      "step": 142
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.308653401203147e-06,
+      "loss": 1.0905,
+      "step": 143
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.3317908375751965e-06,
+      "loss": 1.304,
+      "step": 144
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.354928273947247e-06,
+      "loss": 1.3744,
+      "step": 145
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.378065710319297e-06,
+      "loss": 1.0725,
+      "step": 146
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.4012031466913465e-06,
+      "loss": 1.2729,
+      "step": 147
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.424340583063397e-06,
+      "loss": 1.2745,
+      "step": 148
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.447478019435447e-06,
+      "loss": 1.3304,
+      "step": 149
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.4706154558074965e-06,
+      "loss": 1.0841,
+      "step": 150
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.4937528921795465e-06,
+      "loss": 1.3747,
+      "step": 151
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.516890328551597e-06,
+      "loss": 1.4608,
+      "step": 152
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.5400277649236466e-06,
+      "loss": 1.1696,
+      "step": 153
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.563165201295696e-06,
+      "loss": 1.3045,
+      "step": 154
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.586302637667747e-06,
+      "loss": 1.1317,
+      "step": 155
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.6094400740397966e-06,
+      "loss": 0.9296,
+      "step": 156
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.632577510411846e-06,
+      "loss": 1.3942,
+      "step": 157
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.6557149467838966e-06,
+      "loss": 1.1296,
+      "step": 158
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.6788523831559466e-06,
+      "loss": 1.0384,
+      "step": 159
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.701989819527996e-06,
+      "loss": 1.0651,
+      "step": 160
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.7251272559000466e-06,
+      "loss": 0.8654,
+      "step": 161
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.7482646922720966e-06,
+      "loss": 1.0588,
+      "step": 162
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.7714021286441462e-06,
+      "loss": 0.8931,
+      "step": 163
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.7945395650161962e-06,
+      "loss": 1.2863,
+      "step": 164
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.817677001388247e-06,
+      "loss": 1.3769,
+      "step": 165
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.840814437760296e-06,
+      "loss": 1.1978,
+      "step": 166
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.863951874132346e-06,
+      "loss": 1.1207,
+      "step": 167
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.887089310504396e-06,
+      "loss": 1.0075,
+      "step": 168
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.910226746876447e-06,
+      "loss": 1.0253,
+      "step": 169
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.933364183248496e-06,
+      "loss": 1.1246,
+      "step": 170
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.956501619620547e-06,
+      "loss": 1.2042,
+      "step": 171
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.979639055992596e-06,
+      "loss": 1.4377,
+      "step": 172
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.002776492364646e-06,
+      "loss": 1.2121,
+      "step": 173
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.025913928736696e-06,
+      "loss": 1.0497,
+      "step": 174
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.049051365108746e-06,
+      "loss": 1.329,
+      "step": 175
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.072188801480796e-06,
+      "loss": 0.9537,
+      "step": 176
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.095326237852846e-06,
+      "loss": 1.1713,
+      "step": 177
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.118463674224896e-06,
+      "loss": 1.0067,
+      "step": 178
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.141601110596946e-06,
+      "loss": 1.2178,
+      "step": 179
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.1647385469689955e-06,
+      "loss": 1.2996,
+      "step": 180
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.187875983341046e-06,
+      "loss": 1.1532,
+      "step": 181
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.211013419713096e-06,
+      "loss": 0.9791,
+      "step": 182
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.234150856085146e-06,
+      "loss": 1.2418,
+      "step": 183
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.257288292457196e-06,
+      "loss": 0.8282,
+      "step": 184
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.280425728829246e-06,
+      "loss": 1.0467,
+      "step": 185
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.3035631652012956e-06,
+      "loss": 1.0072,
+      "step": 186
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.326700601573346e-06,
+      "loss": 1.1077,
+      "step": 187
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.3498380379453956e-06,
+      "loss": 1.0178,
+      "step": 188
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.372975474317446e-06,
+      "loss": 1.0905,
+      "step": 189
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.396112910689496e-06,
+      "loss": 0.9498,
+      "step": 190
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.419250347061546e-06,
+      "loss": 0.9839,
+      "step": 191
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.442387783433596e-06,
+      "loss": 1.5259,
+      "step": 192
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.465525219805645e-06,
+      "loss": 1.3288,
+      "step": 193
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.488662656177696e-06,
+      "loss": 0.8782,
+      "step": 194
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.511800092549746e-06,
+      "loss": 1.4033,
+      "step": 195
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.534937528921796e-06,
+      "loss": 1.1533,
+      "step": 196
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.558074965293846e-06,
+      "loss": 1.1287,
+      "step": 197
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.581212401665896e-06,
+      "loss": 1.2555,
+      "step": 198
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.604349838037945e-06,
+      "loss": 1.2434,
+      "step": 199
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.627487274409996e-06,
+      "loss": 1.307,
+      "step": 200
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.650624710782046e-06,
+      "loss": 1.0619,
+      "step": 201
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.673762147154096e-06,
+      "loss": 1.1504,
+      "step": 202
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.696899583526145e-06,
+      "loss": 0.8354,
+      "step": 203
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.720037019898196e-06,
+      "loss": 0.9128,
+      "step": 204
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.743174456270245e-06,
+      "loss": 1.1187,
+      "step": 205
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.766311892642295e-06,
+      "loss": 0.9669,
+      "step": 206
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.789449329014345e-06,
+      "loss": 1.2645,
+      "step": 207
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.812586765386396e-06,
+      "loss": 1.175,
+      "step": 208
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.835724201758445e-06,
+      "loss": 1.3559,
+      "step": 209
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.858861638130496e-06,
+      "loss": 1.1424,
+      "step": 210
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.881999074502545e-06,
+      "loss": 0.9844,
+      "step": 211
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.905136510874595e-06,
+      "loss": 1.068,
+      "step": 212
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.928273947246645e-06,
+      "loss": 1.4047,
+      "step": 213
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.951411383618696e-06,
+      "loss": 1.1444,
+      "step": 214
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.974548819990745e-06,
+      "loss": 1.0875,
+      "step": 215
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.997686256362795e-06,
+      "loss": 1.1708,
+      "step": 216
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.020823692734845e-06,
+      "loss": 0.9237,
+      "step": 217
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.043961129106895e-06,
+      "loss": 0.9595,
+      "step": 218
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.0670985654789445e-06,
+      "loss": 1.2414,
+      "step": 219
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.090236001850995e-06,
+      "loss": 1.1046,
+      "step": 220
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.113373438223045e-06,
+      "loss": 0.8773,
+      "step": 221
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.136510874595095e-06,
+      "loss": 0.8709,
+      "step": 222
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.159648310967145e-06,
+      "loss": 0.9741,
+      "step": 223
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.182785747339195e-06,
+      "loss": 1.0956,
+      "step": 224
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.205923183711245e-06,
+      "loss": 1.4146,
+      "step": 225
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.229060620083295e-06,
+      "loss": 1.2276,
+      "step": 226
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.2521980564553455e-06,
+      "loss": 1.1406,
+      "step": 227
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.275335492827395e-06,
+      "loss": 1.2461,
+      "step": 228
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.298472929199445e-06,
+      "loss": 1.2654,
+      "step": 229
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.321610365571495e-06,
+      "loss": 1.0659,
+      "step": 230
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.344747801943545e-06,
+      "loss": 1.0843,
+      "step": 231
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.367885238315594e-06,
+      "loss": 1.1341,
+      "step": 232
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.391022674687645e-06,
+      "loss": 1.0897,
+      "step": 233
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.414160111059695e-06,
+      "loss": 0.9808,
+      "step": 234
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.437297547431745e-06,
+      "loss": 1.1313,
+      "step": 235
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.460434983803795e-06,
+      "loss": 0.7079,
+      "step": 236
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.483572420175845e-06,
+      "loss": 1.0893,
+      "step": 237
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.506709856547894e-06,
+      "loss": 1.2033,
+      "step": 238
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.529847292919945e-06,
+      "loss": 0.9488,
+      "step": 239
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.552984729291995e-06,
+      "loss": 0.9034,
+      "step": 240
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.576122165664045e-06,
+      "loss": 1.059,
+      "step": 241
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.599259602036094e-06,
+      "loss": 1.1509,
+      "step": 242
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.622397038408145e-06,
+      "loss": 1.3517,
+      "step": 243
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.645534474780194e-06,
+      "loss": 1.1454,
+      "step": 244
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.668671911152245e-06,
+      "loss": 0.9655,
+      "step": 245
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.691809347524295e-06,
+      "loss": 0.8914,
+      "step": 246
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.714946783896345e-06,
+      "loss": 1.091,
+      "step": 247
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.738084220268394e-06,
+      "loss": 0.9608,
+      "step": 248
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.761221656640445e-06,
+      "loss": 1.3378,
+      "step": 249
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.784359093012494e-06,
+      "loss": 0.911,
+      "step": 250
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.807496529384544e-06,
+      "loss": 0.739,
+      "step": 251
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.830633965756594e-06,
+      "loss": 1.2666,
+      "step": 252
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.853771402128645e-06,
+      "loss": 0.7059,
+      "step": 253
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.876908838500694e-06,
+      "loss": 1.0488,
+      "step": 254
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.900046274872744e-06,
+      "loss": 0.9548,
+      "step": 255
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.923183711244794e-06,
+      "loss": 0.9575,
+      "step": 256
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.946321147616844e-06,
+      "loss": 1.279,
+      "step": 257
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.9694585839888944e-06,
+      "loss": 1.195,
+      "step": 258
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.992596020360945e-06,
+      "loss": 0.9757,
+      "step": 259
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 6.0157334567329944e-06,
+      "loss": 1.2013,
+      "step": 260
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 6.038870893105044e-06,
+      "loss": 0.9423,
+      "step": 261
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 6.0620083294770945e-06,
+      "loss": 0.7589,
+      "step": 262
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 6.085145765849144e-06,
+      "loss": 1.122,
+      "step": 263
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 6.108283202221194e-06,
+      "loss": 0.8647,
+      "step": 264
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 6.131420638593244e-06,
+      "loss": 0.9681,
+      "step": 265
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 6.1545580749652945e-06,
+      "loss": 0.9987,
+      "step": 266
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 6.177695511337344e-06,
+      "loss": 1.102,
+      "step": 267
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 6.200832947709394e-06,
+      "loss": 1.201,
+      "step": 268
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 6.223970384081444e-06,
+      "loss": 0.9598,
+      "step": 269
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 6.247107820453494e-06,
+      "loss": 0.9705,
+      "step": 270
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 6.270245256825544e-06,
+      "loss": 1.1066,
+      "step": 271
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 6.2933826931975945e-06,
+      "loss": 0.9859,
+      "step": 272
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 6.316520129569645e-06,
+      "loss": 1.0215,
+      "step": 273
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 6.339657565941694e-06,
+      "loss": 0.9498,
+      "step": 274
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 6.362795002313744e-06,
+      "loss": 1.0404,
+      "step": 275
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 6.385932438685793e-06,
+      "loss": 0.8756,
+      "step": 276
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 6.409069875057843e-06,
+      "loss": 1.1386,
+      "step": 277
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 6.432207311429895e-06,
+      "loss": 0.9924,
+      "step": 278
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 6.455344747801943e-06,
+      "loss": 0.9757,
+      "step": 279
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 6.478482184173994e-06,
+      "loss": 1.0732,
+      "step": 280
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 6.501619620546044e-06,
+      "loss": 0.8586,
+      "step": 281
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 6.524757056918093e-06,
+      "loss": 0.7227,
+      "step": 282
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 6.547894493290143e-06,
+      "loss": 0.9605,
+      "step": 283
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 6.571031929662194e-06,
+      "loss": 1.0546,
+      "step": 284
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 6.594169366034243e-06,
+      "loss": 1.2566,
+      "step": 285
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 6.617306802406294e-06,
+      "loss": 1.2027,
+      "step": 286
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 6.640444238778344e-06,
+      "loss": 1.01,
+      "step": 287
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 6.663581675150393e-06,
+      "loss": 1.0104,
+      "step": 288
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 6.686719111522443e-06,
+      "loss": 0.8356,
+      "step": 289
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 6.709856547894494e-06,
+      "loss": 1.1121,
+      "step": 290
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 6.732993984266543e-06,
+      "loss": 0.6995,
+      "step": 291
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 6.756131420638594e-06,
+      "loss": 1.0911,
+      "step": 292
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 6.779268857010644e-06,
+      "loss": 1.1742,
+      "step": 293
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 6.802406293382693e-06,
+      "loss": 0.8072,
+      "step": 294
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 6.8255437297547435e-06,
+      "loss": 0.986,
+      "step": 295
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 6.848681166126794e-06,
+      "loss": 0.8473,
+      "step": 296
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 6.8718186024988435e-06,
+      "loss": 0.9369,
+      "step": 297
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 6.894956038870894e-06,
+      "loss": 1.1897,
+      "step": 298
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 6.918093475242944e-06,
+      "loss": 0.7412,
+      "step": 299
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 6.941230911614993e-06,
+      "loss": 1.1456,
+      "step": 300
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 6.9643683479870435e-06,
+      "loss": 0.8921,
+      "step": 301
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 6.987505784359093e-06,
+      "loss": 1.0807,
+      "step": 302
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 7.0106432207311435e-06,
+      "loss": 1.0221,
+      "step": 303
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 7.033780657103194e-06,
+      "loss": 0.8869,
+      "step": 304
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 7.056918093475243e-06,
+      "loss": 1.2423,
+      "step": 305
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 7.080055529847293e-06,
+      "loss": 0.973,
+      "step": 306
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 7.1031929662193435e-06,
+      "loss": 1.1119,
+      "step": 307
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 7.126330402591392e-06,
+      "loss": 0.7567,
+      "step": 308
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 7.149467838963443e-06,
+      "loss": 1.0483,
+      "step": 309
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 7.172605275335494e-06,
+      "loss": 0.7977,
+      "step": 310
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 7.195742711707543e-06,
+      "loss": 1.0775,
+      "step": 311
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 7.218880148079593e-06,
+      "loss": 1.1649,
+      "step": 312
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 7.242017584451644e-06,
+      "loss": 0.92,
+      "step": 313
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 7.265155020823692e-06,
+      "loss": 1.1956,
+      "step": 314
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 7.288292457195743e-06,
+      "loss": 1.0916,
+      "step": 315
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 7.311429893567793e-06,
+      "loss": 0.8749,
+      "step": 316
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 7.334567329939843e-06,
+      "loss": 1.0233,
+      "step": 317
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 7.357704766311893e-06,
+      "loss": 1.2872,
+      "step": 318
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 7.380842202683944e-06,
+      "loss": 0.9862,
+      "step": 319
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 7.403979639055992e-06,
+      "loss": 1.1555,
+      "step": 320
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 7.427117075428043e-06,
+      "loss": 1.0549,
+      "step": 321
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 7.450254511800093e-06,
+      "loss": 1.0768,
+      "step": 322
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 7.473391948172143e-06,
+      "loss": 0.8135,
+      "step": 323
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 7.496529384544193e-06,
+      "loss": 1.0621,
+      "step": 324
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 7.519666820916244e-06,
+      "loss": 0.9901,
+      "step": 325
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 7.5428042572882924e-06,
+      "loss": 0.9674,
+      "step": 326
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 7.565941693660343e-06,
+      "loss": 1.1409,
+      "step": 327
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 7.5890791300323924e-06,
+      "loss": 0.821,
+      "step": 328
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 7.612216566404443e-06,
+      "loss": 1.0255,
+      "step": 329
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 7.635354002776493e-06,
+      "loss": 0.969,
+      "step": 330
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 7.658491439148543e-06,
+      "loss": 0.5942,
+      "step": 331
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 7.681628875520592e-06,
+      "loss": 1.1231,
+      "step": 332
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 7.704766311892644e-06,
+      "loss": 1.0718,
+      "step": 333
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 7.727903748264692e-06,
+      "loss": 0.9198,
+      "step": 334
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 7.751041184636743e-06,
+      "loss": 1.3495,
+      "step": 335
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 7.774178621008793e-06,
+      "loss": 0.8871,
+      "step": 336
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 7.797316057380842e-06,
+      "loss": 0.8504,
+      "step": 337
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 7.820453493752893e-06,
+      "loss": 0.8492,
+      "step": 338
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 7.843590930124943e-06,
+      "loss": 0.8439,
+      "step": 339
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 7.866728366496993e-06,
+      "loss": 0.9298,
+      "step": 340
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 7.889865802869042e-06,
+      "loss": 1.0891,
+      "step": 341
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 7.913003239241093e-06,
+      "loss": 0.8102,
+      "step": 342
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 7.936140675613141e-06,
+      "loss": 1.1947,
+      "step": 343
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 7.959278111985193e-06,
+      "loss": 1.0314,
+      "step": 344
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 7.982415548357242e-06,
+      "loss": 0.899,
+      "step": 345
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 8.005552984729292e-06,
+      "loss": 0.6742,
+      "step": 346
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 8.028690421101343e-06,
+      "loss": 0.9081,
+      "step": 347
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 8.051827857473393e-06,
+      "loss": 0.8232,
+      "step": 348
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 8.074965293845442e-06,
+      "loss": 0.9419,
+      "step": 349
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 8.098102730217492e-06,
+      "loss": 1.0544,
+      "step": 350
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 8.121240166589543e-06,
+      "loss": 1.2408,
+      "step": 351
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 8.144377602961593e-06,
+      "loss": 0.8061,
+      "step": 352
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 8.167515039333642e-06,
+      "loss": 0.6625,
+      "step": 353
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 8.190652475705692e-06,
+      "loss": 1.164,
+      "step": 354
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 8.213789912077741e-06,
+      "loss": 1.1102,
+      "step": 355
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 8.236927348449793e-06,
+      "loss": 0.7567,
+      "step": 356
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 8.260064784821842e-06,
+      "loss": 0.6873,
+      "step": 357
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 8.283202221193892e-06,
+      "loss": 1.2124,
+      "step": 358
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 8.306339657565943e-06,
+      "loss": 1.0918,
+      "step": 359
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 8.329477093937991e-06,
+      "loss": 1.0605,
+      "step": 360
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 8.352614530310042e-06,
+      "loss": 0.9795,
+      "step": 361
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 8.375751966682092e-06,
+      "loss": 0.6297,
+      "step": 362
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 8.398889403054141e-06,
+      "loss": 1.0195,
+      "step": 363
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 8.422026839426193e-06,
+      "loss": 0.7728,
+      "step": 364
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 8.445164275798242e-06,
+      "loss": 0.983,
+      "step": 365
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 8.468301712170292e-06,
+      "loss": 1.2131,
+      "step": 366
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 8.491439148542341e-06,
+      "loss": 1.2354,
+      "step": 367
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 8.514576584914393e-06,
+      "loss": 1.1086,
+      "step": 368
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 8.53771402128644e-06,
+      "loss": 0.797,
+      "step": 369
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 8.560851457658492e-06,
+      "loss": 1.0107,
+      "step": 370
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 8.583988894030543e-06,
+      "loss": 0.8861,
+      "step": 371
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 8.607126330402591e-06,
+      "loss": 0.9202,
+      "step": 372
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 8.630263766774642e-06,
+      "loss": 0.9691,
+      "step": 373
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 8.653401203146692e-06,
+      "loss": 0.875,
+      "step": 374
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 8.676538639518742e-06,
+      "loss": 1.147,
+      "step": 375
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 8.699676075890791e-06,
+      "loss": 0.9208,
+      "step": 376
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 8.722813512262842e-06,
+      "loss": 0.8437,
+      "step": 377
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 8.745950948634892e-06,
+      "loss": 0.8104,
+      "step": 378
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 8.769088385006942e-06,
+      "loss": 0.7666,
+      "step": 379
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 8.792225821378991e-06,
+      "loss": 0.6819,
+      "step": 380
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 8.81536325775104e-06,
+      "loss": 0.8213,
+      "step": 381
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 8.838500694123092e-06,
+      "loss": 0.5924,
+      "step": 382
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 8.861638130495142e-06,
+      "loss": 1.1096,
+      "step": 383
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 8.884775566867191e-06,
+      "loss": 1.1299,
+      "step": 384
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 8.907913003239242e-06,
+      "loss": 1.0653,
+      "step": 385
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 8.93105043961129e-06,
+      "loss": 0.8944,
+      "step": 386
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 8.954187875983342e-06,
+      "loss": 0.8022,
+      "step": 387
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 8.977325312355391e-06,
+      "loss": 0.8733,
+      "step": 388
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.00046274872744e-06,
+      "loss": 0.8776,
+      "step": 389
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.023600185099492e-06,
+      "loss": 0.8305,
+      "step": 390
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.046737621471542e-06,
+      "loss": 0.9756,
+      "step": 391
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.069875057843591e-06,
+      "loss": 0.9924,
+      "step": 392
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.09301249421564e-06,
+      "loss": 1.0119,
+      "step": 393
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.116149930587692e-06,
+      "loss": 0.8514,
+      "step": 394
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.139287366959742e-06,
+      "loss": 1.0101,
+      "step": 395
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.162424803331791e-06,
+      "loss": 0.9178,
+      "step": 396
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.185562239703843e-06,
+      "loss": 1.151,
+      "step": 397
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.20869967607589e-06,
+      "loss": 1.1629,
+      "step": 398
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.231837112447942e-06,
+      "loss": 0.9439,
+      "step": 399
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.254974548819991e-06,
+      "loss": 1.0994,
+      "step": 400
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.278111985192041e-06,
+      "loss": 1.0261,
+      "step": 401
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.301249421564092e-06,
+      "loss": 0.7376,
+      "step": 402
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.324386857936142e-06,
+      "loss": 0.9094,
+      "step": 403
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.347524294308191e-06,
+      "loss": 0.9773,
+      "step": 404
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.370661730680241e-06,
+      "loss": 0.9097,
+      "step": 405
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.39379916705229e-06,
+      "loss": 0.6853,
+      "step": 406
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.41693660342434e-06,
+      "loss": 0.985,
+      "step": 407
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.440074039796391e-06,
+      "loss": 0.7796,
+      "step": 408
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.463211476168441e-06,
+      "loss": 1.0122,
+      "step": 409
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.48634891254049e-06,
+      "loss": 1.0505,
+      "step": 410
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.509486348912542e-06,
+      "loss": 1.1896,
+      "step": 411
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.53262378528459e-06,
+      "loss": 0.875,
+      "step": 412
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.555761221656641e-06,
+      "loss": 0.9891,
+      "step": 413
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.57889865802869e-06,
+      "loss": 0.8847,
+      "step": 414
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.60203609440074e-06,
+      "loss": 1.0455,
+      "step": 415
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.625173530772791e-06,
+      "loss": 0.8417,
+      "step": 416
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.648310967144841e-06,
+      "loss": 1.1299,
+      "step": 417
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.67144840351689e-06,
+      "loss": 0.8169,
+      "step": 418
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.69458583988894e-06,
+      "loss": 1.0576,
+      "step": 419
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.717723276260991e-06,
+      "loss": 0.666,
+      "step": 420
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.740860712633041e-06,
+      "loss": 0.8677,
+      "step": 421
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.76399814900509e-06,
+      "loss": 0.8259,
+      "step": 422
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.787135585377142e-06,
+      "loss": 0.7872,
+      "step": 423
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.81027302174919e-06,
+      "loss": 1.3515,
+      "step": 424
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.833410458121241e-06,
+      "loss": 0.9359,
+      "step": 425
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.85654789449329e-06,
+      "loss": 0.8786,
+      "step": 426
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.87968533086534e-06,
+      "loss": 1.0664,
+      "step": 427
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.902822767237392e-06,
+      "loss": 0.6992,
+      "step": 428
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.925960203609441e-06,
+      "loss": 0.8896,
+      "step": 429
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.94909763998149e-06,
+      "loss": 0.9666,
+      "step": 430
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.97223507635354e-06,
+      "loss": 1.0299,
+      "step": 431
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.99537251272559e-06,
+      "loss": 0.8579,
+      "step": 432
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.001850994909764e-05,
+      "loss": 1.1955,
+      "step": 433
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.004164738546969e-05,
+      "loss": 1.0314,
+      "step": 434
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.006478482184174e-05,
+      "loss": 1.2409,
+      "step": 435
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.008792225821379e-05,
+      "loss": 0.7099,
+      "step": 436
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.0111059694585841e-05,
+      "loss": 0.7694,
+      "step": 437
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.0134197130957889e-05,
+      "loss": 0.6464,
+      "step": 438
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.015733456732994e-05,
+      "loss": 0.992,
+      "step": 439
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.018047200370199e-05,
+      "loss": 0.8482,
+      "step": 440
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.020360944007404e-05,
+      "loss": 0.7581,
+      "step": 441
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.022674687644609e-05,
+      "loss": 1.0007,
+      "step": 442
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.024988431281814e-05,
+      "loss": 0.7263,
+      "step": 443
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.027302174919019e-05,
+      "loss": 0.8218,
+      "step": 444
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.029615918556224e-05,
+      "loss": 0.8457,
+      "step": 445
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.031929662193429e-05,
+      "loss": 0.8838,
+      "step": 446
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.034243405830634e-05,
+      "loss": 1.2307,
+      "step": 447
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.036557149467839e-05,
+      "loss": 0.831,
+      "step": 448
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.0388708931050441e-05,
+      "loss": 1.2833,
+      "step": 449
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.041184636742249e-05,
+      "loss": 0.8781,
+      "step": 450
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.043498380379454e-05,
+      "loss": 1.2021,
+      "step": 451
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.045812124016659e-05,
+      "loss": 0.7778,
+      "step": 452
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.048125867653864e-05,
+      "loss": 0.754,
+      "step": 453
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.0504396112910691e-05,
+      "loss": 0.9399,
+      "step": 454
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.052753354928274e-05,
+      "loss": 0.7676,
+      "step": 455
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.055067098565479e-05,
+      "loss": 1.1569,
+      "step": 456
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.057380842202684e-05,
+      "loss": 0.8394,
+      "step": 457
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.059694585839889e-05,
+      "loss": 1.0346,
+      "step": 458
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.062008329477094e-05,
+      "loss": 1.0657,
+      "step": 459
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.064322073114299e-05,
+      "loss": 0.7984,
+      "step": 460
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.066635816751504e-05,
+      "loss": 0.9058,
+      "step": 461
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.068949560388709e-05,
+      "loss": 0.755,
+      "step": 462
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.071263304025914e-05,
+      "loss": 0.8462,
+      "step": 463
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.0735770476631188e-05,
+      "loss": 1.1406,
+      "step": 464
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.075890791300324e-05,
+      "loss": 1.1356,
+      "step": 465
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.078204534937529e-05,
+      "loss": 1.1298,
+      "step": 466
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.0805182785747339e-05,
+      "loss": 0.8863,
+      "step": 467
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.082832022211939e-05,
+      "loss": 0.8116,
+      "step": 468
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.085145765849144e-05,
+      "loss": 1.0697,
+      "step": 469
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.087459509486349e-05,
+      "loss": 0.7066,
+      "step": 470
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.0897732531235539e-05,
+      "loss": 0.9363,
+      "step": 471
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.092086996760759e-05,
+      "loss": 1.0091,
+      "step": 472
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.094400740397964e-05,
+      "loss": 1.176,
+      "step": 473
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.096714484035169e-05,
+      "loss": 0.95,
+      "step": 474
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.099028227672374e-05,
+      "loss": 0.9212,
+      "step": 475
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.1013419713095789e-05,
+      "loss": 0.8715,
+      "step": 476
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.103655714946784e-05,
+      "loss": 1.0183,
+      "step": 477
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.105969458583989e-05,
+      "loss": 0.8858,
+      "step": 478
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.1082832022211939e-05,
+      "loss": 1.0217,
+      "step": 479
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.110596945858399e-05,
+      "loss": 1.0078,
+      "step": 480
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.112910689495604e-05,
+      "loss": 1.0924,
+      "step": 481
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.115224433132809e-05,
+      "loss": 0.8452,
+      "step": 482
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.1175381767700139e-05,
+      "loss": 0.8138,
+      "step": 483
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.1198519204072189e-05,
+      "loss": 1.0654,
+      "step": 484
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.122165664044424e-05,
+      "loss": 0.967,
+      "step": 485
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.124479407681629e-05,
+      "loss": 0.9123,
+      "step": 486
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.1267931513188339e-05,
+      "loss": 0.7107,
+      "step": 487
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.1291068949560389e-05,
+      "loss": 1.1795,
+      "step": 488
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.131420638593244e-05,
+      "loss": 0.7771,
+      "step": 489
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.133734382230449e-05,
+      "loss": 0.9283,
+      "step": 490
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.1360481258676539e-05,
+      "loss": 1.1103,
+      "step": 491
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.138361869504859e-05,
+      "loss": 0.9516,
+      "step": 492
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.1406756131420638e-05,
+      "loss": 1.0757,
+      "step": 493
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.142989356779269e-05,
+      "loss": 1.337,
+      "step": 494
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.1453031004164739e-05,
+      "loss": 1.0169,
+      "step": 495
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.1476168440536789e-05,
+      "loss": 1.0275,
+      "step": 496
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.1499305876908838e-05,
+      "loss": 1.2685,
+      "step": 497
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.152244331328089e-05,
+      "loss": 0.7138,
+      "step": 498
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.1545580749652939e-05,
+      "loss": 1.0793,
+      "step": 499
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.1568718186024989e-05,
+      "loss": 0.9866,
+      "step": 500
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.159185562239704e-05,
+      "loss": 0.8242,
+      "step": 501
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.1614993058769088e-05,
+      "loss": 1.1762,
+      "step": 502
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.163813049514114e-05,
+      "loss": 0.8452,
+      "step": 503
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.1661267931513189e-05,
+      "loss": 0.9484,
+      "step": 504
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.1684405367885238e-05,
+      "loss": 0.9518,
+      "step": 505
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.170754280425729e-05,
+      "loss": 1.0721,
+      "step": 506
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.173068024062934e-05,
+      "loss": 1.0216,
+      "step": 507
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.1753817677001389e-05,
+      "loss": 0.7855,
+      "step": 508
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.1776955113373438e-05,
+      "loss": 0.6947,
+      "step": 509
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.1800092549745488e-05,
+      "loss": 0.9368,
+      "step": 510
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.182322998611754e-05,
+      "loss": 0.9753,
+      "step": 511
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.1846367422489589e-05,
+      "loss": 0.5313,
+      "step": 512
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.1869504858861638e-05,
+      "loss": 0.8976,
+      "step": 513
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.1892642295233688e-05,
+      "loss": 0.8306,
+      "step": 514
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.191577973160574e-05,
+      "loss": 0.9593,
+      "step": 515
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.1938917167977789e-05,
+      "loss": 0.8411,
+      "step": 516
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.1962054604349838e-05,
+      "loss": 1.2087,
+      "step": 517
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.198519204072189e-05,
+      "loss": 0.783,
+      "step": 518
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.2008329477093938e-05,
+      "loss": 1.1158,
+      "step": 519
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.2031466913465989e-05,
+      "loss": 0.7955,
+      "step": 520
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.2054604349838038e-05,
+      "loss": 1.0401,
+      "step": 521
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.2077741786210088e-05,
+      "loss": 1.0085,
+      "step": 522
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.210087922258214e-05,
+      "loss": 0.7719,
+      "step": 523
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.2124016658954189e-05,
+      "loss": 1.124,
+      "step": 524
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.2147154095326239e-05,
+      "loss": 0.8796,
+      "step": 525
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.2170291531698288e-05,
+      "loss": 1.2324,
+      "step": 526
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.219342896807034e-05,
+      "loss": 0.9056,
+      "step": 527
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.2216566404442387e-05,
+      "loss": 0.9919,
+      "step": 528
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.2239703840814439e-05,
+      "loss": 0.8899,
+      "step": 529
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.2262841277186488e-05,
+      "loss": 0.5921,
+      "step": 530
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.2285978713558538e-05,
+      "loss": 0.9693,
+      "step": 531
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.2309116149930589e-05,
+      "loss": 1.1627,
+      "step": 532
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.2332253586302639e-05,
+      "loss": 0.7926,
+      "step": 533
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.2355391022674688e-05,
+      "loss": 1.038,
+      "step": 534
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.2378528459046738e-05,
+      "loss": 0.8759,
+      "step": 535
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.2401665895418787e-05,
+      "loss": 0.8726,
+      "step": 536
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.2424803331790839e-05,
+      "loss": 1.1416,
+      "step": 537
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.2447940768162888e-05,
+      "loss": 0.8116,
+      "step": 538
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.2471078204534938e-05,
+      "loss": 0.7306,
+      "step": 539
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.2494215640906987e-05,
+      "loss": 1.0202,
+      "step": 540
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.2517353077279037e-05,
+      "loss": 0.7682,
+      "step": 541
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.2540490513651088e-05,
+      "loss": 0.655,
+      "step": 542
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.2563627950023138e-05,
+      "loss": 0.8306,
+      "step": 543
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.2586765386395189e-05,
+      "loss": 0.6581,
+      "step": 544
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.2609902822767239e-05,
+      "loss": 0.4832,
+      "step": 545
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.263304025913929e-05,
+      "loss": 1.0262,
+      "step": 546
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.2656177695511338e-05,
+      "loss": 0.8517,
+      "step": 547
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.2679315131883387e-05,
+      "loss": 1.0881,
+      "step": 548
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.2702452568255439e-05,
+      "loss": 0.7317,
+      "step": 549
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.2725590004627488e-05,
+      "loss": 0.7099,
+      "step": 550
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.274872744099954e-05,
+      "loss": 0.803,
+      "step": 551
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.2771864877371586e-05,
+      "loss": 0.9776,
+      "step": 552
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.2795002313743637e-05,
+      "loss": 0.9649,
+      "step": 553
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.2818139750115687e-05,
+      "loss": 0.8289,
+      "step": 554
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.2841277186487738e-05,
+      "loss": 0.8461,
+      "step": 555
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.286441462285979e-05,
+      "loss": 1.0355,
+      "step": 556
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.2887552059231839e-05,
+      "loss": 1.139,
+      "step": 557
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.2910689495603887e-05,
+      "loss": 1.0358,
+      "step": 558
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.2933826931975936e-05,
+      "loss": 0.8162,
+      "step": 559
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.2956964368347988e-05,
+      "loss": 1.1095,
+      "step": 560
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.2980101804720037e-05,
+      "loss": 0.7123,
+      "step": 561
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.3003239241092088e-05,
+      "loss": 0.7868,
+      "step": 562
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.302637667746414e-05,
+      "loss": 0.9727,
+      "step": 563
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.3049514113836186e-05,
+      "loss": 0.8077,
+      "step": 564
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.3072651550208237e-05,
+      "loss": 0.9401,
+      "step": 565
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.3095788986580287e-05,
+      "loss": 0.9121,
+      "step": 566
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.3118926422952338e-05,
+      "loss": 1.2319,
+      "step": 567
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.3142063859324388e-05,
+      "loss": 0.9234,
+      "step": 568
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.3165201295696439e-05,
+      "loss": 0.9561,
+      "step": 569
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.3188338732068487e-05,
+      "loss": 0.936,
+      "step": 570
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.3211476168440536e-05,
+      "loss": 0.5937,
+      "step": 571
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.3234613604812588e-05,
+      "loss": 1.03,
+      "step": 572
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.3257751041184637e-05,
+      "loss": 0.7958,
+      "step": 573
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.3280888477556688e-05,
+      "loss": 0.8842,
+      "step": 574
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.3304025913928736e-05,
+      "loss": 0.7796,
+      "step": 575
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.3327163350300786e-05,
+      "loss": 1.106,
+      "step": 576
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.3350300786672837e-05,
+      "loss": 0.8944,
+      "step": 577
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.3373438223044887e-05,
+      "loss": 0.8804,
+      "step": 578
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.3396575659416938e-05,
+      "loss": 0.8148,
+      "step": 579
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.3419713095788988e-05,
+      "loss": 1.0511,
+      "step": 580
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.3442850532161036e-05,
+      "loss": 0.9109,
+      "step": 581
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.3465987968533087e-05,
+      "loss": 0.9486,
+      "step": 582
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.3489125404905136e-05,
+      "loss": 0.9826,
+      "step": 583
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.3512262841277188e-05,
+      "loss": 1.1563,
+      "step": 584
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.3535400277649237e-05,
+      "loss": 0.7312,
+      "step": 585
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.3558537714021289e-05,
+      "loss": 0.9467,
+      "step": 586
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.3581675150393336e-05,
+      "loss": 1.1666,
+      "step": 587
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.3604812586765386e-05,
+      "loss": 1.0411,
+      "step": 588
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.3627950023137437e-05,
+      "loss": 0.6907,
+      "step": 589
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.3651087459509487e-05,
+      "loss": 0.8098,
+      "step": 590
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.3674224895881538e-05,
+      "loss": 0.7887,
+      "step": 591
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.3697362332253588e-05,
+      "loss": 1.0605,
+      "step": 592
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.3720499768625636e-05,
+      "loss": 0.7464,
+      "step": 593
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.3743637204997687e-05,
+      "loss": 0.7819,
+      "step": 594
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.3766774641369737e-05,
+      "loss": 0.9683,
+      "step": 595
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.3789912077741788e-05,
+      "loss": 0.5984,
+      "step": 596
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.3813049514113837e-05,
+      "loss": 1.0748,
+      "step": 597
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.3836186950485889e-05,
+      "loss": 0.7605,
+      "step": 598
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.3859324386857937e-05,
+      "loss": 0.8113,
+      "step": 599
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.3882461823229986e-05,
+      "loss": 0.9463,
+      "step": 600
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.3905599259602037e-05,
+      "loss": 0.7045,
+      "step": 601
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.3928736695974087e-05,
+      "loss": 0.7553,
+      "step": 602
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.3951874132346138e-05,
+      "loss": 1.1007,
+      "step": 603
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.3975011568718186e-05,
+      "loss": 1.0148,
+      "step": 604
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.3998149005090236e-05,
+      "loss": 0.867,
+      "step": 605
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.4021286441462287e-05,
+      "loss": 0.8186,
+      "step": 606
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.4044423877834337e-05,
+      "loss": 0.7441,
+      "step": 607
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.4067561314206388e-05,
+      "loss": 0.7961,
+      "step": 608
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.4090698750578437e-05,
+      "loss": 0.9641,
+      "step": 609
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.4113836186950485e-05,
+      "loss": 1.3233,
+      "step": 610
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.4136973623322537e-05,
+      "loss": 0.7685,
+      "step": 611
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.4160111059694586e-05,
+      "loss": 0.6832,
+      "step": 612
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.4183248496066638e-05,
+      "loss": 1.0236,
+      "step": 613
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.4206385932438687e-05,
+      "loss": 0.8594,
+      "step": 614
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.4229523368810738e-05,
+      "loss": 0.9228,
+      "step": 615
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.4252660805182785e-05,
+      "loss": 1.0539,
+      "step": 616
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.4275798241554836e-05,
+      "loss": 0.7472,
+      "step": 617
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.4298935677926885e-05,
+      "loss": 0.9027,
+      "step": 618
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.4322073114298937e-05,
+      "loss": 0.9871,
+      "step": 619
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.4345210550670988e-05,
+      "loss": 1.0479,
+      "step": 620
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.4368347987043038e-05,
+      "loss": 0.9964,
+      "step": 621
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.4391485423415085e-05,
+      "loss": 0.8687,
+      "step": 622
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.4414622859787135e-05,
+      "loss": 1.0386,
+      "step": 623
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.4437760296159186e-05,
+      "loss": 0.6078,
+      "step": 624
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.4460897732531236e-05,
+      "loss": 0.8005,
+      "step": 625
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.4484035168903287e-05,
+      "loss": 1.1851,
+      "step": 626
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.4507172605275335e-05,
+      "loss": 0.8887,
+      "step": 627
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.4530310041647385e-05,
+      "loss": 0.8574,
+      "step": 628
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.4553447478019436e-05,
+      "loss": 0.6695,
+      "step": 629
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.4576584914391486e-05,
+      "loss": 0.8159,
+      "step": 630
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.4599722350763537e-05,
+      "loss": 1.1722,
+      "step": 631
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.4622859787135586e-05,
+      "loss": 0.7752,
+      "step": 632
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.4645997223507634e-05,
+      "loss": 0.7326,
+      "step": 633
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.4669134659879686e-05,
+      "loss": 0.8585,
+      "step": 634
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.4692272096251735e-05,
+      "loss": 0.9279,
+      "step": 635
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.4715409532623786e-05,
+      "loss": 0.822,
+      "step": 636
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.4738546968995836e-05,
+      "loss": 0.8048,
+      "step": 637
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.4761684405367887e-05,
+      "loss": 0.6157,
+      "step": 638
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.4784821841739935e-05,
+      "loss": 0.7864,
+      "step": 639
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.4807959278111985e-05,
+      "loss": 0.8519,
+      "step": 640
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.4831096714484036e-05,
+      "loss": 0.6477,
+      "step": 641
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.4854234150856086e-05,
+      "loss": 0.9033,
+      "step": 642
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.4877371587228137e-05,
+      "loss": 0.9389,
+      "step": 643
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.4900509023600186e-05,
+      "loss": 0.9705,
+      "step": 644
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.4923646459972234e-05,
+      "loss": 0.9229,
+      "step": 645
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.4946783896344286e-05,
+      "loss": 0.9943,
+      "step": 646
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.4969921332716335e-05,
+      "loss": 0.9781,
+      "step": 647
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.4993058769088387e-05,
+      "loss": 1.1193,
+      "step": 648
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.5016196205460436e-05,
+      "loss": 0.9694,
+      "step": 649
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.5039333641832487e-05,
+      "loss": 1.0135,
+      "step": 650
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.5062471078204535e-05,
+      "loss": 0.7057,
+      "step": 651
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.5085608514576585e-05,
+      "loss": 1.1981,
+      "step": 652
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.5108745950948636e-05,
+      "loss": 0.6797,
+      "step": 653
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.5131883387320686e-05,
+      "loss": 0.8815,
+      "step": 654
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.5155020823692737e-05,
+      "loss": 1.0267,
+      "step": 655
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.5178158260064785e-05,
+      "loss": 0.8427,
+      "step": 656
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.5201295696436834e-05,
+      "loss": 1.2851,
+      "step": 657
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.5224433132808886e-05,
+      "loss": 0.7709,
+      "step": 658
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.5247570569180935e-05,
+      "loss": 0.8399,
+      "step": 659
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.5270708005552987e-05,
+      "loss": 0.7666,
+      "step": 660
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.5293845441925036e-05,
+      "loss": 0.7,
+      "step": 661
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.5316982878297086e-05,
+      "loss": 1.0955,
+      "step": 662
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.5340120314669135e-05,
+      "loss": 0.9426,
+      "step": 663
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.5363257751041185e-05,
+      "loss": 0.8161,
+      "step": 664
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.5386395187413235e-05,
+      "loss": 0.6339,
+      "step": 665
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.5409532623785288e-05,
+      "loss": 0.9608,
+      "step": 666
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.5432670060157337e-05,
+      "loss": 0.9277,
+      "step": 667
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.5455807496529383e-05,
+      "loss": 0.8596,
+      "step": 668
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.5478944932901436e-05,
+      "loss": 0.9157,
+      "step": 669
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.5502082369273486e-05,
+      "loss": 0.9271,
+      "step": 670
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.5525219805645535e-05,
+      "loss": 0.6873,
+      "step": 671
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.5548357242017585e-05,
+      "loss": 0.8245,
+      "step": 672
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.5571494678389638e-05,
+      "loss": 0.8824,
+      "step": 673
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.5594632114761684e-05,
+      "loss": 1.0254,
+      "step": 674
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.5617769551133734e-05,
+      "loss": 0.7843,
+      "step": 675
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.5640906987505787e-05,
+      "loss": 0.9246,
+      "step": 676
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.5664044423877836e-05,
+      "loss": 1.1929,
+      "step": 677
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.5687181860249886e-05,
+      "loss": 1.1556,
+      "step": 678
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.5710319296621932e-05,
+      "loss": 0.9944,
+      "step": 679
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.5733456732993985e-05,
+      "loss": 1.0149,
+      "step": 680
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.5756594169366035e-05,
+      "loss": 1.061,
+      "step": 681
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.5779731605738084e-05,
+      "loss": 0.8333,
+      "step": 682
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.5802869042110134e-05,
+      "loss": 1.0048,
+      "step": 683
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.5826006478482187e-05,
+      "loss": 0.9395,
+      "step": 684
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.5849143914854233e-05,
+      "loss": 0.749,
+      "step": 685
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.5872281351226283e-05,
+      "loss": 1.1399,
+      "step": 686
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.5895418787598336e-05,
+      "loss": 1.0073,
+      "step": 687
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.5918556223970385e-05,
+      "loss": 0.8437,
+      "step": 688
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.5941693660342435e-05,
+      "loss": 1.2253,
+      "step": 689
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.5964831096714484e-05,
+      "loss": 1.0558,
+      "step": 690
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.5987968533086534e-05,
+      "loss": 0.6828,
+      "step": 691
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.6011105969458583e-05,
+      "loss": 1.1633,
+      "step": 692
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.6034243405830633e-05,
+      "loss": 0.8118,
+      "step": 693
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.6057380842202686e-05,
+      "loss": 0.9587,
+      "step": 694
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.6080518278574736e-05,
+      "loss": 0.8644,
+      "step": 695
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.6103655714946785e-05,
+      "loss": 0.654,
+      "step": 696
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.6126793151318835e-05,
+      "loss": 0.8588,
+      "step": 697
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.6149930587690884e-05,
+      "loss": 1.1519,
+      "step": 698
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.6173068024062934e-05,
+      "loss": 1.1005,
+      "step": 699
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.6196205460434984e-05,
+      "loss": 0.9991,
+      "step": 700
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.6219342896807037e-05,
+      "loss": 0.9205,
+      "step": 701
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.6242480333179086e-05,
+      "loss": 0.9383,
+      "step": 702
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.6265617769551132e-05,
+      "loss": 1.0004,
+      "step": 703
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.6288755205923185e-05,
+      "loss": 1.0587,
+      "step": 704
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.6311892642295235e-05,
+      "loss": 0.6862,
+      "step": 705
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.6335030078667284e-05,
+      "loss": 0.7295,
+      "step": 706
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.6358167515039334e-05,
+      "loss": 0.8751,
+      "step": 707
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.6381304951411384e-05,
+      "loss": 0.9122,
+      "step": 708
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.6404442387783433e-05,
+      "loss": 0.8285,
+      "step": 709
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.6427579824155483e-05,
+      "loss": 0.9504,
+      "step": 710
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.6450717260527536e-05,
+      "loss": 0.7425,
+      "step": 711
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.6473854696899585e-05,
+      "loss": 0.6834,
+      "step": 712
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.6496992133271635e-05,
+      "loss": 1.3705,
+      "step": 713
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.6520129569643685e-05,
+      "loss": 0.8585,
+      "step": 714
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.6543267006015734e-05,
+      "loss": 0.9346,
+      "step": 715
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.6566404442387784e-05,
+      "loss": 1.1208,
+      "step": 716
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.6589541878759833e-05,
+      "loss": 0.8528,
+      "step": 717
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.6612679315131886e-05,
+      "loss": 0.9716,
+      "step": 718
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.6635816751503936e-05,
+      "loss": 0.8273,
+      "step": 719
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.6658954187875982e-05,
+      "loss": 1.0924,
+      "step": 720
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.6682091624248035e-05,
+      "loss": 1.0163,
+      "step": 721
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.6705229060620085e-05,
+      "loss": 0.8134,
+      "step": 722
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.6728366496992134e-05,
+      "loss": 0.9979,
+      "step": 723
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.6751503933364184e-05,
+      "loss": 0.8416,
+      "step": 724
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.6774641369736237e-05,
+      "loss": 1.2498,
+      "step": 725
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.6797778806108283e-05,
+      "loss": 1.1264,
+      "step": 726
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.6820916242480332e-05,
+      "loss": 0.9648,
+      "step": 727
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.6844053678852385e-05,
+      "loss": 0.8727,
+      "step": 728
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.6867191115224435e-05,
+      "loss": 0.6119,
+      "step": 729
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.6890328551596485e-05,
+      "loss": 0.8927,
+      "step": 730
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.6913465987968534e-05,
+      "loss": 0.7299,
+      "step": 731
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.6936603424340584e-05,
+      "loss": 0.9975,
+      "step": 732
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.6959740860712633e-05,
+      "loss": 1.0023,
+      "step": 733
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.6982878297084683e-05,
+      "loss": 1.0133,
+      "step": 734
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.7006015733456736e-05,
+      "loss": 0.7394,
+      "step": 735
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.7029153169828786e-05,
+      "loss": 0.8647,
+      "step": 736
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.7052290606200832e-05,
+      "loss": 1.0815,
+      "step": 737
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.707542804257288e-05,
+      "loss": 0.8579,
+      "step": 738
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.7098565478944934e-05,
+      "loss": 0.772,
+      "step": 739
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.7121702915316984e-05,
+      "loss": 0.8214,
+      "step": 740
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.7144840351689033e-05,
+      "loss": 0.8735,
+      "step": 741
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.7167977788061086e-05,
+      "loss": 0.8787,
+      "step": 742
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.7191115224433133e-05,
+      "loss": 1.057,
+      "step": 743
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.7214252660805182e-05,
+      "loss": 0.4968,
+      "step": 744
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.7237390097177232e-05,
+      "loss": 0.9796,
+      "step": 745
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.7260527533549285e-05,
+      "loss": 0.825,
+      "step": 746
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.7283664969921334e-05,
+      "loss": 0.8012,
+      "step": 747
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.7306802406293384e-05,
+      "loss": 1.1419,
+      "step": 748
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.7329939842665434e-05,
+      "loss": 1.0819,
+      "step": 749
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.7353077279037483e-05,
+      "loss": 0.7924,
+      "step": 750
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.7376214715409533e-05,
+      "loss": 0.7404,
+      "step": 751
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.7399352151781582e-05,
+      "loss": 0.8414,
+      "step": 752
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.7422489588153635e-05,
+      "loss": 1.1968,
+      "step": 753
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.7445627024525685e-05,
+      "loss": 0.7712,
+      "step": 754
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.746876446089773e-05,
+      "loss": 1.0146,
+      "step": 755
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.7491901897269784e-05,
+      "loss": 0.99,
+      "step": 756
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.7515039333641834e-05,
+      "loss": 0.8071,
+      "step": 757
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.7538176770013883e-05,
+      "loss": 0.971,
+      "step": 758
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.7561314206385933e-05,
+      "loss": 0.8042,
+      "step": 759
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.7584451642757982e-05,
+      "loss": 0.9901,
+      "step": 760
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.7607589079130032e-05,
+      "loss": 0.7315,
+      "step": 761
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.763072651550208e-05,
+      "loss": 0.6902,
+      "step": 762
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.7653863951874134e-05,
+      "loss": 0.8691,
+      "step": 763
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.7677001388246184e-05,
+      "loss": 1.198,
+      "step": 764
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.7700138824618234e-05,
+      "loss": 1.2012,
+      "step": 765
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.7723276260990283e-05,
+      "loss": 0.9897,
+      "step": 766
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.7746413697362333e-05,
+      "loss": 0.8305,
+      "step": 767
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.7769551133734382e-05,
+      "loss": 1.0476,
+      "step": 768
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.7792688570106432e-05,
+      "loss": 0.7687,
+      "step": 769
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.7815826006478485e-05,
+      "loss": 0.9307,
+      "step": 770
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.7838963442850535e-05,
+      "loss": 0.9387,
+      "step": 771
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.786210087922258e-05,
+      "loss": 0.9355,
+      "step": 772
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.7885238315594634e-05,
+      "loss": 1.0391,
+      "step": 773
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.7908375751966683e-05,
+      "loss": 0.6754,
+      "step": 774
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.7931513188338733e-05,
+      "loss": 0.8133,
+      "step": 775
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.7954650624710782e-05,
+      "loss": 0.8252,
+      "step": 776
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.7977788061082835e-05,
+      "loss": 0.8227,
+      "step": 777
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.800092549745488e-05,
+      "loss": 1.0508,
+      "step": 778
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.802406293382693e-05,
+      "loss": 1.127,
+      "step": 779
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.8047200370198984e-05,
+      "loss": 1.1487,
+      "step": 780
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.8070337806571034e-05,
+      "loss": 1.0351,
+      "step": 781
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.8093475242943083e-05,
+      "loss": 0.7615,
+      "step": 782
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.8116612679315133e-05,
+      "loss": 0.8265,
+      "step": 783
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8139750115687183e-05,
+      "loss": 0.714,
+      "step": 784
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8162887552059232e-05,
+      "loss": 0.6451,
+      "step": 785
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.818602498843128e-05,
+      "loss": 0.8381,
+      "step": 786
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8209162424803335e-05,
+      "loss": 0.9708,
+      "step": 787
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8232299861175384e-05,
+      "loss": 0.9125,
+      "step": 788
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.825543729754743e-05,
+      "loss": 0.8597,
+      "step": 789
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8278574733919483e-05,
+      "loss": 0.9264,
+      "step": 790
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8301712170291533e-05,
+      "loss": 0.9796,
+      "step": 791
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8324849606663583e-05,
+      "loss": 0.7372,
+      "step": 792
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8347987043035632e-05,
+      "loss": 1.0595,
+      "step": 793
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8371124479407685e-05,
+      "loss": 0.656,
+      "step": 794
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.839426191577973e-05,
+      "loss": 0.7575,
+      "step": 795
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.841739935215178e-05,
+      "loss": 1.1511,
+      "step": 796
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8440536788523834e-05,
+      "loss": 0.7146,
+      "step": 797
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8463674224895883e-05,
+      "loss": 0.8631,
+      "step": 798
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8486811661267933e-05,
+      "loss": 0.6841,
+      "step": 799
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8509949097639983e-05,
+      "loss": 0.7277,
+      "step": 800
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8533086534012032e-05,
+      "loss": 0.86,
+      "step": 801
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8556223970384082e-05,
+      "loss": 0.9133,
+      "step": 802
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.857936140675613e-05,
+      "loss": 1.0323,
+      "step": 803
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8602498843128184e-05,
+      "loss": 1.0204,
+      "step": 804
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8625636279500234e-05,
+      "loss": 1.111,
+      "step": 805
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8648773715872284e-05,
+      "loss": 0.9918,
+      "step": 806
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.867191115224433e-05,
+      "loss": 0.7901,
+      "step": 807
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8695048588616383e-05,
+      "loss": 0.7467,
+      "step": 808
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8718186024988432e-05,
+      "loss": 0.8381,
+      "step": 809
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8741323461360482e-05,
+      "loss": 0.9055,
+      "step": 810
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.876446089773253e-05,
+      "loss": 0.7295,
+      "step": 811
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.878759833410458e-05,
+      "loss": 1.0172,
+      "step": 812
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.881073577047663e-05,
+      "loss": 0.8684,
+      "step": 813
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.883387320684868e-05,
+      "loss": 0.9878,
+      "step": 814
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8857010643220733e-05,
+      "loss": 0.9298,
+      "step": 815
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8880148079592783e-05,
+      "loss": 0.9382,
+      "step": 816
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8903285515964832e-05,
+      "loss": 1.0389,
+      "step": 817
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.8926422952336882e-05,
+      "loss": 1.0342,
+      "step": 818
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.894956038870893e-05,
+      "loss": 1.1308,
+      "step": 819
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.897269782508098e-05,
+      "loss": 0.7478,
+      "step": 820
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.899583526145303e-05,
+      "loss": 0.8863,
+      "step": 821
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9018972697825084e-05,
+      "loss": 1.083,
+      "step": 822
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9042110134197133e-05,
+      "loss": 1.1181,
+      "step": 823
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.906524757056918e-05,
+      "loss": 1.0114,
+      "step": 824
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9088385006941232e-05,
+      "loss": 1.0925,
+      "step": 825
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9111522443313282e-05,
+      "loss": 0.66,
+      "step": 826
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.913465987968533e-05,
+      "loss": 1.0294,
+      "step": 827
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.915779731605738e-05,
+      "loss": 1.0615,
+      "step": 828
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9180934752429434e-05,
+      "loss": 1.1253,
+      "step": 829
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.920407218880148e-05,
+      "loss": 0.9194,
+      "step": 830
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.922720962517353e-05,
+      "loss": 1.201,
+      "step": 831
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9250347061545583e-05,
+      "loss": 1.0432,
+      "step": 832
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9273484497917632e-05,
+      "loss": 0.7592,
+      "step": 833
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9296621934289682e-05,
+      "loss": 0.8567,
+      "step": 834
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.931975937066173e-05,
+      "loss": 0.8206,
+      "step": 835
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.934289680703378e-05,
+      "loss": 0.8654,
+      "step": 836
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.936603424340583e-05,
+      "loss": 0.6572,
+      "step": 837
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.938917167977788e-05,
+      "loss": 0.9053,
+      "step": 838
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9412309116149933e-05,
+      "loss": 0.9368,
+      "step": 839
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9435446552521983e-05,
+      "loss": 0.8994,
+      "step": 840
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.945858398889403e-05,
+      "loss": 1.1748,
+      "step": 841
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9481721425266082e-05,
+      "loss": 0.798,
+      "step": 842
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9504858861638132e-05,
+      "loss": 0.9439,
+      "step": 843
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.952799629801018e-05,
+      "loss": 0.7896,
+      "step": 844
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.955113373438223e-05,
+      "loss": 0.9313,
+      "step": 845
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9574271170754284e-05,
+      "loss": 0.7075,
+      "step": 846
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.959740860712633e-05,
+      "loss": 0.6415,
+      "step": 847
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.962054604349838e-05,
+      "loss": 1.094,
+      "step": 848
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9643683479870433e-05,
+      "loss": 0.9684,
+      "step": 849
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9666820916242482e-05,
+      "loss": 0.8331,
+      "step": 850
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9689958352614532e-05,
+      "loss": 1.0455,
+      "step": 851
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.971309578898658e-05,
+      "loss": 0.89,
+      "step": 852
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.973623322535863e-05,
+      "loss": 1.1062,
+      "step": 853
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.975937066173068e-05,
+      "loss": 0.7472,
+      "step": 854
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.978250809810273e-05,
+      "loss": 0.8953,
+      "step": 855
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9805645534474783e-05,
+      "loss": 0.8632,
+      "step": 856
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9828782970846833e-05,
+      "loss": 0.646,
+      "step": 857
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9851920407218882e-05,
+      "loss": 1.1045,
+      "step": 858
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.987505784359093e-05,
+      "loss": 1.1304,
+      "step": 859
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.989819527996298e-05,
+      "loss": 0.7475,
+      "step": 860
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.992133271633503e-05,
+      "loss": 0.8933,
+      "step": 861
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.994447015270708e-05,
+      "loss": 1.0295,
+      "step": 862
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9967607589079134e-05,
+      "loss": 1.1041,
+      "step": 863
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.999074502545118e-05,
+      "loss": 1.0127,
+      "step": 864
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.001388246182323e-05,
+      "loss": 0.7023,
+      "step": 865
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.003701989819528e-05,
+      "loss": 1.0707,
+      "step": 866
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.0060157334567332e-05,
+      "loss": 0.9783,
+      "step": 867
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.008329477093938e-05,
+      "loss": 0.8202,
+      "step": 868
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.010643220731143e-05,
+      "loss": 1.2431,
+      "step": 869
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.012956964368348e-05,
+      "loss": 0.6881,
+      "step": 870
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.015270708005553e-05,
+      "loss": 0.7877,
+      "step": 871
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.017584451642758e-05,
+      "loss": 0.677,
+      "step": 872
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.019898195279963e-05,
+      "loss": 0.9814,
+      "step": 873
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.0222119389171682e-05,
+      "loss": 0.923,
+      "step": 874
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.0245256825543732e-05,
+      "loss": 0.9504,
+      "step": 875
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.0268394261915778e-05,
+      "loss": 1.2292,
+      "step": 876
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.029153169828783e-05,
+      "loss": 0.756,
+      "step": 877
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.031466913465988e-05,
+      "loss": 0.7827,
+      "step": 878
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.033780657103193e-05,
+      "loss": 0.8547,
+      "step": 879
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.036094400740398e-05,
+      "loss": 0.8545,
+      "step": 880
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.0384081443776033e-05,
+      "loss": 0.7906,
+      "step": 881
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.040721888014808e-05,
+      "loss": 0.7689,
+      "step": 882
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.043035631652013e-05,
+      "loss": 1.0898,
+      "step": 883
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.045349375289218e-05,
+      "loss": 1.1955,
+      "step": 884
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.047663118926423e-05,
+      "loss": 0.6,
+      "step": 885
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.049976862563628e-05,
+      "loss": 0.6576,
+      "step": 886
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.052290606200833e-05,
+      "loss": 0.7689,
+      "step": 887
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.054604349838038e-05,
+      "loss": 1.0726,
+      "step": 888
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.056918093475243e-05,
+      "loss": 0.8515,
+      "step": 889
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.059231837112448e-05,
+      "loss": 1.0834,
+      "step": 890
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.0615455807496532e-05,
+      "loss": 0.9233,
+      "step": 891
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.063859324386858e-05,
+      "loss": 0.851,
+      "step": 892
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.0661730680240628e-05,
+      "loss": 1.049,
+      "step": 893
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.068486811661268e-05,
+      "loss": 0.8121,
+      "step": 894
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.070800555298473e-05,
+      "loss": 1.1726,
+      "step": 895
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.073114298935678e-05,
+      "loss": 0.791,
+      "step": 896
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.075428042572883e-05,
+      "loss": 0.8898,
+      "step": 897
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.0777417862100883e-05,
+      "loss": 1.068,
+      "step": 898
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.080055529847293e-05,
+      "loss": 0.8446,
+      "step": 899
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.082369273484498e-05,
+      "loss": 0.7203,
+      "step": 900
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.084683017121703e-05,
+      "loss": 0.8674,
+      "step": 901
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.086996760758908e-05,
+      "loss": 1.0196,
+      "step": 902
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.089310504396113e-05,
+      "loss": 0.5742,
+      "step": 903
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.091624248033318e-05,
+      "loss": 0.753,
+      "step": 904
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.093937991670523e-05,
+      "loss": 1.1116,
+      "step": 905
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.096251735307728e-05,
+      "loss": 1.0145,
+      "step": 906
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.098565478944933e-05,
+      "loss": 0.9857,
+      "step": 907
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.1008792225821382e-05,
+      "loss": 0.8266,
+      "step": 908
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.103192966219343e-05,
+      "loss": 0.6835,
+      "step": 909
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.105506709856548e-05,
+      "loss": 0.7988,
+      "step": 910
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.107820453493753e-05,
+      "loss": 0.6316,
+      "step": 911
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.110134197130958e-05,
+      "loss": 0.9152,
+      "step": 912
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.112447940768163e-05,
+      "loss": 0.8021,
+      "step": 913
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.114761684405368e-05,
+      "loss": 0.8775,
+      "step": 914
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.1170754280425732e-05,
+      "loss": 1.1772,
+      "step": 915
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.119389171679778e-05,
+      "loss": 0.7634,
+      "step": 916
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.1217029153169828e-05,
+      "loss": 0.5813,
+      "step": 917
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.124016658954188e-05,
+      "loss": 1.2916,
+      "step": 918
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.126330402591393e-05,
+      "loss": 1.1848,
+      "step": 919
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.128644146228598e-05,
+      "loss": 0.9542,
+      "step": 920
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.130957889865803e-05,
+      "loss": 1.0353,
+      "step": 921
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.133271633503008e-05,
+      "loss": 1.2075,
+      "step": 922
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.135585377140213e-05,
+      "loss": 0.9461,
+      "step": 923
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.137899120777418e-05,
+      "loss": 0.8815,
+      "step": 924
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.140212864414623e-05,
+      "loss": 0.7896,
+      "step": 925
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.142526608051828e-05,
+      "loss": 1.0097,
+      "step": 926
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.144840351689033e-05,
+      "loss": 0.8174,
+      "step": 927
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.1471540953262377e-05,
+      "loss": 0.6874,
+      "step": 928
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.149467838963443e-05,
+      "loss": 0.7009,
+      "step": 929
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.151781582600648e-05,
+      "loss": 1.146,
+      "step": 930
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.154095326237853e-05,
+      "loss": 0.7866,
+      "step": 931
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.156409069875058e-05,
+      "loss": 0.6452,
+      "step": 932
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.158722813512263e-05,
+      "loss": 0.6601,
+      "step": 933
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.1610365571494678e-05,
+      "loss": 0.9909,
+      "step": 934
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.1633503007866727e-05,
+      "loss": 0.9481,
+      "step": 935
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.165664044423878e-05,
+      "loss": 0.9502,
+      "step": 936
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.167977788061083e-05,
+      "loss": 1.1082,
+      "step": 937
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.170291531698288e-05,
+      "loss": 0.8332,
+      "step": 938
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.172605275335493e-05,
+      "loss": 0.8711,
+      "step": 939
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.174919018972698e-05,
+      "loss": 0.9582,
+      "step": 940
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.1772327626099028e-05,
+      "loss": 0.7927,
+      "step": 941
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.1795465062471078e-05,
+      "loss": 0.9055,
+      "step": 942
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.181860249884313e-05,
+      "loss": 1.1168,
+      "step": 943
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.184173993521518e-05,
+      "loss": 0.8702,
+      "step": 944
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.1864877371587227e-05,
+      "loss": 0.8911,
+      "step": 945
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.188801480795928e-05,
+      "loss": 1.1573,
+      "step": 946
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.191115224433133e-05,
+      "loss": 1.172,
+      "step": 947
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.193428968070338e-05,
+      "loss": 0.6915,
+      "step": 948
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.195742711707543e-05,
+      "loss": 0.6115,
+      "step": 949
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.198056455344748e-05,
+      "loss": 0.8388,
+      "step": 950
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.2003701989819527e-05,
+      "loss": 0.9954,
+      "step": 951
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.2026839426191577e-05,
+      "loss": 0.8681,
+      "step": 952
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.204997686256363e-05,
+      "loss": 1.251,
+      "step": 953
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.207311429893568e-05,
+      "loss": 0.8666,
+      "step": 954
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.209625173530773e-05,
+      "loss": 0.6917,
+      "step": 955
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.211938917167978e-05,
+      "loss": 1.1706,
+      "step": 956
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.214252660805183e-05,
+      "loss": 0.7673,
+      "step": 957
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.2165664044423878e-05,
+      "loss": 0.7311,
+      "step": 958
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.2188801480795928e-05,
+      "loss": 1.104,
+      "step": 959
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.221193891716798e-05,
+      "loss": 0.5688,
+      "step": 960
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.223507635354003e-05,
+      "loss": 0.8247,
+      "step": 961
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.225821378991208e-05,
+      "loss": 1.0324,
+      "step": 962
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.228135122628413e-05,
+      "loss": 1.0242,
+      "step": 963
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.230448866265618e-05,
+      "loss": 0.6754,
+      "step": 964
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.232762609902823e-05,
+      "loss": 0.7513,
+      "step": 965
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.2350763535400278e-05,
+      "loss": 0.7338,
+      "step": 966
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.237390097177233e-05,
+      "loss": 0.8271,
+      "step": 967
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.2397038408144377e-05,
+      "loss": 1.0398,
+      "step": 968
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.2420175844516427e-05,
+      "loss": 0.5297,
+      "step": 969
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.244331328088848e-05,
+      "loss": 1.0848,
+      "step": 970
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.246645071726053e-05,
+      "loss": 0.8384,
+      "step": 971
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.248958815363258e-05,
+      "loss": 0.7335,
+      "step": 972
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.251272559000463e-05,
+      "loss": 0.6599,
+      "step": 973
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.2535863026376678e-05,
+      "loss": 0.9804,
+      "step": 974
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.2559000462748728e-05,
+      "loss": 0.791,
+      "step": 975
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.2582137899120777e-05,
+      "loss": 0.7744,
+      "step": 976
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.260527533549283e-05,
+      "loss": 0.7843,
+      "step": 977
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.262841277186488e-05,
+      "loss": 1.1388,
+      "step": 978
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.265155020823693e-05,
+      "loss": 1.147,
+      "step": 979
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.267468764460898e-05,
+      "loss": 1.2474,
+      "step": 980
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.269782508098103e-05,
+      "loss": 1.2611,
+      "step": 981
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.2720962517353078e-05,
+      "loss": 0.9131,
+      "step": 982
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.2744099953725128e-05,
+      "loss": 0.7311,
+      "step": 983
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.276723739009718e-05,
+      "loss": 0.9251,
+      "step": 984
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.279037482646923e-05,
+      "loss": 1.133,
+      "step": 985
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.2813512262841277e-05,
+      "loss": 0.9381,
+      "step": 986
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.2836649699213326e-05,
+      "loss": 1.036,
+      "step": 987
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.285978713558538e-05,
+      "loss": 0.9708,
+      "step": 988
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.288292457195743e-05,
+      "loss": 0.8352,
+      "step": 989
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.2906062008329478e-05,
+      "loss": 0.9538,
+      "step": 990
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.2929199444701528e-05,
+      "loss": 1.0155,
+      "step": 991
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.2952336881073577e-05,
+      "loss": 0.8434,
+      "step": 992
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.2975474317445627e-05,
+      "loss": 1.0551,
+      "step": 993
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.2998611753817677e-05,
+      "loss": 0.6711,
+      "step": 994
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.302174919018973e-05,
+      "loss": 0.9597,
+      "step": 995
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.304488662656178e-05,
+      "loss": 0.5947,
+      "step": 996
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.3068024062933825e-05,
+      "loss": 0.8372,
+      "step": 997
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.3091161499305878e-05,
+      "loss": 0.8419,
+      "step": 998
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.3114298935677928e-05,
+      "loss": 1.1515,
+      "step": 999
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.3137436372049977e-05,
+      "loss": 0.7127,
+      "step": 1000
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.3160573808422027e-05,
+      "loss": 0.9702,
+      "step": 1001
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.318371124479408e-05,
+      "loss": 0.9636,
+      "step": 1002
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.3206848681166126e-05,
+      "loss": 1.1177,
+      "step": 1003
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.3229986117538176e-05,
+      "loss": 0.8849,
+      "step": 1004
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.325312355391023e-05,
+      "loss": 0.8906,
+      "step": 1005
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.327626099028228e-05,
+      "loss": 1.1691,
+      "step": 1006
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.3299398426654328e-05,
+      "loss": 0.9443,
+      "step": 1007
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.3322535863026378e-05,
+      "loss": 0.92,
+      "step": 1008
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.3345673299398427e-05,
+      "loss": 1.035,
+      "step": 1009
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.3368810735770477e-05,
+      "loss": 0.8343,
+      "step": 1010
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.3391948172142526e-05,
+      "loss": 0.9261,
+      "step": 1011
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.341508560851458e-05,
+      "loss": 1.177,
+      "step": 1012
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.343822304488663e-05,
+      "loss": 0.8763,
+      "step": 1013
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.346136048125868e-05,
+      "loss": 1.0501,
+      "step": 1014
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.3484497917630728e-05,
+      "loss": 1.0311,
+      "step": 1015
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.3507635354002778e-05,
+      "loss": 0.8804,
+      "step": 1016
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.3530772790374827e-05,
+      "loss": 0.7445,
+      "step": 1017
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.3553910226746877e-05,
+      "loss": 0.7637,
+      "step": 1018
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.357704766311893e-05,
+      "loss": 1.0877,
+      "step": 1019
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.3600185099490976e-05,
+      "loss": 0.7552,
+      "step": 1020
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.3623322535863026e-05,
+      "loss": 1.023,
+      "step": 1021
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.364645997223508e-05,
+      "loss": 0.7658,
+      "step": 1022
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.3669597408607128e-05,
+      "loss": 0.934,
+      "step": 1023
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.3692734844979178e-05,
+      "loss": 0.8484,
+      "step": 1024
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.3715872281351227e-05,
+      "loss": 0.8477,
+      "step": 1025
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.3739009717723277e-05,
+      "loss": 0.957,
+      "step": 1026
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.3762147154095326e-05,
+      "loss": 1.0728,
+      "step": 1027
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.3785284590467376e-05,
+      "loss": 0.6957,
+      "step": 1028
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.380842202683943e-05,
+      "loss": 0.9257,
+      "step": 1029
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.383155946321148e-05,
+      "loss": 0.7411,
+      "step": 1030
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.3854696899583528e-05,
+      "loss": 1.0418,
+      "step": 1031
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.3877834335955578e-05,
+      "loss": 0.8005,
+      "step": 1032
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.3900971772327627e-05,
+      "loss": 0.8818,
+      "step": 1033
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.3924109208699677e-05,
+      "loss": 0.8109,
+      "step": 1034
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.3947246645071726e-05,
+      "loss": 1.0315,
+      "step": 1035
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.397038408144378e-05,
+      "loss": 0.9709,
+      "step": 1036
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.399352151781583e-05,
+      "loss": 1.0357,
+      "step": 1037
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.4016658954187875e-05,
+      "loss": 0.987,
+      "step": 1038
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.4039796390559928e-05,
+      "loss": 0.9939,
+      "step": 1039
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.4062933826931978e-05,
+      "loss": 1.037,
+      "step": 1040
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.4086071263304027e-05,
+      "loss": 1.047,
+      "step": 1041
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.4109208699676077e-05,
+      "loss": 0.6723,
+      "step": 1042
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.4132346136048127e-05,
+      "loss": 1.0413,
+      "step": 1043
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.4155483572420176e-05,
+      "loss": 0.8296,
+      "step": 1044
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.4178621008792226e-05,
+      "loss": 0.9873,
+      "step": 1045
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.420175844516428e-05,
+      "loss": 0.9859,
+      "step": 1046
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.4224895881536328e-05,
+      "loss": 0.8115,
+      "step": 1047
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.4248033317908378e-05,
+      "loss": 0.6988,
+      "step": 1048
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.4271170754280424e-05,
+      "loss": 0.7309,
+      "step": 1049
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.4294308190652477e-05,
+      "loss": 1.069,
+      "step": 1050
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.4317445627024527e-05,
+      "loss": 0.9689,
+      "step": 1051
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.4340583063396576e-05,
+      "loss": 0.8366,
+      "step": 1052
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.436372049976863e-05,
+      "loss": 0.7382,
+      "step": 1053
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.438685793614068e-05,
+      "loss": 1.1256,
+      "step": 1054
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.4409995372512725e-05,
+      "loss": 0.9324,
+      "step": 1055
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.4433132808884775e-05,
+      "loss": 0.8199,
+      "step": 1056
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.4456270245256827e-05,
+      "loss": 1.0265,
+      "step": 1057
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.4479407681628877e-05,
+      "loss": 1.1326,
+      "step": 1058
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.4502545118000927e-05,
+      "loss": 0.8782,
+      "step": 1059
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.4525682554372976e-05,
+      "loss": 0.8094,
+      "step": 1060
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.4548819990745026e-05,
+      "loss": 0.7313,
+      "step": 1061
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.4571957427117075e-05,
+      "loss": 0.9021,
+      "step": 1062
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.4595094863489125e-05,
+      "loss": 0.9404,
+      "step": 1063
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.4618232299861178e-05,
+      "loss": 1.0019,
+      "step": 1064
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.4641369736233228e-05,
+      "loss": 0.8017,
+      "step": 1065
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.4664507172605277e-05,
+      "loss": 0.8976,
+      "step": 1066
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.4687644608977327e-05,
+      "loss": 0.7664,
+      "step": 1067
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.4710782045349376e-05,
+      "loss": 1.3953,
+      "step": 1068
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.4733919481721426e-05,
+      "loss": 0.8126,
+      "step": 1069
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.4757056918093475e-05,
+      "loss": 1.0063,
+      "step": 1070
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.478019435446553e-05,
+      "loss": 0.9342,
+      "step": 1071
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.4803331790837575e-05,
+      "loss": 0.8985,
+      "step": 1072
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.4826469227209624e-05,
+      "loss": 0.865,
+      "step": 1073
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.4849606663581677e-05,
+      "loss": 0.993,
+      "step": 1074
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.4872744099953727e-05,
+      "loss": 0.6864,
+      "step": 1075
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.4895881536325776e-05,
+      "loss": 0.966,
+      "step": 1076
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.4919018972697826e-05,
+      "loss": 1.1562,
+      "step": 1077
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.4942156409069876e-05,
+      "loss": 0.616,
+      "step": 1078
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.4965293845441925e-05,
+      "loss": 0.9211,
+      "step": 1079
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.4988431281813975e-05,
+      "loss": 1.3408,
+      "step": 1080
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.5011568718186024e-05,
+      "loss": 0.6968,
+      "step": 1081
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.5034706154558074e-05,
+      "loss": 0.9365,
+      "step": 1082
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.5057843590930123e-05,
+      "loss": 0.7825,
+      "step": 1083
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.5080981027302176e-05,
+      "loss": 0.7918,
+      "step": 1084
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.5104118463674226e-05,
+      "loss": 1.1188,
+      "step": 1085
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.5127255900046276e-05,
+      "loss": 0.4752,
+      "step": 1086
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.5150393336418325e-05,
+      "loss": 0.9497,
+      "step": 1087
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.5173530772790378e-05,
+      "loss": 0.9594,
+      "step": 1088
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.5196668209162428e-05,
+      "loss": 0.6286,
+      "step": 1089
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.5219805645534477e-05,
+      "loss": 0.8208,
+      "step": 1090
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.5242943081906527e-05,
+      "loss": 0.8513,
+      "step": 1091
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.526608051827858e-05,
+      "loss": 0.733,
+      "step": 1092
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.5289217954650623e-05,
+      "loss": 0.7109,
+      "step": 1093
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.5312355391022676e-05,
+      "loss": 0.7018,
+      "step": 1094
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.5335492827394725e-05,
+      "loss": 0.8966,
+      "step": 1095
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.5358630263766775e-05,
+      "loss": 0.9321,
+      "step": 1096
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.5381767700138824e-05,
+      "loss": 0.86,
+      "step": 1097
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.5404905136510877e-05,
+      "loss": 0.9543,
+      "step": 1098
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.5428042572882927e-05,
+      "loss": 0.8723,
+      "step": 1099
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.5451180009254977e-05,
+      "loss": 0.8858,
+      "step": 1100
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.5474317445627026e-05,
+      "loss": 0.9342,
+      "step": 1101
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.549745488199908e-05,
+      "loss": 0.6899,
+      "step": 1102
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.552059231837113e-05,
+      "loss": 0.6501,
+      "step": 1103
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.554372975474317e-05,
+      "loss": 0.839,
+      "step": 1104
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.5566867191115224e-05,
+      "loss": 1.038,
+      "step": 1105
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.5590004627487274e-05,
+      "loss": 0.8777,
+      "step": 1106
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.5613142063859324e-05,
+      "loss": 0.9654,
+      "step": 1107
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.5636279500231373e-05,
+      "loss": 1.0826,
+      "step": 1108
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.5659416936603426e-05,
+      "loss": 0.7455,
+      "step": 1109
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.5682554372975476e-05,
+      "loss": 0.9843,
+      "step": 1110
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.5705691809347525e-05,
+      "loss": 0.8804,
+      "step": 1111
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.572882924571958e-05,
+      "loss": 0.7203,
+      "step": 1112
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.5751966682091628e-05,
+      "loss": 1.0488,
+      "step": 1113
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.5775104118463678e-05,
+      "loss": 0.4827,
+      "step": 1114
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.5798241554835727e-05,
+      "loss": 0.7936,
+      "step": 1115
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.5821378991207773e-05,
+      "loss": 0.8567,
+      "step": 1116
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.5844516427579823e-05,
+      "loss": 0.8811,
+      "step": 1117
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.5867653863951872e-05,
+      "loss": 1.1266,
+      "step": 1118
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.5890791300323925e-05,
+      "loss": 0.8708,
+      "step": 1119
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.5913928736695975e-05,
+      "loss": 1.0359,
+      "step": 1120
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.5937066173068025e-05,
+      "loss": 0.8327,
+      "step": 1121
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.5960203609440074e-05,
+      "loss": 0.7157,
+      "step": 1122
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.5983341045812127e-05,
+      "loss": 0.8399,
+      "step": 1123
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.6006478482184177e-05,
+      "loss": 0.8349,
+      "step": 1124
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.6029615918556226e-05,
+      "loss": 1.0517,
+      "step": 1125
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.605275335492828e-05,
+      "loss": 0.7526,
+      "step": 1126
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.6075890791300322e-05,
+      "loss": 0.9301,
+      "step": 1127
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.6099028227672372e-05,
+      "loss": 0.7156,
+      "step": 1128
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.6122165664044425e-05,
+      "loss": 0.7115,
+      "step": 1129
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.6145303100416474e-05,
+      "loss": 0.833,
+      "step": 1130
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.6168440536788524e-05,
+      "loss": 0.74,
+      "step": 1131
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.6191577973160573e-05,
+      "loss": 1.0222,
+      "step": 1132
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.6214715409532626e-05,
+      "loss": 0.8656,
+      "step": 1133
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.6237852845904676e-05,
+      "loss": 0.727,
+      "step": 1134
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.6260990282276726e-05,
+      "loss": 0.8848,
+      "step": 1135
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.6284127718648775e-05,
+      "loss": 0.6405,
+      "step": 1136
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.6307265155020828e-05,
+      "loss": 0.7852,
+      "step": 1137
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.6330402591392878e-05,
+      "loss": 0.5377,
+      "step": 1138
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.6353540027764924e-05,
+      "loss": 0.801,
+      "step": 1139
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.6376677464136973e-05,
+      "loss": 0.8865,
+      "step": 1140
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.6399814900509023e-05,
+      "loss": 0.9631,
+      "step": 1141
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.6422952336881073e-05,
+      "loss": 0.6693,
+      "step": 1142
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.6446089773253126e-05,
+      "loss": 1.0063,
+      "step": 1143
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.6469227209625175e-05,
+      "loss": 1.1552,
+      "step": 1144
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.6492364645997225e-05,
+      "loss": 1.1479,
+      "step": 1145
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.6515502082369274e-05,
+      "loss": 0.9074,
+      "step": 1146
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.6538639518741327e-05,
+      "loss": 0.8906,
+      "step": 1147
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.6561776955113377e-05,
+      "loss": 0.9257,
+      "step": 1148
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.6584914391485427e-05,
+      "loss": 0.7337,
+      "step": 1149
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.6608051827857473e-05,
+      "loss": 0.9046,
+      "step": 1150
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.6631189264229522e-05,
+      "loss": 0.8703,
+      "step": 1151
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.6654326700601572e-05,
+      "loss": 1.1667,
+      "step": 1152
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.6677464136973625e-05,
+      "loss": 0.8414,
+      "step": 1153
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.6700601573345674e-05,
+      "loss": 0.9009,
+      "step": 1154
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.6723739009717724e-05,
+      "loss": 0.8795,
+      "step": 1155
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.6746876446089774e-05,
+      "loss": 0.8537,
+      "step": 1156
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.6770013882461827e-05,
+      "loss": 0.8583,
+      "step": 1157
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.6793151318833876e-05,
+      "loss": 0.8459,
+      "step": 1158
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.6816288755205926e-05,
+      "loss": 0.8882,
+      "step": 1159
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.6839426191577975e-05,
+      "loss": 1.0875,
+      "step": 1160
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.686256362795003e-05,
+      "loss": 1.0544,
+      "step": 1161
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.688570106432207e-05,
+      "loss": 0.8731,
+      "step": 1162
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.690883850069412e-05,
+      "loss": 1.0955,
+      "step": 1163
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.6931975937066174e-05,
+      "loss": 0.8327,
+      "step": 1164
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.6955113373438223e-05,
+      "loss": 0.7402,
+      "step": 1165
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.6978250809810273e-05,
+      "loss": 0.7917,
+      "step": 1166
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.7001388246182326e-05,
+      "loss": 0.8143,
+      "step": 1167
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.7024525682554375e-05,
+      "loss": 0.991,
+      "step": 1168
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.7047663118926425e-05,
+      "loss": 0.714,
+      "step": 1169
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.7070800555298475e-05,
+      "loss": 0.8062,
+      "step": 1170
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.7093937991670528e-05,
+      "loss": 0.7328,
+      "step": 1171
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.7117075428042577e-05,
+      "loss": 0.7,
+      "step": 1172
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.714021286441462e-05,
+      "loss": 0.5823,
+      "step": 1173
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.7163350300786673e-05,
+      "loss": 0.8134,
+      "step": 1174
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.7186487737158723e-05,
+      "loss": 1.0537,
+      "step": 1175
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.7209625173530772e-05,
+      "loss": 0.7651,
+      "step": 1176
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.723276260990282e-05,
+      "loss": 1.157,
+      "step": 1177
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.7255900046274875e-05,
+      "loss": 0.8064,
+      "step": 1178
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.7279037482646924e-05,
+      "loss": 1.2746,
+      "step": 1179
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.7302174919018974e-05,
+      "loss": 0.6265,
+      "step": 1180
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.7325312355391023e-05,
+      "loss": 0.6031,
+      "step": 1181
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.7348449791763076e-05,
+      "loss": 0.8972,
+      "step": 1182
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.7371587228135126e-05,
+      "loss": 1.1993,
+      "step": 1183
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.7394724664507176e-05,
+      "loss": 1.1156,
+      "step": 1184
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.7417862100879222e-05,
+      "loss": 0.7244,
+      "step": 1185
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.744099953725127e-05,
+      "loss": 0.918,
+      "step": 1186
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.746413697362332e-05,
+      "loss": 0.928,
+      "step": 1187
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.7487274409995374e-05,
+      "loss": 1.1073,
+      "step": 1188
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.7510411846367423e-05,
+      "loss": 1.1168,
+      "step": 1189
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.7533549282739473e-05,
+      "loss": 0.6833,
+      "step": 1190
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.7556686719111523e-05,
+      "loss": 0.8347,
+      "step": 1191
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.7579824155483576e-05,
+      "loss": 1.1063,
+      "step": 1192
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.7602961591855625e-05,
+      "loss": 0.8092,
+      "step": 1193
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.7626099028227675e-05,
+      "loss": 0.6074,
+      "step": 1194
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.7649236464599724e-05,
+      "loss": 0.9542,
+      "step": 1195
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.7672373900971777e-05,
+      "loss": 1.1723,
+      "step": 1196
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.769551133734382e-05,
+      "loss": 0.8798,
+      "step": 1197
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.7718648773715873e-05,
+      "loss": 1.2244,
+      "step": 1198
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.7741786210087923e-05,
+      "loss": 0.7599,
+      "step": 1199
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.7764923646459972e-05,
+      "loss": 0.9154,
+      "step": 1200
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.7788061082832022e-05,
+      "loss": 0.882,
+      "step": 1201
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.7811198519204075e-05,
+      "loss": 0.9999,
+      "step": 1202
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.7834335955576124e-05,
+      "loss": 0.8188,
+      "step": 1203
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.7857473391948174e-05,
+      "loss": 0.9013,
+      "step": 1204
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.7880610828320224e-05,
+      "loss": 0.8653,
+      "step": 1205
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.7903748264692277e-05,
+      "loss": 0.8213,
+      "step": 1206
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.7926885701064326e-05,
+      "loss": 0.8922,
+      "step": 1207
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.7950023137436372e-05,
+      "loss": 0.8389,
+      "step": 1208
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.7973160573808422e-05,
+      "loss": 0.9719,
+      "step": 1209
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.799629801018047e-05,
+      "loss": 0.8814,
+      "step": 1210
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.801943544655252e-05,
+      "loss": 0.9604,
+      "step": 1211
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.8042572882924574e-05,
+      "loss": 0.8161,
+      "step": 1212
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.8065710319296624e-05,
+      "loss": 1.0261,
+      "step": 1213
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.8088847755668673e-05,
+      "loss": 0.9772,
+      "step": 1214
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.8111985192040723e-05,
+      "loss": 0.7531,
+      "step": 1215
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.8135122628412776e-05,
+      "loss": 1.0247,
+      "step": 1216
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.8158260064784825e-05,
+      "loss": 0.8185,
+      "step": 1217
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.8181397501156875e-05,
+      "loss": 0.6912,
+      "step": 1218
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.8204534937528925e-05,
+      "loss": 0.8822,
+      "step": 1219
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.822767237390097e-05,
+      "loss": 1.0835,
+      "step": 1220
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.825080981027302e-05,
+      "loss": 1.1557,
+      "step": 1221
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.8273947246645073e-05,
+      "loss": 0.8552,
+      "step": 1222
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.8297084683017123e-05,
+      "loss": 1.0986,
+      "step": 1223
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.8320222119389172e-05,
+      "loss": 0.7043,
+      "step": 1224
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.8343359555761222e-05,
+      "loss": 1.0338,
+      "step": 1225
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.8366496992133275e-05,
+      "loss": 0.99,
+      "step": 1226
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.8389634428505325e-05,
+      "loss": 0.774,
+      "step": 1227
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.8412771864877374e-05,
+      "loss": 0.9782,
+      "step": 1228
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.8435909301249424e-05,
+      "loss": 0.9809,
+      "step": 1229
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.8459046737621477e-05,
+      "loss": 0.831,
+      "step": 1230
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.848218417399352e-05,
+      "loss": 0.5556,
+      "step": 1231
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.850532161036557e-05,
+      "loss": 1.1141,
+      "step": 1232
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.8528459046737622e-05,
+      "loss": 0.9438,
+      "step": 1233
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.8551596483109672e-05,
+      "loss": 0.9052,
+      "step": 1234
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.857473391948172e-05,
+      "loss": 0.9299,
+      "step": 1235
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.859787135585377e-05,
+      "loss": 0.8617,
+      "step": 1236
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.8621008792225824e-05,
+      "loss": 1.0645,
+      "step": 1237
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.8644146228597873e-05,
+      "loss": 0.7772,
+      "step": 1238
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.8667283664969923e-05,
+      "loss": 1.2618,
+      "step": 1239
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.8690421101341976e-05,
+      "loss": 0.7531,
+      "step": 1240
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.8713558537714026e-05,
+      "loss": 1.0803,
+      "step": 1241
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.8736695974086075e-05,
+      "loss": 1.0891,
+      "step": 1242
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.875983341045812e-05,
+      "loss": 1.1216,
+      "step": 1243
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.878297084683017e-05,
+      "loss": 0.9528,
+      "step": 1244
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.880610828320222e-05,
+      "loss": 0.7346,
+      "step": 1245
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.882924571957427e-05,
+      "loss": 0.7453,
+      "step": 1246
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.8852383155946323e-05,
+      "loss": 1.1026,
+      "step": 1247
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.8875520592318373e-05,
+      "loss": 0.7684,
+      "step": 1248
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.8898658028690422e-05,
+      "loss": 0.9754,
+      "step": 1249
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.8921795465062472e-05,
+      "loss": 0.7494,
+      "step": 1250
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.8944932901434525e-05,
+      "loss": 0.6834,
+      "step": 1251
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.8968070337806574e-05,
+      "loss": 0.7649,
+      "step": 1252
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.8991207774178624e-05,
+      "loss": 1.0369,
+      "step": 1253
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.901434521055067e-05,
+      "loss": 0.9815,
+      "step": 1254
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.903748264692272e-05,
+      "loss": 0.9705,
+      "step": 1255
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.906062008329477e-05,
+      "loss": 1.1907,
+      "step": 1256
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.9083757519666822e-05,
+      "loss": 0.9564,
+      "step": 1257
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.9106894956038872e-05,
+      "loss": 0.8116,
+      "step": 1258
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.913003239241092e-05,
+      "loss": 0.5773,
+      "step": 1259
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.915316982878297e-05,
+      "loss": 1.0773,
+      "step": 1260
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.9176307265155024e-05,
+      "loss": 0.7716,
+      "step": 1261
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.9199444701527074e-05,
+      "loss": 1.1212,
+      "step": 1262
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.9222582137899123e-05,
+      "loss": 0.8496,
+      "step": 1263
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.9245719574271173e-05,
+      "loss": 0.5506,
+      "step": 1264
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.9268857010643226e-05,
+      "loss": 0.9919,
+      "step": 1265
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.929199444701527e-05,
+      "loss": 0.7335,
+      "step": 1266
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.931513188338732e-05,
+      "loss": 0.886,
+      "step": 1267
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.933826931975937e-05,
+      "loss": 1.0058,
+      "step": 1268
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.936140675613142e-05,
+      "loss": 0.8539,
+      "step": 1269
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.938454419250347e-05,
+      "loss": 0.8426,
+      "step": 1270
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.9407681628875523e-05,
+      "loss": 0.6705,
+      "step": 1271
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.9430819065247573e-05,
+      "loss": 1.2234,
+      "step": 1272
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.9453956501619622e-05,
+      "loss": 0.939,
+      "step": 1273
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.9477093937991672e-05,
+      "loss": 0.8571,
+      "step": 1274
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.9500231374363725e-05,
+      "loss": 0.7742,
+      "step": 1275
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.9523368810735775e-05,
+      "loss": 0.7399,
+      "step": 1276
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.954650624710782e-05,
+      "loss": 0.9186,
+      "step": 1277
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.956964368347987e-05,
+      "loss": 0.746,
+      "step": 1278
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.959278111985192e-05,
+      "loss": 1.0123,
+      "step": 1279
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.961591855622397e-05,
+      "loss": 0.7172,
+      "step": 1280
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.9639055992596022e-05,
+      "loss": 0.7818,
+      "step": 1281
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.9662193428968072e-05,
+      "loss": 0.665,
+      "step": 1282
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.968533086534012e-05,
+      "loss": 0.8838,
+      "step": 1283
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.970846830171217e-05,
+      "loss": 0.9709,
+      "step": 1284
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.9731605738084224e-05,
+      "loss": 0.6371,
+      "step": 1285
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.9754743174456274e-05,
+      "loss": 0.9203,
+      "step": 1286
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.9777880610828323e-05,
+      "loss": 0.596,
+      "step": 1287
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.9801018047200373e-05,
+      "loss": 0.5919,
+      "step": 1288
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.982415548357242e-05,
+      "loss": 1.3623,
+      "step": 1289
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.984729291994447e-05,
+      "loss": 0.8562,
+      "step": 1290
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.987043035631652e-05,
+      "loss": 0.8401,
+      "step": 1291
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.989356779268857e-05,
+      "loss": 0.8398,
+      "step": 1292
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.991670522906062e-05,
+      "loss": 1.0597,
+      "step": 1293
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.993984266543267e-05,
+      "loss": 0.9937,
+      "step": 1294
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.9962980101804723e-05,
+      "loss": 1.4358,
+      "step": 1295
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.9986117538176773e-05,
+      "loss": 0.9124,
+      "step": 1296
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.0009254974548823e-05,
+      "loss": 0.5197,
+      "step": 1297
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.0032392410920872e-05,
+      "loss": 0.8847,
+      "step": 1298
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.0055529847292925e-05,
+      "loss": 0.9197,
+      "step": 1299
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.0078667283664975e-05,
+      "loss": 0.7633,
+      "step": 1300
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.0101804720037018e-05,
+      "loss": 0.8902,
+      "step": 1301
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.012494215640907e-05,
+      "loss": 0.8535,
+      "step": 1302
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.014807959278112e-05,
+      "loss": 0.7973,
+      "step": 1303
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.017121702915317e-05,
+      "loss": 0.8633,
+      "step": 1304
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.019435446552522e-05,
+      "loss": 0.8136,
+      "step": 1305
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.0217491901897272e-05,
+      "loss": 0.7282,
+      "step": 1306
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.0240629338269322e-05,
+      "loss": 1.0944,
+      "step": 1307
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.026376677464137e-05,
+      "loss": 0.7788,
+      "step": 1308
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.028690421101342e-05,
+      "loss": 0.9498,
+      "step": 1309
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.0310041647385474e-05,
+      "loss": 0.9584,
+      "step": 1310
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.0333179083757524e-05,
+      "loss": 0.8788,
+      "step": 1311
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.035631652012957e-05,
+      "loss": 0.8785,
+      "step": 1312
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.037945395650162e-05,
+      "loss": 1.1304,
+      "step": 1313
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.040259139287367e-05,
+      "loss": 0.7293,
+      "step": 1314
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.042572882924572e-05,
+      "loss": 1.0329,
+      "step": 1315
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.044886626561777e-05,
+      "loss": 1.0675,
+      "step": 1316
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.047200370198982e-05,
+      "loss": 1.1109,
+      "step": 1317
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.049514113836187e-05,
+      "loss": 1.1578,
+      "step": 1318
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.051827857473392e-05,
+      "loss": 0.9178,
+      "step": 1319
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.054141601110597e-05,
+      "loss": 0.7402,
+      "step": 1320
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.056455344747802e-05,
+      "loss": 0.9588,
+      "step": 1321
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.058769088385007e-05,
+      "loss": 1.0199,
+      "step": 1322
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.0610828320222125e-05,
+      "loss": 1.1373,
+      "step": 1323
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.063396575659417e-05,
+      "loss": 0.9007,
+      "step": 1324
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.065710319296622e-05,
+      "loss": 0.9381,
+      "step": 1325
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.068024062933827e-05,
+      "loss": 1.0272,
+      "step": 1326
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.070337806571032e-05,
+      "loss": 1.1941,
+      "step": 1327
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.072651550208237e-05,
+      "loss": 0.9672,
+      "step": 1328
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.074965293845442e-05,
+      "loss": 0.9907,
+      "step": 1329
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.077279037482647e-05,
+      "loss": 0.6486,
+      "step": 1330
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.079592781119852e-05,
+      "loss": 0.7801,
+      "step": 1331
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.0819065247570575e-05,
+      "loss": 0.7415,
+      "step": 1332
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.084220268394262e-05,
+      "loss": 0.879,
+      "step": 1333
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.0865340120314674e-05,
+      "loss": 1.0314,
+      "step": 1334
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.088847755668672e-05,
+      "loss": 0.9666,
+      "step": 1335
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.0911614993058767e-05,
+      "loss": 0.8118,
+      "step": 1336
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.093475242943082e-05,
+      "loss": 0.6915,
+      "step": 1337
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.095788986580287e-05,
+      "loss": 0.7948,
+      "step": 1338
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.098102730217492e-05,
+      "loss": 1.0658,
+      "step": 1339
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.100416473854697e-05,
+      "loss": 0.7023,
+      "step": 1340
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.102730217491902e-05,
+      "loss": 0.9294,
+      "step": 1341
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.105043961129107e-05,
+      "loss": 1.0446,
+      "step": 1342
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.1073577047663124e-05,
+      "loss": 0.5862,
+      "step": 1343
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.109671448403517e-05,
+      "loss": 0.9235,
+      "step": 1344
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.111985192040722e-05,
+      "loss": 1.0561,
+      "step": 1345
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.1142989356779276e-05,
+      "loss": 0.8917,
+      "step": 1346
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.1166126793151315e-05,
+      "loss": 1.195,
+      "step": 1347
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.118926422952337e-05,
+      "loss": 0.8927,
+      "step": 1348
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.121240166589542e-05,
+      "loss": 1.0308,
+      "step": 1349
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.123553910226747e-05,
+      "loss": 0.8651,
+      "step": 1350
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.125867653863952e-05,
+      "loss": 1.2286,
+      "step": 1351
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.1281813975011573e-05,
+      "loss": 0.83,
+      "step": 1352
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.130495141138362e-05,
+      "loss": 0.9495,
+      "step": 1353
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.132808884775567e-05,
+      "loss": 1.1324,
+      "step": 1354
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.135122628412772e-05,
+      "loss": 0.7312,
+      "step": 1355
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.137436372049977e-05,
+      "loss": 0.7755,
+      "step": 1356
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.1397501156871825e-05,
+      "loss": 0.8645,
+      "step": 1357
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.1420638593243864e-05,
+      "loss": 1.0664,
+      "step": 1358
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.144377602961592e-05,
+      "loss": 0.7789,
+      "step": 1359
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.146691346598797e-05,
+      "loss": 0.7839,
+      "step": 1360
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.1490050902360016e-05,
+      "loss": 0.7659,
+      "step": 1361
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.151318833873207e-05,
+      "loss": 0.8611,
+      "step": 1362
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.153632577510412e-05,
+      "loss": 0.9832,
+      "step": 1363
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.155946321147617e-05,
+      "loss": 1.1911,
+      "step": 1364
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.158260064784822e-05,
+      "loss": 0.6258,
+      "step": 1365
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.160573808422027e-05,
+      "loss": 1.1626,
+      "step": 1366
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.162887552059232e-05,
+      "loss": 0.9353,
+      "step": 1367
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.1652012956964374e-05,
+      "loss": 1.3011,
+      "step": 1368
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.167515039333642e-05,
+      "loss": 0.9901,
+      "step": 1369
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.1698287829708466e-05,
+      "loss": 0.6965,
+      "step": 1370
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.172142526608052e-05,
+      "loss": 0.877,
+      "step": 1371
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.1744562702452565e-05,
+      "loss": 0.9885,
+      "step": 1372
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.176770013882462e-05,
+      "loss": 1.0311,
+      "step": 1373
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.179083757519667e-05,
+      "loss": 0.729,
+      "step": 1374
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.181397501156872e-05,
+      "loss": 1.0924,
+      "step": 1375
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.183711244794077e-05,
+      "loss": 0.8353,
+      "step": 1376
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.186024988431282e-05,
+      "loss": 1.0146,
+      "step": 1377
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.188338732068487e-05,
+      "loss": 0.7579,
+      "step": 1378
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.190652475705692e-05,
+      "loss": 1.0963,
+      "step": 1379
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.192966219342897e-05,
+      "loss": 0.7143,
+      "step": 1380
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.1952799629801015e-05,
+      "loss": 0.6562,
+      "step": 1381
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.197593706617307e-05,
+      "loss": 0.7655,
+      "step": 1382
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.199907450254512e-05,
+      "loss": 0.9005,
+      "step": 1383
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.202221193891717e-05,
+      "loss": 0.7023,
+      "step": 1384
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.204534937528922e-05,
+      "loss": 0.8512,
+      "step": 1385
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.2068486811661266e-05,
+      "loss": 0.8147,
+      "step": 1386
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.209162424803332e-05,
+      "loss": 0.8472,
+      "step": 1387
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.211476168440537e-05,
+      "loss": 0.9087,
+      "step": 1388
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.213789912077742e-05,
+      "loss": 0.7451,
+      "step": 1389
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.216103655714947e-05,
+      "loss": 1.1316,
+      "step": 1390
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.2184173993521524e-05,
+      "loss": 1.2758,
+      "step": 1391
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.220731142989357e-05,
+      "loss": 0.7545,
+      "step": 1392
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.223044886626562e-05,
+      "loss": 0.8471,
+      "step": 1393
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.225358630263767e-05,
+      "loss": 0.8049,
+      "step": 1394
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.2276723739009716e-05,
+      "loss": 0.7214,
+      "step": 1395
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.229986117538177e-05,
+      "loss": 0.8657,
+      "step": 1396
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.232299861175382e-05,
+      "loss": 0.8861,
+      "step": 1397
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.234613604812587e-05,
+      "loss": 0.808,
+      "step": 1398
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.236927348449792e-05,
+      "loss": 0.9998,
+      "step": 1399
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.239241092086997e-05,
+      "loss": 0.8147,
+      "step": 1400
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.241554835724202e-05,
+      "loss": 1.0036,
+      "step": 1401
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.243868579361407e-05,
+      "loss": 0.9507,
+      "step": 1402
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.246182322998612e-05,
+      "loss": 0.9563,
+      "step": 1403
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.248496066635817e-05,
+      "loss": 0.8626,
+      "step": 1404
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.250809810273022e-05,
+      "loss": 0.8186,
+      "step": 1405
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.2531235539102265e-05,
+      "loss": 0.5736,
+      "step": 1406
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.255437297547432e-05,
+      "loss": 1.0114,
+      "step": 1407
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.257751041184637e-05,
+      "loss": 1.3379,
+      "step": 1408
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.260064784821842e-05,
+      "loss": 0.9133,
+      "step": 1409
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.262378528459047e-05,
+      "loss": 0.9148,
+      "step": 1410
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.264692272096252e-05,
+      "loss": 0.9107,
+      "step": 1411
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.267006015733457e-05,
+      "loss": 0.6921,
+      "step": 1412
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.269319759370662e-05,
+      "loss": 0.9367,
+      "step": 1413
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.271633503007867e-05,
+      "loss": 0.6604,
+      "step": 1414
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.273947246645072e-05,
+      "loss": 0.8272,
+      "step": 1415
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.276260990282277e-05,
+      "loss": 1.0633,
+      "step": 1416
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.2785747339194813e-05,
+      "loss": 0.7969,
+      "step": 1417
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.2808884775566866e-05,
+      "loss": 0.7415,
+      "step": 1418
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.283202221193892e-05,
+      "loss": 0.6091,
+      "step": 1419
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.2855159648310966e-05,
+      "loss": 1.2128,
+      "step": 1420
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.287829708468302e-05,
+      "loss": 1.0642,
+      "step": 1421
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.290143452105507e-05,
+      "loss": 0.7696,
+      "step": 1422
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.292457195742712e-05,
+      "loss": 0.7314,
+      "step": 1423
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.294770939379917e-05,
+      "loss": 0.5556,
+      "step": 1424
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.2970846830171224e-05,
+      "loss": 0.8148,
+      "step": 1425
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.299398426654327e-05,
+      "loss": 0.9136,
+      "step": 1426
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.301712170291532e-05,
+      "loss": 0.9459,
+      "step": 1427
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.304025913928737e-05,
+      "loss": 1.0239,
+      "step": 1428
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.3063396575659415e-05,
+      "loss": 1.022,
+      "step": 1429
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.308653401203147e-05,
+      "loss": 0.793,
+      "step": 1430
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.3109671448403514e-05,
+      "loss": 0.7126,
+      "step": 1431
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.313280888477557e-05,
+      "loss": 0.8208,
+      "step": 1432
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.315594632114762e-05,
+      "loss": 0.7638,
+      "step": 1433
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.3179083757519667e-05,
+      "loss": 0.8714,
+      "step": 1434
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.320222119389172e-05,
+      "loss": 0.6922,
+      "step": 1435
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.322535863026377e-05,
+      "loss": 0.8088,
+      "step": 1436
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.324849606663582e-05,
+      "loss": 1.2721,
+      "step": 1437
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.327163350300787e-05,
+      "loss": 0.9247,
+      "step": 1438
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.329477093937992e-05,
+      "loss": 0.766,
+      "step": 1439
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.3317908375751964e-05,
+      "loss": 1.1769,
+      "step": 1440
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.334104581212402e-05,
+      "loss": 0.8582,
+      "step": 1441
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.336418324849607e-05,
+      "loss": 1.1919,
+      "step": 1442
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.3387320684868116e-05,
+      "loss": 0.9247,
+      "step": 1443
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.341045812124017e-05,
+      "loss": 0.8954,
+      "step": 1444
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.3433595557612215e-05,
+      "loss": 1.0226,
+      "step": 1445
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.345673299398427e-05,
+      "loss": 1.0394,
+      "step": 1446
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.347987043035632e-05,
+      "loss": 0.7955,
+      "step": 1447
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.350300786672837e-05,
+      "loss": 0.9673,
+      "step": 1448
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.352614530310042e-05,
+      "loss": 0.735,
+      "step": 1449
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.3549282739472473e-05,
+      "loss": 1.0003,
+      "step": 1450
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.357242017584451e-05,
+      "loss": 0.8314,
+      "step": 1451
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.3595557612216566e-05,
+      "loss": 0.923,
+      "step": 1452
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.361869504858862e-05,
+      "loss": 1.0048,
+      "step": 1453
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.3641832484960665e-05,
+      "loss": 0.8247,
+      "step": 1454
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.366496992133272e-05,
+      "loss": 0.7324,
+      "step": 1455
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.368810735770477e-05,
+      "loss": 0.8027,
+      "step": 1456
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.371124479407682e-05,
+      "loss": 0.902,
+      "step": 1457
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.373438223044887e-05,
+      "loss": 0.6032,
+      "step": 1458
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.3757519666820916e-05,
+      "loss": 0.9024,
+      "step": 1459
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.378065710319297e-05,
+      "loss": 0.724,
+      "step": 1460
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.380379453956502e-05,
+      "loss": 0.7535,
+      "step": 1461
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.382693197593707e-05,
+      "loss": 0.9003,
+      "step": 1462
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.3850069412309115e-05,
+      "loss": 0.785,
+      "step": 1463
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.387320684868117e-05,
+      "loss": 1.5132,
+      "step": 1464
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.3896344285053214e-05,
+      "loss": 0.4963,
+      "step": 1465
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.391948172142527e-05,
+      "loss": 1.0914,
+      "step": 1466
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.394261915779732e-05,
+      "loss": 0.7022,
+      "step": 1467
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.3965756594169366e-05,
+      "loss": 0.796,
+      "step": 1468
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.398889403054142e-05,
+      "loss": 1.0747,
+      "step": 1469
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.401203146691347e-05,
+      "loss": 0.7173,
+      "step": 1470
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.403516890328552e-05,
+      "loss": 0.9357,
+      "step": 1471
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.405830633965757e-05,
+      "loss": 1.2923,
+      "step": 1472
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.408144377602962e-05,
+      "loss": 1.1185,
+      "step": 1473
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.4104581212401663e-05,
+      "loss": 0.9665,
+      "step": 1474
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.4127718648773716e-05,
+      "loss": 0.8355,
+      "step": 1475
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.415085608514576e-05,
+      "loss": 1.0974,
+      "step": 1476
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.4173993521517816e-05,
+      "loss": 1.0509,
+      "step": 1477
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.419713095788987e-05,
+      "loss": 0.8141,
+      "step": 1478
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.4220268394261915e-05,
+      "loss": 0.7299,
+      "step": 1479
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.424340583063397e-05,
+      "loss": 0.8662,
+      "step": 1480
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.426654326700602e-05,
+      "loss": 0.6164,
+      "step": 1481
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.428968070337807e-05,
+      "loss": 0.8849,
+      "step": 1482
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.431281813975012e-05,
+      "loss": 0.931,
+      "step": 1483
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.433595557612217e-05,
+      "loss": 0.8916,
+      "step": 1484
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.435909301249421e-05,
+      "loss": 1.1634,
+      "step": 1485
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.4382230448866265e-05,
+      "loss": 0.5857,
+      "step": 1486
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.440536788523832e-05,
+      "loss": 1.0188,
+      "step": 1487
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.4428505321610364e-05,
+      "loss": 0.9607,
+      "step": 1488
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.445164275798242e-05,
+      "loss": 0.8461,
+      "step": 1489
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.4474780194354464e-05,
+      "loss": 1.021,
+      "step": 1490
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.4497917630726517e-05,
+      "loss": 0.6421,
+      "step": 1491
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.452105506709857e-05,
+      "loss": 0.9398,
+      "step": 1492
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.4544192503470616e-05,
+      "loss": 1.1264,
+      "step": 1493
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.456732993984267e-05,
+      "loss": 0.7175,
+      "step": 1494
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.459046737621472e-05,
+      "loss": 0.8215,
+      "step": 1495
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.461360481258677e-05,
+      "loss": 1.136,
+      "step": 1496
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.4636742248958814e-05,
+      "loss": 1.0529,
+      "step": 1497
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.465987968533087e-05,
+      "loss": 0.9233,
+      "step": 1498
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.468301712170291e-05,
+      "loss": 0.85,
+      "step": 1499
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.4706154558074966e-05,
+      "loss": 0.8422,
+      "step": 1500
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.472929199444702e-05,
+      "loss": 0.8568,
+      "step": 1501
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.4752429430819065e-05,
+      "loss": 1.1711,
+      "step": 1502
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.477556686719112e-05,
+      "loss": 0.8971,
+      "step": 1503
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.4798704303563165e-05,
+      "loss": 0.6858,
+      "step": 1504
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.482184173993522e-05,
+      "loss": 1.0534,
+      "step": 1505
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.484497917630727e-05,
+      "loss": 0.8585,
+      "step": 1506
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.486811661267932e-05,
+      "loss": 0.7974,
+      "step": 1507
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.489125404905137e-05,
+      "loss": 0.838,
+      "step": 1508
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.4914391485423416e-05,
+      "loss": 1.0655,
+      "step": 1509
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.493752892179546e-05,
+      "loss": 1.0569,
+      "step": 1510
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.4960666358167515e-05,
+      "loss": 0.9309,
+      "step": 1511
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.498380379453957e-05,
+      "loss": 0.9369,
+      "step": 1512
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.5006941230911614e-05,
+      "loss": 1.0866,
+      "step": 1513
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.503007866728367e-05,
+      "loss": 0.7247,
+      "step": 1514
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.505321610365572e-05,
+      "loss": 0.7984,
+      "step": 1515
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.5076353540027766e-05,
+      "loss": 1.1203,
+      "step": 1516
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.509949097639982e-05,
+      "loss": 0.6203,
+      "step": 1517
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.5122628412771865e-05,
+      "loss": 0.7482,
+      "step": 1518
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.514576584914392e-05,
+      "loss": 1.1706,
+      "step": 1519
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.5168903285515965e-05,
+      "loss": 0.9335,
+      "step": 1520
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.519204072188802e-05,
+      "loss": 0.965,
+      "step": 1521
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.5215178158260064e-05,
+      "loss": 1.0372,
+      "step": 1522
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.523831559463212e-05,
+      "loss": 0.8274,
+      "step": 1523
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.526145303100416e-05,
+      "loss": 0.8771,
+      "step": 1524
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.5284590467376216e-05,
+      "loss": 1.1933,
+      "step": 1525
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.530772790374827e-05,
+      "loss": 0.8578,
+      "step": 1526
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.5330865340120315e-05,
+      "loss": 1.0015,
+      "step": 1527
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.535400277649237e-05,
+      "loss": 0.8885,
+      "step": 1528
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.537714021286442e-05,
+      "loss": 1.1717,
+      "step": 1529
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.540027764923647e-05,
+      "loss": 0.9203,
+      "step": 1530
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.542341508560852e-05,
+      "loss": 0.8169,
+      "step": 1531
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.5446552521980566e-05,
+      "loss": 0.9418,
+      "step": 1532
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.546968995835261e-05,
+      "loss": 0.8055,
+      "step": 1533
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.5492827394724666e-05,
+      "loss": 1.0647,
+      "step": 1534
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.551596483109672e-05,
+      "loss": 0.4239,
+      "step": 1535
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.5539102267468765e-05,
+      "loss": 0.9036,
+      "step": 1536
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.556223970384082e-05,
+      "loss": 0.7002,
+      "step": 1537
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.5585377140212864e-05,
+      "loss": 0.8219,
+      "step": 1538
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.560851457658492e-05,
+      "loss": 0.6694,
+      "step": 1539
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.563165201295697e-05,
+      "loss": 0.8467,
+      "step": 1540
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.5654789449329016e-05,
+      "loss": 0.8995,
+      "step": 1541
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.567792688570107e-05,
+      "loss": 1.1137,
+      "step": 1542
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.5701064322073115e-05,
+      "loss": 0.9434,
+      "step": 1543
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.572420175844516e-05,
+      "loss": 0.5241,
+      "step": 1544
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.5747339194817214e-05,
+      "loss": 1.0383,
+      "step": 1545
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.577047663118927e-05,
+      "loss": 0.8461,
+      "step": 1546
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.5793614067561314e-05,
+      "loss": 0.7481,
+      "step": 1547
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.5816751503933367e-05,
+      "loss": 1.0669,
+      "step": 1548
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.583988894030541e-05,
+      "loss": 0.9198,
+      "step": 1549
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.5863026376677466e-05,
+      "loss": 0.7552,
+      "step": 1550
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.588616381304952e-05,
+      "loss": 1.059,
+      "step": 1551
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.5909301249421565e-05,
+      "loss": 1.1653,
+      "step": 1552
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.593243868579362e-05,
+      "loss": 1.0655,
+      "step": 1553
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.595557612216567e-05,
+      "loss": 0.8143,
+      "step": 1554
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.597871355853771e-05,
+      "loss": 0.9544,
+      "step": 1555
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.600185099490976e-05,
+      "loss": 0.8747,
+      "step": 1556
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.6024988431281816e-05,
+      "loss": 0.9812,
+      "step": 1557
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.604812586765386e-05,
+      "loss": 1.2049,
+      "step": 1558
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.6071263304025915e-05,
+      "loss": 0.9093,
+      "step": 1559
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.609440074039797e-05,
+      "loss": 0.9996,
+      "step": 1560
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.6117538176770015e-05,
+      "loss": 0.9856,
+      "step": 1561
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.614067561314207e-05,
+      "loss": 1.2293,
+      "step": 1562
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.6163813049514114e-05,
+      "loss": 0.9923,
+      "step": 1563
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.618695048588617e-05,
+      "loss": 0.7462,
+      "step": 1564
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.621008792225822e-05,
+      "loss": 0.7925,
+      "step": 1565
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.6233225358630266e-05,
+      "loss": 1.0265,
+      "step": 1566
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.625636279500231e-05,
+      "loss": 1.0414,
+      "step": 1567
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.6279500231374365e-05,
+      "loss": 0.9631,
+      "step": 1568
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.630263766774641e-05,
+      "loss": 0.8591,
+      "step": 1569
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.6325775104118464e-05,
+      "loss": 0.7599,
+      "step": 1570
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.634891254049052e-05,
+      "loss": 0.8891,
+      "step": 1571
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.637204997686256e-05,
+      "loss": 1.0511,
+      "step": 1572
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.6395187413234616e-05,
+      "loss": 0.9133,
+      "step": 1573
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.641832484960667e-05,
+      "loss": 0.6759,
+      "step": 1574
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.6441462285978716e-05,
+      "loss": 1.0646,
+      "step": 1575
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.646459972235077e-05,
+      "loss": 0.8144,
+      "step": 1576
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.6487737158722815e-05,
+      "loss": 0.922,
+      "step": 1577
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.651087459509486e-05,
+      "loss": 0.8904,
+      "step": 1578
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.6534012031466914e-05,
+      "loss": 1.0025,
+      "step": 1579
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.655714946783897e-05,
+      "loss": 0.8544,
+      "step": 1580
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.658028690421101e-05,
+      "loss": 1.0467,
+      "step": 1581
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.6603424340583066e-05,
+      "loss": 0.8084,
+      "step": 1582
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.662656177695511e-05,
+      "loss": 0.761,
+      "step": 1583
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.6649699213327165e-05,
+      "loss": 0.7044,
+      "step": 1584
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.667283664969922e-05,
+      "loss": 0.8537,
+      "step": 1585
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.6695974086071264e-05,
+      "loss": 0.8301,
+      "step": 1586
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.671911152244332e-05,
+      "loss": 1.0183,
+      "step": 1587
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.674224895881537e-05,
+      "loss": 0.8951,
+      "step": 1588
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.676538639518741e-05,
+      "loss": 0.7527,
+      "step": 1589
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.678852383155946e-05,
+      "loss": 1.0077,
+      "step": 1590
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.6811661267931516e-05,
+      "loss": 1.0904,
+      "step": 1591
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.683479870430356e-05,
+      "loss": 1.2317,
+      "step": 1592
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.6857936140675615e-05,
+      "loss": 0.8309,
+      "step": 1593
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.688107357704767e-05,
+      "loss": 0.7939,
+      "step": 1594
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.6904211013419714e-05,
+      "loss": 0.728,
+      "step": 1595
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.692734844979177e-05,
+      "loss": 1.1121,
+      "step": 1596
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.695048588616381e-05,
+      "loss": 0.8632,
+      "step": 1597
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.6973623322535866e-05,
+      "loss": 0.8101,
+      "step": 1598
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.699676075890792e-05,
+      "loss": 0.6639,
+      "step": 1599
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.7019898195279965e-05,
+      "loss": 1.1523,
+      "step": 1600
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.704303563165201e-05,
+      "loss": 1.2459,
+      "step": 1601
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.7066173068024064e-05,
+      "loss": 0.6986,
+      "step": 1602
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.708931050439611e-05,
+      "loss": 0.7989,
+      "step": 1603
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.7112447940768164e-05,
+      "loss": 1.0451,
+      "step": 1604
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.7135585377140217e-05,
+      "loss": 0.8127,
+      "step": 1605
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.715872281351226e-05,
+      "loss": 1.1339,
+      "step": 1606
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.7181860249884316e-05,
+      "loss": 0.8179,
+      "step": 1607
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.720499768625637e-05,
+      "loss": 0.9227,
+      "step": 1608
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.7228135122628415e-05,
+      "loss": 0.8763,
+      "step": 1609
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.725127255900047e-05,
+      "loss": 0.7556,
+      "step": 1610
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.7274409995372514e-05,
+      "loss": 0.6677,
+      "step": 1611
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.729754743174457e-05,
+      "loss": 1.051,
+      "step": 1612
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.732068486811661e-05,
+      "loss": 0.6961,
+      "step": 1613
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.734382230448866e-05,
+      "loss": 1.1601,
+      "step": 1614
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.736695974086071e-05,
+      "loss": 0.7246,
+      "step": 1615
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.7390097177232765e-05,
+      "loss": 1.1172,
+      "step": 1616
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.741323461360481e-05,
+      "loss": 1.0858,
+      "step": 1617
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.7436372049976865e-05,
+      "loss": 0.6512,
+      "step": 1618
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.745950948634892e-05,
+      "loss": 1.0323,
+      "step": 1619
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.7482646922720964e-05,
+      "loss": 0.9723,
+      "step": 1620
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.750578435909302e-05,
+      "loss": 0.7866,
+      "step": 1621
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.752892179546506e-05,
+      "loss": 0.945,
+      "step": 1622
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.7552059231837116e-05,
+      "loss": 0.8392,
+      "step": 1623
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.757519666820916e-05,
+      "loss": 0.849,
+      "step": 1624
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.7598334104581215e-05,
+      "loss": 1.125,
+      "step": 1625
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.762147154095326e-05,
+      "loss": 1.003,
+      "step": 1626
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.7644608977325314e-05,
+      "loss": 0.9893,
+      "step": 1627
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.766774641369736e-05,
+      "loss": 0.6976,
+      "step": 1628
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.7690883850069413e-05,
+      "loss": 0.8016,
+      "step": 1629
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.7714021286441466e-05,
+      "loss": 0.8302,
+      "step": 1630
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.773715872281351e-05,
+      "loss": 0.9258,
+      "step": 1631
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.7760296159185566e-05,
+      "loss": 0.8524,
+      "step": 1632
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.778343359555762e-05,
+      "loss": 0.7462,
+      "step": 1633
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.7806571031929665e-05,
+      "loss": 0.7668,
+      "step": 1634
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.782970846830172e-05,
+      "loss": 1.015,
+      "step": 1635
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.7852845904673764e-05,
+      "loss": 0.6353,
+      "step": 1636
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.787598334104581e-05,
+      "loss": 0.7312,
+      "step": 1637
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.789912077741786e-05,
+      "loss": 0.6865,
+      "step": 1638
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.7922258213789916e-05,
+      "loss": 0.6137,
+      "step": 1639
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.794539565016196e-05,
+      "loss": 0.7458,
+      "step": 1640
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.7968533086534015e-05,
+      "loss": 0.8757,
+      "step": 1641
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.799167052290606e-05,
+      "loss": 0.9785,
+      "step": 1642
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.8014807959278114e-05,
+      "loss": 0.6802,
+      "step": 1643
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.803794539565017e-05,
+      "loss": 0.6544,
+      "step": 1644
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.8061082832022214e-05,
+      "loss": 1.0145,
+      "step": 1645
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.8084220268394267e-05,
+      "loss": 0.9302,
+      "step": 1646
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.810735770476631e-05,
+      "loss": 0.9289,
+      "step": 1647
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.813049514113836e-05,
+      "loss": 0.9519,
+      "step": 1648
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.815363257751041e-05,
+      "loss": 0.963,
+      "step": 1649
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.8176770013882465e-05,
+      "loss": 0.7789,
+      "step": 1650
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.819990745025451e-05,
+      "loss": 0.7597,
+      "step": 1651
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.8223044886626564e-05,
+      "loss": 1.0507,
+      "step": 1652
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.824618232299862e-05,
+      "loss": 0.9675,
+      "step": 1653
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.826931975937066e-05,
+      "loss": 0.8547,
+      "step": 1654
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.8292457195742716e-05,
+      "loss": 1.0886,
+      "step": 1655
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.831559463211476e-05,
+      "loss": 1.0111,
+      "step": 1656
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.8338732068486815e-05,
+      "loss": 0.8659,
+      "step": 1657
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.836186950485887e-05,
+      "loss": 0.859,
+      "step": 1658
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.838500694123091e-05,
+      "loss": 0.9923,
+      "step": 1659
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.840814437760296e-05,
+      "loss": 0.9836,
+      "step": 1660
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.8431281813975014e-05,
+      "loss": 0.8309,
+      "step": 1661
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.845441925034706e-05,
+      "loss": 0.7139,
+      "step": 1662
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.847755668671911e-05,
+      "loss": 0.7446,
+      "step": 1663
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.8500694123091166e-05,
+      "loss": 0.8448,
+      "step": 1664
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.852383155946321e-05,
+      "loss": 0.8589,
+      "step": 1665
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.8546968995835265e-05,
+      "loss": 1.0794,
+      "step": 1666
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.857010643220732e-05,
+      "loss": 0.8194,
+      "step": 1667
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.8593243868579364e-05,
+      "loss": 0.5254,
+      "step": 1668
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.861638130495142e-05,
+      "loss": 0.7362,
+      "step": 1669
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.863951874132346e-05,
+      "loss": 0.4489,
+      "step": 1670
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.866265617769551e-05,
+      "loss": 0.6865,
+      "step": 1671
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.868579361406756e-05,
+      "loss": 0.9161,
+      "step": 1672
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.870893105043961e-05,
+      "loss": 1.0003,
+      "step": 1673
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.873206848681166e-05,
+      "loss": 0.6785,
+      "step": 1674
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.8755205923183715e-05,
+      "loss": 1.2009,
+      "step": 1675
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.877834335955576e-05,
+      "loss": 0.7446,
+      "step": 1676
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.8801480795927814e-05,
+      "loss": 0.9351,
+      "step": 1677
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.882461823229987e-05,
+      "loss": 0.7055,
+      "step": 1678
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.884775566867191e-05,
+      "loss": 0.6497,
+      "step": 1679
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.8870893105043966e-05,
+      "loss": 0.621,
+      "step": 1680
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.889403054141602e-05,
+      "loss": 0.6865,
+      "step": 1681
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.891716797778806e-05,
+      "loss": 0.7704,
+      "step": 1682
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.894030541416011e-05,
+      "loss": 0.8163,
+      "step": 1683
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.8963442850532164e-05,
+      "loss": 1.1199,
+      "step": 1684
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.898658028690421e-05,
+      "loss": 0.6319,
+      "step": 1685
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.9009717723276263e-05,
+      "loss": 0.6416,
+      "step": 1686
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.903285515964831e-05,
+      "loss": 0.8021,
+      "step": 1687
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.905599259602036e-05,
+      "loss": 0.7985,
+      "step": 1688
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.9079130032392416e-05,
+      "loss": 0.7313,
+      "step": 1689
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.910226746876446e-05,
+      "loss": 0.9323,
+      "step": 1690
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.9125404905136515e-05,
+      "loss": 1.0701,
+      "step": 1691
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.914854234150857e-05,
+      "loss": 0.7199,
+      "step": 1692
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.917167977788061e-05,
+      "loss": 0.8846,
+      "step": 1693
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.919481721425266e-05,
+      "loss": 0.7677,
+      "step": 1694
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.921795465062471e-05,
+      "loss": 0.6825,
+      "step": 1695
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.924109208699676e-05,
+      "loss": 0.4492,
+      "step": 1696
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.926422952336881e-05,
+      "loss": 0.9207,
+      "step": 1697
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.9287366959740865e-05,
+      "loss": 1.3896,
+      "step": 1698
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.931050439611291e-05,
+      "loss": 0.9506,
+      "step": 1699
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.9333641832484964e-05,
+      "loss": 0.5903,
+      "step": 1700
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.935677926885701e-05,
+      "loss": 0.6777,
+      "step": 1701
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.9379916705229064e-05,
+      "loss": 0.6791,
+      "step": 1702
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.9403054141601117e-05,
+      "loss": 1.0171,
+      "step": 1703
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.942619157797316e-05,
+      "loss": 0.8692,
+      "step": 1704
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.944932901434521e-05,
+      "loss": 1.0557,
+      "step": 1705
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.947246645071726e-05,
+      "loss": 0.6875,
+      "step": 1706
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.949560388708931e-05,
+      "loss": 0.7659,
+      "step": 1707
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.951874132346136e-05,
+      "loss": 0.6578,
+      "step": 1708
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.9541878759833414e-05,
+      "loss": 0.79,
+      "step": 1709
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.956501619620546e-05,
+      "loss": 1.0307,
+      "step": 1710
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.958815363257751e-05,
+      "loss": 1.0125,
+      "step": 1711
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.9611291068949566e-05,
+      "loss": 1.066,
+      "step": 1712
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.963442850532161e-05,
+      "loss": 1.4327,
+      "step": 1713
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.9657565941693665e-05,
+      "loss": 0.9419,
+      "step": 1714
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.968070337806571e-05,
+      "loss": 0.9237,
+      "step": 1715
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.9703840814437765e-05,
+      "loss": 0.6553,
+      "step": 1716
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.972697825080981e-05,
+      "loss": 0.9723,
+      "step": 1717
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.975011568718186e-05,
+      "loss": 0.936,
+      "step": 1718
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.977325312355391e-05,
+      "loss": 1.0948,
+      "step": 1719
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.979639055992596e-05,
+      "loss": 1.034,
+      "step": 1720
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.981952799629801e-05,
+      "loss": 1.018,
+      "step": 1721
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.984266543267006e-05,
+      "loss": 0.8374,
+      "step": 1722
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.9865802869042115e-05,
+      "loss": 0.9044,
+      "step": 1723
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.988894030541416e-05,
+      "loss": 0.8748,
+      "step": 1724
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.9912077741786214e-05,
+      "loss": 0.8235,
+      "step": 1725
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.993521517815827e-05,
+      "loss": 1.2558,
+      "step": 1726
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.995835261453031e-05,
+      "loss": 0.8198,
+      "step": 1727
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.998149005090236e-05,
+      "loss": 0.8576,
+      "step": 1728
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.000462748727441e-05,
+      "loss": 0.9787,
+      "step": 1729
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.002776492364646e-05,
+      "loss": 0.5124,
+      "step": 1730
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.005090236001851e-05,
+      "loss": 0.9693,
+      "step": 1731
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.007403979639056e-05,
+      "loss": 0.6236,
+      "step": 1732
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.009717723276261e-05,
+      "loss": 0.5781,
+      "step": 1733
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.0120314669134664e-05,
+      "loss": 1.0959,
+      "step": 1734
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.014345210550671e-05,
+      "loss": 0.9999,
+      "step": 1735
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.016658954187876e-05,
+      "loss": 0.9354,
+      "step": 1736
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.0189726978250816e-05,
+      "loss": 0.9216,
+      "step": 1737
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.021286441462286e-05,
+      "loss": 0.8602,
+      "step": 1738
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.0236001850994915e-05,
+      "loss": 0.9394,
+      "step": 1739
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.025913928736696e-05,
+      "loss": 0.8144,
+      "step": 1740
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.028227672373901e-05,
+      "loss": 0.9743,
+      "step": 1741
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.030541416011106e-05,
+      "loss": 0.7908,
+      "step": 1742
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.0328551596483113e-05,
+      "loss": 0.9739,
+      "step": 1743
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.035168903285516e-05,
+      "loss": 0.6305,
+      "step": 1744
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.037482646922721e-05,
+      "loss": 0.9586,
+      "step": 1745
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.039796390559926e-05,
+      "loss": 0.9299,
+      "step": 1746
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.042110134197131e-05,
+      "loss": 0.7716,
+      "step": 1747
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.0444238778343365e-05,
+      "loss": 0.9247,
+      "step": 1748
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.046737621471541e-05,
+      "loss": 0.8215,
+      "step": 1749
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.0490513651087464e-05,
+      "loss": 0.678,
+      "step": 1750
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.051365108745951e-05,
+      "loss": 0.8502,
+      "step": 1751
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.0536788523831556e-05,
+      "loss": 0.764,
+      "step": 1752
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.055992596020361e-05,
+      "loss": 1.2172,
+      "step": 1753
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.058306339657566e-05,
+      "loss": 0.771,
+      "step": 1754
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.060620083294771e-05,
+      "loss": 0.7198,
+      "step": 1755
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.062933826931976e-05,
+      "loss": 1.0228,
+      "step": 1756
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.0652475705691814e-05,
+      "loss": 0.9291,
+      "step": 1757
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.067561314206386e-05,
+      "loss": 1.0193,
+      "step": 1758
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.0698750578435914e-05,
+      "loss": 0.728,
+      "step": 1759
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.072188801480796e-05,
+      "loss": 0.9953,
+      "step": 1760
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.074502545118001e-05,
+      "loss": 0.631,
+      "step": 1761
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.0768162887552066e-05,
+      "loss": 0.6676,
+      "step": 1762
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.079130032392411e-05,
+      "loss": 1.2396,
+      "step": 1763
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.081443776029616e-05,
+      "loss": 0.7846,
+      "step": 1764
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.083757519666821e-05,
+      "loss": 0.8235,
+      "step": 1765
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.086071263304026e-05,
+      "loss": 1.1941,
+      "step": 1766
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.088385006941231e-05,
+      "loss": 0.6768,
+      "step": 1767
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.090698750578436e-05,
+      "loss": 0.98,
+      "step": 1768
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.093012494215641e-05,
+      "loss": 1.0484,
+      "step": 1769
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.095326237852846e-05,
+      "loss": 1.0801,
+      "step": 1770
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.0976399814900515e-05,
+      "loss": 1.3685,
+      "step": 1771
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.099953725127256e-05,
+      "loss": 0.8642,
+      "step": 1772
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.1022674687644615e-05,
+      "loss": 0.9494,
+      "step": 1773
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.104581212401666e-05,
+      "loss": 0.9138,
+      "step": 1774
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.106894956038871e-05,
+      "loss": 0.7234,
+      "step": 1775
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.109208699676076e-05,
+      "loss": 0.7122,
+      "step": 1776
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.111522443313281e-05,
+      "loss": 0.6492,
+      "step": 1777
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.113836186950486e-05,
+      "loss": 1.0178,
+      "step": 1778
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.116149930587691e-05,
+      "loss": 0.9841,
+      "step": 1779
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.118463674224896e-05,
+      "loss": 0.8983,
+      "step": 1780
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.120777417862101e-05,
+      "loss": 0.9141,
+      "step": 1781
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.1230911614993064e-05,
+      "loss": 1.0303,
+      "step": 1782
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.125404905136511e-05,
+      "loss": 0.7545,
+      "step": 1783
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.127718648773716e-05,
+      "loss": 0.8564,
+      "step": 1784
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.1300323924109216e-05,
+      "loss": 0.8142,
+      "step": 1785
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.1323461360481256e-05,
+      "loss": 0.8622,
+      "step": 1786
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.134659879685331e-05,
+      "loss": 1.0066,
+      "step": 1787
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.136973623322536e-05,
+      "loss": 1.2333,
+      "step": 1788
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.139287366959741e-05,
+      "loss": 0.6061,
+      "step": 1789
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.141601110596946e-05,
+      "loss": 1.105,
+      "step": 1790
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.143914854234151e-05,
+      "loss": 0.9181,
+      "step": 1791
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.146228597871356e-05,
+      "loss": 0.463,
+      "step": 1792
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.148542341508561e-05,
+      "loss": 0.9876,
+      "step": 1793
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.150856085145766e-05,
+      "loss": 0.9102,
+      "step": 1794
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.153169828782971e-05,
+      "loss": 0.99,
+      "step": 1795
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.1554835724201765e-05,
+      "loss": 0.9241,
+      "step": 1796
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.1577973160573805e-05,
+      "loss": 0.85,
+      "step": 1797
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.160111059694586e-05,
+      "loss": 0.8802,
+      "step": 1798
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.162424803331791e-05,
+      "loss": 0.6042,
+      "step": 1799
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.164738546968996e-05,
+      "loss": 0.6855,
+      "step": 1800
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.167052290606201e-05,
+      "loss": 0.8838,
+      "step": 1801
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.169366034243406e-05,
+      "loss": 0.6758,
+      "step": 1802
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.171679777880611e-05,
+      "loss": 0.8292,
+      "step": 1803
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.173993521517816e-05,
+      "loss": 0.906,
+      "step": 1804
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.176307265155021e-05,
+      "loss": 0.96,
+      "step": 1805
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.178621008792226e-05,
+      "loss": 0.8058,
+      "step": 1806
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.1809347524294314e-05,
+      "loss": 1.0917,
+      "step": 1807
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.183248496066636e-05,
+      "loss": 0.8853,
+      "step": 1808
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.1855622397038406e-05,
+      "loss": 0.8226,
+      "step": 1809
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.187875983341046e-05,
+      "loss": 0.6393,
+      "step": 1810
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.1901897269782506e-05,
+      "loss": 1.0088,
+      "step": 1811
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.192503470615456e-05,
+      "loss": 1.2596,
+      "step": 1812
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.194817214252661e-05,
+      "loss": 0.6756,
+      "step": 1813
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.197130957889866e-05,
+      "loss": 1.1208,
+      "step": 1814
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.199444701527071e-05,
+      "loss": 1.0226,
+      "step": 1815
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.2017584451642764e-05,
+      "loss": 0.9812,
+      "step": 1816
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.204072188801481e-05,
+      "loss": 0.8446,
+      "step": 1817
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.206385932438686e-05,
+      "loss": 0.8617,
+      "step": 1818
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.208699676075891e-05,
+      "loss": 0.841,
+      "step": 1819
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.211013419713096e-05,
+      "loss": 0.4887,
+      "step": 1820
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.213327163350301e-05,
+      "loss": 0.7578,
+      "step": 1821
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.215640906987506e-05,
+      "loss": 0.8403,
+      "step": 1822
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.217954650624711e-05,
+      "loss": 0.7042,
+      "step": 1823
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.220268394261916e-05,
+      "loss": 0.9044,
+      "step": 1824
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.2225821378991206e-05,
+      "loss": 0.8079,
+      "step": 1825
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.224895881536326e-05,
+      "loss": 0.8498,
+      "step": 1826
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.227209625173531e-05,
+      "loss": 0.6827,
+      "step": 1827
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.229523368810736e-05,
+      "loss": 0.9415,
+      "step": 1828
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.231837112447941e-05,
+      "loss": 0.8596,
+      "step": 1829
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.2341508560851465e-05,
+      "loss": 0.98,
+      "step": 1830
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.236464599722351e-05,
+      "loss": 0.9815,
+      "step": 1831
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.238778343359556e-05,
+      "loss": 0.7294,
+      "step": 1832
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.241092086996761e-05,
+      "loss": 0.8684,
+      "step": 1833
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.2434058306339656e-05,
+      "loss": 0.9112,
+      "step": 1834
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.245719574271171e-05,
+      "loss": 0.8061,
+      "step": 1835
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.248033317908376e-05,
+      "loss": 0.8087,
+      "step": 1836
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.250347061545581e-05,
+      "loss": 0.9755,
+      "step": 1837
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.252660805182786e-05,
+      "loss": 0.8912,
+      "step": 1838
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.254974548819991e-05,
+      "loss": 0.5805,
+      "step": 1839
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.257288292457196e-05,
+      "loss": 0.8427,
+      "step": 1840
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.2596020360944013e-05,
+      "loss": 1.1116,
+      "step": 1841
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.261915779731606e-05,
+      "loss": 1.0641,
+      "step": 1842
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.264229523368811e-05,
+      "loss": 0.7193,
+      "step": 1843
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.266543267006016e-05,
+      "loss": 0.9294,
+      "step": 1844
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.2688570106432205e-05,
+      "loss": 0.8081,
+      "step": 1845
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.271170754280426e-05,
+      "loss": 0.5892,
+      "step": 1846
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.273484497917631e-05,
+      "loss": 1.0871,
+      "step": 1847
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.275798241554836e-05,
+      "loss": 0.7842,
+      "step": 1848
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.278111985192041e-05,
+      "loss": 0.6885,
+      "step": 1849
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.280425728829246e-05,
+      "loss": 0.6406,
+      "step": 1850
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.282739472466451e-05,
+      "loss": 0.9717,
+      "step": 1851
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.285053216103656e-05,
+      "loss": 0.6913,
+      "step": 1852
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.287366959740861e-05,
+      "loss": 0.714,
+      "step": 1853
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.289680703378066e-05,
+      "loss": 0.7341,
+      "step": 1854
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.291994447015271e-05,
+      "loss": 0.8412,
+      "step": 1855
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.2943081906524754e-05,
+      "loss": 0.9623,
+      "step": 1856
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.296621934289681e-05,
+      "loss": 0.8553,
+      "step": 1857
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.298935677926886e-05,
+      "loss": 1.0334,
+      "step": 1858
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.3012494215640906e-05,
+      "loss": 0.9079,
+      "step": 1859
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.303563165201296e-05,
+      "loss": 1.0721,
+      "step": 1860
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.305876908838501e-05,
+      "loss": 1.0393,
+      "step": 1861
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.308190652475706e-05,
+      "loss": 0.7158,
+      "step": 1862
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.310504396112911e-05,
+      "loss": 0.8753,
+      "step": 1863
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.312818139750116e-05,
+      "loss": 0.8829,
+      "step": 1864
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.315131883387321e-05,
+      "loss": 0.9548,
+      "step": 1865
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.317445627024526e-05,
+      "loss": 0.7451,
+      "step": 1866
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.319759370661731e-05,
+      "loss": 0.8909,
+      "step": 1867
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.3220731142989356e-05,
+      "loss": 0.6779,
+      "step": 1868
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.324386857936141e-05,
+      "loss": 0.6971,
+      "step": 1869
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.3267006015733455e-05,
+      "loss": 0.8626,
+      "step": 1870
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.329014345210551e-05,
+      "loss": 1.01,
+      "step": 1871
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.331328088847756e-05,
+      "loss": 0.9603,
+      "step": 1872
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.333641832484961e-05,
+      "loss": 0.6386,
+      "step": 1873
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.335955576122166e-05,
+      "loss": 0.8444,
+      "step": 1874
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.338269319759371e-05,
+      "loss": 1.0642,
+      "step": 1875
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.340583063396576e-05,
+      "loss": 1.0149,
+      "step": 1876
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.342896807033781e-05,
+      "loss": 0.6489,
+      "step": 1877
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.345210550670986e-05,
+      "loss": 0.684,
+      "step": 1878
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.3475242943081904e-05,
+      "loss": 1.0372,
+      "step": 1879
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.349838037945396e-05,
+      "loss": 1.0302,
+      "step": 1880
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.352151781582601e-05,
+      "loss": 0.8004,
+      "step": 1881
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.3544655252198057e-05,
+      "loss": 0.9733,
+      "step": 1882
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.356779268857011e-05,
+      "loss": 1.1008,
+      "step": 1883
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.3590930124942156e-05,
+      "loss": 0.9961,
+      "step": 1884
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.361406756131421e-05,
+      "loss": 0.8715,
+      "step": 1885
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.363720499768626e-05,
+      "loss": 0.6775,
+      "step": 1886
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.366034243405831e-05,
+      "loss": 0.929,
+      "step": 1887
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.368347987043036e-05,
+      "loss": 1.1079,
+      "step": 1888
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.3706617306802414e-05,
+      "loss": 1.1502,
+      "step": 1889
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.372975474317445e-05,
+      "loss": 0.88,
+      "step": 1890
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.3752892179546506e-05,
+      "loss": 0.6682,
+      "step": 1891
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.377602961591856e-05,
+      "loss": 0.9755,
+      "step": 1892
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.3799167052290605e-05,
+      "loss": 0.7602,
+      "step": 1893
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.382230448866266e-05,
+      "loss": 0.8984,
+      "step": 1894
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.384544192503471e-05,
+      "loss": 0.9517,
+      "step": 1895
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.386857936140676e-05,
+      "loss": 0.604,
+      "step": 1896
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.389171679777881e-05,
+      "loss": 0.5887,
+      "step": 1897
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.391485423415086e-05,
+      "loss": 1.2132,
+      "step": 1898
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.393799167052291e-05,
+      "loss": 0.9513,
+      "step": 1899
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.396112910689496e-05,
+      "loss": 0.4289,
+      "step": 1900
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.3984266543267e-05,
+      "loss": 1.2336,
+      "step": 1901
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.4007403979639055e-05,
+      "loss": 1.0498,
+      "step": 1902
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.403054141601111e-05,
+      "loss": 0.9651,
+      "step": 1903
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.4053678852383154e-05,
+      "loss": 0.6762,
+      "step": 1904
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.407681628875521e-05,
+      "loss": 0.8161,
+      "step": 1905
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.409995372512726e-05,
+      "loss": 0.4459,
+      "step": 1906
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.4123091161499306e-05,
+      "loss": 0.913,
+      "step": 1907
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.414622859787136e-05,
+      "loss": 1.1976,
+      "step": 1908
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.416936603424341e-05,
+      "loss": 1.0287,
+      "step": 1909
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.419250347061546e-05,
+      "loss": 1.1099,
+      "step": 1910
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.421564090698751e-05,
+      "loss": 0.813,
+      "step": 1911
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.423877834335956e-05,
+      "loss": 0.9204,
+      "step": 1912
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.4261915779731604e-05,
+      "loss": 0.6646,
+      "step": 1913
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.428505321610366e-05,
+      "loss": 0.6852,
+      "step": 1914
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.43081906524757e-05,
+      "loss": 1.0788,
+      "step": 1915
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.4331328088847756e-05,
+      "loss": 0.9663,
+      "step": 1916
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.435446552521981e-05,
+      "loss": 0.8841,
+      "step": 1917
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.4377602961591855e-05,
+      "loss": 0.6784,
+      "step": 1918
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.440074039796391e-05,
+      "loss": 0.739,
+      "step": 1919
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.442387783433596e-05,
+      "loss": 0.8474,
+      "step": 1920
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.444701527070801e-05,
+      "loss": 0.7768,
+      "step": 1921
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.447015270708006e-05,
+      "loss": 0.8917,
+      "step": 1922
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.449329014345211e-05,
+      "loss": 0.9284,
+      "step": 1923
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.451642757982416e-05,
+      "loss": 0.7173,
+      "step": 1924
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.4539565016196206e-05,
+      "loss": 0.6391,
+      "step": 1925
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.456270245256826e-05,
+      "loss": 1.0426,
+      "step": 1926
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.4585839888940305e-05,
+      "loss": 0.6401,
+      "step": 1927
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.460897732531236e-05,
+      "loss": 0.857,
+      "step": 1928
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.4632114761684404e-05,
+      "loss": 0.7412,
+      "step": 1929
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.465525219805646e-05,
+      "loss": 0.7429,
+      "step": 1930
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.467838963442851e-05,
+      "loss": 0.7416,
+      "step": 1931
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.4701527070800556e-05,
+      "loss": 0.774,
+      "step": 1932
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.472466450717261e-05,
+      "loss": 0.64,
+      "step": 1933
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.474780194354466e-05,
+      "loss": 0.9058,
+      "step": 1934
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.477093937991671e-05,
+      "loss": 0.6252,
+      "step": 1935
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.4794076816288754e-05,
+      "loss": 0.802,
+      "step": 1936
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.481721425266081e-05,
+      "loss": 1.1985,
+      "step": 1937
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.4840351689032854e-05,
+      "loss": 0.8751,
+      "step": 1938
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.4863489125404907e-05,
+      "loss": 0.8906,
+      "step": 1939
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.488662656177696e-05,
+      "loss": 0.8694,
+      "step": 1940
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.4909763998149006e-05,
+      "loss": 1.443,
+      "step": 1941
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.493290143452106e-05,
+      "loss": 1.0475,
+      "step": 1942
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.4956038870893105e-05,
+      "loss": 0.7117,
+      "step": 1943
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.497917630726516e-05,
+      "loss": 0.9276,
+      "step": 1944
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.500231374363721e-05,
+      "loss": 0.7955,
+      "step": 1945
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.502545118000926e-05,
+      "loss": 0.8843,
+      "step": 1946
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.504858861638131e-05,
+      "loss": 0.7538,
+      "step": 1947
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.5071726052753356e-05,
+      "loss": 0.8148,
+      "step": 1948
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.50948634891254e-05,
+      "loss": 0.9373,
+      "step": 1949
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.5118000925497455e-05,
+      "loss": 0.6908,
+      "step": 1950
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.514113836186951e-05,
+      "loss": 0.9836,
+      "step": 1951
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.5164275798241555e-05,
+      "loss": 1.2134,
+      "step": 1952
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.518741323461361e-05,
+      "loss": 1.0649,
+      "step": 1953
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.521055067098566e-05,
+      "loss": 1.0711,
+      "step": 1954
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.523368810735771e-05,
+      "loss": 0.7114,
+      "step": 1955
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.525682554372976e-05,
+      "loss": 0.8058,
+      "step": 1956
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.5279962980101806e-05,
+      "loss": 0.8619,
+      "step": 1957
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.530310041647386e-05,
+      "loss": 0.9599,
+      "step": 1958
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.5326237852845905e-05,
+      "loss": 0.619,
+      "step": 1959
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.534937528921796e-05,
+      "loss": 0.758,
+      "step": 1960
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.5372512725590004e-05,
+      "loss": 1.0637,
+      "step": 1961
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.539565016196206e-05,
+      "loss": 0.6132,
+      "step": 1962
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.54187875983341e-05,
+      "loss": 0.6735,
+      "step": 1963
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.5441925034706156e-05,
+      "loss": 0.801,
+      "step": 1964
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.546506247107821e-05,
+      "loss": 0.7446,
+      "step": 1965
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.5488199907450256e-05,
+      "loss": 1.0268,
+      "step": 1966
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.551133734382231e-05,
+      "loss": 0.8231,
+      "step": 1967
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.553447478019436e-05,
+      "loss": 0.6734,
+      "step": 1968
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.555761221656641e-05,
+      "loss": 0.6912,
+      "step": 1969
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.558074965293846e-05,
+      "loss": 1.0237,
+      "step": 1970
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.560388708931051e-05,
+      "loss": 0.8015,
+      "step": 1971
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.562702452568255e-05,
+      "loss": 0.7081,
+      "step": 1972
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.5650161962054606e-05,
+      "loss": 0.5965,
+      "step": 1973
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.567329939842665e-05,
+      "loss": 1.0984,
+      "step": 1974
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.5696436834798705e-05,
+      "loss": 0.8609,
+      "step": 1975
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.571957427117076e-05,
+      "loss": 0.8049,
+      "step": 1976
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.5742711707542804e-05,
+      "loss": 0.749,
+      "step": 1977
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.576584914391486e-05,
+      "loss": 0.486,
+      "step": 1978
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.578898658028691e-05,
+      "loss": 0.7741,
+      "step": 1979
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.5812124016658956e-05,
+      "loss": 1.0349,
+      "step": 1980
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.583526145303101e-05,
+      "loss": 0.9529,
+      "step": 1981
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.5858398889403056e-05,
+      "loss": 0.6298,
+      "step": 1982
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.58815363257751e-05,
+      "loss": 1.1373,
+      "step": 1983
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.5904673762147155e-05,
+      "loss": 0.8886,
+      "step": 1984
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.592781119851921e-05,
+      "loss": 1.0553,
+      "step": 1985
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.5950948634891254e-05,
+      "loss": 1.0199,
+      "step": 1986
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.597408607126331e-05,
+      "loss": 1.1894,
+      "step": 1987
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.599722350763535e-05,
+      "loss": 1.0267,
+      "step": 1988
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.6020360944007406e-05,
+      "loss": 0.8349,
+      "step": 1989
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.604349838037946e-05,
+      "loss": 0.6227,
+      "step": 1990
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.6066635816751505e-05,
+      "loss": 0.9376,
+      "step": 1991
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.608977325312356e-05,
+      "loss": 1.0732,
+      "step": 1992
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.611291068949561e-05,
+      "loss": 0.9149,
+      "step": 1993
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.613604812586765e-05,
+      "loss": 0.9279,
+      "step": 1994
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.6159185562239704e-05,
+      "loss": 0.9647,
+      "step": 1995
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.6182322998611757e-05,
+      "loss": 0.8883,
+      "step": 1996
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.62054604349838e-05,
+      "loss": 0.7122,
+      "step": 1997
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.6228597871355856e-05,
+      "loss": 0.7385,
+      "step": 1998
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.625173530772791e-05,
+      "loss": 1.1913,
+      "step": 1999
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.6274872744099955e-05,
+      "loss": 0.6978,
+      "step": 2000
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.629801018047201e-05,
+      "loss": 0.963,
+      "step": 2001
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.6321147616844054e-05,
+      "loss": 0.9865,
+      "step": 2002
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.634428505321611e-05,
+      "loss": 0.9919,
+      "step": 2003
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.636742248958816e-05,
+      "loss": 1.1666,
+      "step": 2004
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.6390559925960206e-05,
+      "loss": 0.9599,
+      "step": 2005
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.641369736233225e-05,
+      "loss": 0.9353,
+      "step": 2006
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.6436834798704305e-05,
+      "loss": 0.7676,
+      "step": 2007
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.645997223507635e-05,
+      "loss": 0.98,
+      "step": 2008
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.6483109671448405e-05,
+      "loss": 0.7854,
+      "step": 2009
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.650624710782046e-05,
+      "loss": 0.7681,
+      "step": 2010
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.6529384544192504e-05,
+      "loss": 0.8769,
+      "step": 2011
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.655252198056456e-05,
+      "loss": 0.8845,
+      "step": 2012
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.657565941693661e-05,
+      "loss": 0.9725,
+      "step": 2013
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.6598796853308656e-05,
+      "loss": 0.9296,
+      "step": 2014
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.662193428968071e-05,
+      "loss": 1.1596,
+      "step": 2015
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.6645071726052755e-05,
+      "loss": 0.9848,
+      "step": 2016
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.66682091624248e-05,
+      "loss": 0.8661,
+      "step": 2017
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.6691346598796854e-05,
+      "loss": 0.8011,
+      "step": 2018
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.671448403516891e-05,
+      "loss": 0.7467,
+      "step": 2019
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.6737621471540953e-05,
+      "loss": 0.9999,
+      "step": 2020
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.6760758907913006e-05,
+      "loss": 0.6248,
+      "step": 2021
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.678389634428505e-05,
+      "loss": 0.5739,
+      "step": 2022
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.6807033780657106e-05,
+      "loss": 1.2383,
+      "step": 2023
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.683017121702916e-05,
+      "loss": 0.993,
+      "step": 2024
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.6853308653401205e-05,
+      "loss": 0.9799,
+      "step": 2025
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.687644608977326e-05,
+      "loss": 0.9662,
+      "step": 2026
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.689958352614531e-05,
+      "loss": 0.9174,
+      "step": 2027
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.692272096251736e-05,
+      "loss": 0.9519,
+      "step": 2028
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.69458583988894e-05,
+      "loss": 0.8357,
+      "step": 2029
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.6968995835261456e-05,
+      "loss": 0.6842,
+      "step": 2030
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.69921332716335e-05,
+      "loss": 0.8253,
+      "step": 2031
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.7015270708005555e-05,
+      "loss": 1.2159,
+      "step": 2032
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.703840814437761e-05,
+      "loss": 0.9665,
+      "step": 2033
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.7061545580749654e-05,
+      "loss": 0.9188,
+      "step": 2034
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.708468301712171e-05,
+      "loss": 0.6981,
+      "step": 2035
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.7107820453493754e-05,
+      "loss": 0.6873,
+      "step": 2036
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.7130957889865806e-05,
+      "loss": 0.7568,
+      "step": 2037
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.715409532623786e-05,
+      "loss": 0.9423,
+      "step": 2038
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.7177232762609906e-05,
+      "loss": 0.9801,
+      "step": 2039
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.720037019898195e-05,
+      "loss": 0.9946,
+      "step": 2040
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.7223507635354005e-05,
+      "loss": 0.7185,
+      "step": 2041
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.724664507172605e-05,
+      "loss": 0.8387,
+      "step": 2042
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.7269782508098104e-05,
+      "loss": 1.1105,
+      "step": 2043
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.729291994447016e-05,
+      "loss": 0.7113,
+      "step": 2044
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.73160573808422e-05,
+      "loss": 0.839,
+      "step": 2045
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.7339194817214256e-05,
+      "loss": 0.6335,
+      "step": 2046
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.73623322535863e-05,
+      "loss": 0.7803,
+      "step": 2047
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.7385469689958355e-05,
+      "loss": 1.0649,
+      "step": 2048
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.740860712633041e-05,
+      "loss": 0.8359,
+      "step": 2049
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.7431744562702454e-05,
+      "loss": 1.153,
+      "step": 2050
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.745488199907451e-05,
+      "loss": 0.7972,
+      "step": 2051
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.7478019435446554e-05,
+      "loss": 1.2555,
+      "step": 2052
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.75011568718186e-05,
+      "loss": 0.6115,
+      "step": 2053
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.752429430819065e-05,
+      "loss": 1.1857,
+      "step": 2054
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.7547431744562706e-05,
+      "loss": 0.9187,
+      "step": 2055
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.757056918093475e-05,
+      "loss": 0.9186,
+      "step": 2056
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.7593706617306805e-05,
+      "loss": 0.9273,
+      "step": 2057
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.761684405367886e-05,
+      "loss": 0.769,
+      "step": 2058
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.7639981490050904e-05,
+      "loss": 0.8041,
+      "step": 2059
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.766311892642296e-05,
+      "loss": 0.7832,
+      "step": 2060
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.7686256362795e-05,
+      "loss": 0.8485,
+      "step": 2061
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.7709393799167056e-05,
+      "loss": 0.7586,
+      "step": 2062
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.77325312355391e-05,
+      "loss": 0.8851,
+      "step": 2063
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.7755668671911155e-05,
+      "loss": 0.8479,
+      "step": 2064
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.77788061082832e-05,
+      "loss": 1.284,
+      "step": 2065
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.7801943544655255e-05,
+      "loss": 0.899,
+      "step": 2066
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.78250809810273e-05,
+      "loss": 0.7298,
+      "step": 2067
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.7848218417399354e-05,
+      "loss": 1.0355,
+      "step": 2068
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.787135585377141e-05,
+      "loss": 0.8472,
+      "step": 2069
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.789449329014345e-05,
+      "loss": 1.1314,
+      "step": 2070
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.7917630726515506e-05,
+      "loss": 0.9824,
+      "step": 2071
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.794076816288756e-05,
+      "loss": 1.1041,
+      "step": 2072
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.7963905599259605e-05,
+      "loss": 1.1476,
+      "step": 2073
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.798704303563166e-05,
+      "loss": 0.5276,
+      "step": 2074
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.8010180472003704e-05,
+      "loss": 0.9812,
+      "step": 2075
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.803331790837575e-05,
+      "loss": 0.7329,
+      "step": 2076
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.8056455344747803e-05,
+      "loss": 1.0023,
+      "step": 2077
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.8079592781119856e-05,
+      "loss": 0.7054,
+      "step": 2078
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.81027302174919e-05,
+      "loss": 0.8116,
+      "step": 2079
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.8125867653863956e-05,
+      "loss": 0.8241,
+      "step": 2080
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.8149005090236e-05,
+      "loss": 0.8245,
+      "step": 2081
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.8172142526608055e-05,
+      "loss": 0.7539,
+      "step": 2082
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.819527996298011e-05,
+      "loss": 0.9842,
+      "step": 2083
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.8218417399352154e-05,
+      "loss": 0.8855,
+      "step": 2084
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.824155483572421e-05,
+      "loss": 0.7555,
+      "step": 2085
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.826469227209625e-05,
+      "loss": 0.9184,
+      "step": 2086
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.82878297084683e-05,
+      "loss": 0.7316,
+      "step": 2087
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.831096714484035e-05,
+      "loss": 0.9218,
+      "step": 2088
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.8334104581212405e-05,
+      "loss": 0.9252,
+      "step": 2089
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.835724201758445e-05,
+      "loss": 0.6351,
+      "step": 2090
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.8380379453956504e-05,
+      "loss": 0.9417,
+      "step": 2091
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.840351689032856e-05,
+      "loss": 0.8472,
+      "step": 2092
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.8426654326700604e-05,
+      "loss": 0.9327,
+      "step": 2093
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.8449791763072657e-05,
+      "loss": 1.3593,
+      "step": 2094
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.84729291994447e-05,
+      "loss": 1.0199,
+      "step": 2095
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.8496066635816756e-05,
+      "loss": 1.0195,
+      "step": 2096
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.851920407218881e-05,
+      "loss": 1.1053,
+      "step": 2097
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.854234150856085e-05,
+      "loss": 0.9042,
+      "step": 2098
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.85654789449329e-05,
+      "loss": 1.1271,
+      "step": 2099
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.8588616381304954e-05,
+      "loss": 0.6948,
+      "step": 2100
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.8611753817677e-05,
+      "loss": 0.7413,
+      "step": 2101
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.863489125404905e-05,
+      "loss": 0.6161,
+      "step": 2102
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.8658028690421106e-05,
+      "loss": 0.714,
+      "step": 2103
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.868116612679315e-05,
+      "loss": 0.984,
+      "step": 2104
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.8704303563165205e-05,
+      "loss": 0.9557,
+      "step": 2105
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.872744099953726e-05,
+      "loss": 0.9347,
+      "step": 2106
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.8750578435909305e-05,
+      "loss": 0.7485,
+      "step": 2107
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.877371587228136e-05,
+      "loss": 0.6129,
+      "step": 2108
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.8796853308653404e-05,
+      "loss": 0.7945,
+      "step": 2109
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.881999074502545e-05,
+      "loss": 0.8093,
+      "step": 2110
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.88431281813975e-05,
+      "loss": 0.9136,
+      "step": 2111
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.886626561776955e-05,
+      "loss": 0.8508,
+      "step": 2112
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.88894030541416e-05,
+      "loss": 0.7262,
+      "step": 2113
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.8912540490513655e-05,
+      "loss": 0.565,
+      "step": 2114
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.89356779268857e-05,
+      "loss": 0.792,
+      "step": 2115
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.8958815363257754e-05,
+      "loss": 0.7047,
+      "step": 2116
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.898195279962981e-05,
+      "loss": 1.0581,
+      "step": 2117
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.900509023600185e-05,
+      "loss": 0.8053,
+      "step": 2118
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.9028227672373906e-05,
+      "loss": 0.5494,
+      "step": 2119
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.905136510874595e-05,
+      "loss": 1.1308,
+      "step": 2120
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.9074502545118e-05,
+      "loss": 0.8168,
+      "step": 2121
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.909763998149005e-05,
+      "loss": 0.8716,
+      "step": 2122
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.9120777417862105e-05,
+      "loss": 0.7273,
+      "step": 2123
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.914391485423415e-05,
+      "loss": 0.999,
+      "step": 2124
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.9167052290606204e-05,
+      "loss": 0.8621,
+      "step": 2125
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.919018972697825e-05,
+      "loss": 1.0144,
+      "step": 2126
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.92133271633503e-05,
+      "loss": 0.6924,
+      "step": 2127
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.9236464599722356e-05,
+      "loss": 0.7531,
+      "step": 2128
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.92596020360944e-05,
+      "loss": 0.7823,
+      "step": 2129
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.9282739472466455e-05,
+      "loss": 0.9154,
+      "step": 2130
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.930587690883851e-05,
+      "loss": 0.6875,
+      "step": 2131
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.9329014345210554e-05,
+      "loss": 0.7329,
+      "step": 2132
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.93521517815826e-05,
+      "loss": 1.0121,
+      "step": 2133
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.9375289217954653e-05,
+      "loss": 0.8627,
+      "step": 2134
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.93984266543267e-05,
+      "loss": 0.9329,
+      "step": 2135
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.942156409069875e-05,
+      "loss": 0.8992,
+      "step": 2136
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.9444701527070806e-05,
+      "loss": 0.6112,
+      "step": 2137
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.946783896344285e-05,
+      "loss": 0.7878,
+      "step": 2138
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.9490976399814905e-05,
+      "loss": 0.9022,
+      "step": 2139
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.951411383618695e-05,
+      "loss": 0.9546,
+      "step": 2140
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.9537251272559004e-05,
+      "loss": 0.8587,
+      "step": 2141
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.956038870893106e-05,
+      "loss": 0.8702,
+      "step": 2142
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.95835261453031e-05,
+      "loss": 0.7006,
+      "step": 2143
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.960666358167515e-05,
+      "loss": 1.2032,
+      "step": 2144
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.96298010180472e-05,
+      "loss": 0.6739,
+      "step": 2145
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.965293845441925e-05,
+      "loss": 0.8436,
+      "step": 2146
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.96760758907913e-05,
+      "loss": 0.7771,
+      "step": 2147
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.9699213327163354e-05,
+      "loss": 0.883,
+      "step": 2148
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.97223507635354e-05,
+      "loss": 1.0047,
+      "step": 2149
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.9745488199907454e-05,
+      "loss": 0.7286,
+      "step": 2150
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.9768625636279507e-05,
+      "loss": 0.6632,
+      "step": 2151
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.979176307265155e-05,
+      "loss": 0.8759,
+      "step": 2152
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.9814900509023606e-05,
+      "loss": 1.1509,
+      "step": 2153
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.983803794539565e-05,
+      "loss": 0.7185,
+      "step": 2154
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.9861175381767705e-05,
+      "loss": 0.7575,
+      "step": 2155
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.988431281813975e-05,
+      "loss": 0.7987,
+      "step": 2156
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.99074502545118e-05,
+      "loss": 0.9612,
+      "step": 2157
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.993058769088385e-05,
+      "loss": 0.6694,
+      "step": 2158
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.99537251272559e-05,
+      "loss": 0.9696,
+      "step": 2159
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.997686256362795e-05,
+      "loss": 0.5454,
+      "step": 2160
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 5e-05,
+      "loss": 0.757,
+      "step": 2161
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.999742838039397e-05,
+      "loss": 0.8759,
+      "step": 2162
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.999485676078794e-05,
+      "loss": 0.6802,
+      "step": 2163
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.999228514118192e-05,
+      "loss": 0.752,
+      "step": 2164
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.9989713521575895e-05,
+      "loss": 0.7966,
+      "step": 2165
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.9987141901969864e-05,
+      "loss": 0.7604,
+      "step": 2166
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.9984570282363834e-05,
+      "loss": 0.8614,
+      "step": 2167
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.9981998662757804e-05,
+      "loss": 0.9523,
+      "step": 2168
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.997942704315178e-05,
+      "loss": 0.8632,
+      "step": 2169
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.997685542354576e-05,
+      "loss": 0.8835,
+      "step": 2170
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.9974283803939726e-05,
+      "loss": 1.1778,
+      "step": 2171
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.9971712184333696e-05,
+      "loss": 0.6772,
+      "step": 2172
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.9969140564727666e-05,
+      "loss": 1.0233,
+      "step": 2173
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.996656894512164e-05,
+      "loss": 0.8655,
+      "step": 2174
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.996399732551561e-05,
+      "loss": 0.9467,
+      "step": 2175
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.996142570590959e-05,
+      "loss": 0.7497,
+      "step": 2176
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.995885408630356e-05,
+      "loss": 0.7676,
+      "step": 2177
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.995628246669753e-05,
+      "loss": 0.9816,
+      "step": 2178
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.99537108470915e-05,
+      "loss": 1.0318,
+      "step": 2179
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.9951139227485474e-05,
+      "loss": 0.9735,
+      "step": 2180
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.9948567607879444e-05,
+      "loss": 0.7318,
+      "step": 2181
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.994599598827342e-05,
+      "loss": 0.9409,
+      "step": 2182
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.994342436866739e-05,
+      "loss": 0.8,
+      "step": 2183
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.994085274906136e-05,
+      "loss": 1.0625,
+      "step": 2184
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.9938281129455336e-05,
+      "loss": 1.0544,
+      "step": 2185
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.9935709509849306e-05,
+      "loss": 0.8755,
+      "step": 2186
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.9933137890243275e-05,
+      "loss": 0.9761,
+      "step": 2187
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.993056627063725e-05,
+      "loss": 0.6374,
+      "step": 2188
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.992799465103122e-05,
+      "loss": 0.7631,
+      "step": 2189
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.99254230314252e-05,
+      "loss": 1.1405,
+      "step": 2190
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.992285141181917e-05,
+      "loss": 0.9553,
+      "step": 2191
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.992027979221314e-05,
+      "loss": 0.72,
+      "step": 2192
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.991770817260711e-05,
+      "loss": 0.987,
+      "step": 2193
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.991513655300108e-05,
+      "loss": 0.9227,
+      "step": 2194
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.991256493339505e-05,
+      "loss": 0.8838,
+      "step": 2195
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.990999331378903e-05,
+      "loss": 0.6081,
+      "step": 2196
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.9907421694183e-05,
+      "loss": 0.8523,
+      "step": 2197
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.990485007457697e-05,
+      "loss": 0.9468,
+      "step": 2198
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.990227845497094e-05,
+      "loss": 1.152,
+      "step": 2199
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.9899706835364915e-05,
+      "loss": 0.6659,
+      "step": 2200
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.989713521575889e-05,
+      "loss": 1.0759,
+      "step": 2201
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.989456359615286e-05,
+      "loss": 0.5192,
+      "step": 2202
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.989199197654683e-05,
+      "loss": 0.856,
+      "step": 2203
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.98894203569408e-05,
+      "loss": 1.1643,
+      "step": 2204
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.988684873733478e-05,
+      "loss": 0.8614,
+      "step": 2205
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.9884277117728753e-05,
+      "loss": 0.7629,
+      "step": 2206
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.988170549812272e-05,
+      "loss": 0.9142,
+      "step": 2207
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.987913387851669e-05,
+      "loss": 0.6886,
+      "step": 2208
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.987656225891066e-05,
+      "loss": 0.947,
+      "step": 2209
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.987399063930463e-05,
+      "loss": 1.0648,
+      "step": 2210
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.987141901969861e-05,
+      "loss": 0.9286,
+      "step": 2211
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.9868847400092585e-05,
+      "loss": 0.9489,
+      "step": 2212
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.9866275780486555e-05,
+      "loss": 1.1646,
+      "step": 2213
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.9863704160880524e-05,
+      "loss": 1.1231,
+      "step": 2214
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.9861132541274494e-05,
+      "loss": 0.7359,
+      "step": 2215
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.985856092166847e-05,
+      "loss": 1.4818,
+      "step": 2216
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.985598930206244e-05,
+      "loss": 0.9205,
+      "step": 2217
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.985341768245642e-05,
+      "loss": 1.0147,
+      "step": 2218
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.9850846062850386e-05,
+      "loss": 1.268,
+      "step": 2219
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.9848274443244356e-05,
+      "loss": 0.7952,
+      "step": 2220
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.984570282363833e-05,
+      "loss": 0.9363,
+      "step": 2221
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.98431312040323e-05,
+      "loss": 0.8472,
+      "step": 2222
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.984055958442627e-05,
+      "loss": 1.0827,
+      "step": 2223
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.983798796482025e-05,
+      "loss": 0.8624,
+      "step": 2224
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.983541634521422e-05,
+      "loss": 0.8346,
+      "step": 2225
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.983284472560819e-05,
+      "loss": 0.6931,
+      "step": 2226
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.9830273106002164e-05,
+      "loss": 0.6587,
+      "step": 2227
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.9827701486396134e-05,
+      "loss": 1.0381,
+      "step": 2228
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.9825129866790104e-05,
+      "loss": 0.7302,
+      "step": 2229
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.982255824718408e-05,
+      "loss": 0.6313,
+      "step": 2230
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.981998662757805e-05,
+      "loss": 0.6971,
+      "step": 2231
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.9817415007972026e-05,
+      "loss": 0.9527,
+      "step": 2232
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.9814843388365996e-05,
+      "loss": 0.7685,
+      "step": 2233
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.9812271768759966e-05,
+      "loss": 0.8894,
+      "step": 2234
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.9809700149153935e-05,
+      "loss": 0.8535,
+      "step": 2235
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.980712852954791e-05,
+      "loss": 0.7035,
+      "step": 2236
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.980455690994189e-05,
+      "loss": 0.7077,
+      "step": 2237
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.980198529033586e-05,
+      "loss": 0.9217,
+      "step": 2238
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.979941367072983e-05,
+      "loss": 0.9269,
+      "step": 2239
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.97968420511238e-05,
+      "loss": 1.076,
+      "step": 2240
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.979427043151777e-05,
+      "loss": 1.1201,
+      "step": 2241
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.9791698811911743e-05,
+      "loss": 0.668,
+      "step": 2242
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.978912719230572e-05,
+      "loss": 1.2906,
+      "step": 2243
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.978655557269969e-05,
+      "loss": 1.1101,
+      "step": 2244
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.978398395309366e-05,
+      "loss": 0.7603,
+      "step": 2245
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.978141233348763e-05,
+      "loss": 0.9807,
+      "step": 2246
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.9778840713881605e-05,
+      "loss": 0.7,
+      "step": 2247
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.977626909427558e-05,
+      "loss": 0.9613,
+      "step": 2248
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.977369747466955e-05,
+      "loss": 0.9945,
+      "step": 2249
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.977112585506352e-05,
+      "loss": 0.8087,
+      "step": 2250
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.976855423545749e-05,
+      "loss": 0.9256,
+      "step": 2251
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.976598261585147e-05,
+      "loss": 0.7121,
+      "step": 2252
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.976341099624544e-05,
+      "loss": 0.7916,
+      "step": 2253
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.9760839376639414e-05,
+      "loss": 0.6516,
+      "step": 2254
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.975826775703338e-05,
+      "loss": 0.7656,
+      "step": 2255
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.975569613742735e-05,
+      "loss": 0.6535,
+      "step": 2256
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.975312451782132e-05,
+      "loss": 0.8895,
+      "step": 2257
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.97505528982153e-05,
+      "loss": 0.9961,
+      "step": 2258
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.974798127860927e-05,
+      "loss": 0.5127,
+      "step": 2259
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.9745409659003245e-05,
+      "loss": 0.957,
+      "step": 2260
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.9742838039397215e-05,
+      "loss": 1.1758,
+      "step": 2261
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.9740266419791185e-05,
+      "loss": 0.7491,
+      "step": 2262
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.973769480018516e-05,
+      "loss": 0.8638,
+      "step": 2263
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.973512318057913e-05,
+      "loss": 0.9041,
+      "step": 2264
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.97325515609731e-05,
+      "loss": 1.069,
+      "step": 2265
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.972997994136708e-05,
+      "loss": 0.8446,
+      "step": 2266
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.9727408321761047e-05,
+      "loss": 1.008,
+      "step": 2267
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.972483670215502e-05,
+      "loss": 0.8496,
+      "step": 2268
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.972226508254899e-05,
+      "loss": 1.0673,
+      "step": 2269
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.971969346294296e-05,
+      "loss": 1.0133,
+      "step": 2270
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.971712184333693e-05,
+      "loss": 0.8442,
+      "step": 2271
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.971455022373091e-05,
+      "loss": 0.7687,
+      "step": 2272
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.971197860412488e-05,
+      "loss": 0.9743,
+      "step": 2273
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.9709406984518855e-05,
+      "loss": 0.745,
+      "step": 2274
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.9706835364912824e-05,
+      "loss": 0.5122,
+      "step": 2275
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.9704263745306794e-05,
+      "loss": 0.7,
+      "step": 2276
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.9701692125700764e-05,
+      "loss": 1.0092,
+      "step": 2277
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.969912050609474e-05,
+      "loss": 0.906,
+      "step": 2278
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.969654888648872e-05,
+      "loss": 0.6923,
+      "step": 2279
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.9693977266882686e-05,
+      "loss": 0.9025,
+      "step": 2280
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.9691405647276656e-05,
+      "loss": 1.0253,
+      "step": 2281
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.9688834027670626e-05,
+      "loss": 0.841,
+      "step": 2282
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.96862624080646e-05,
+      "loss": 0.8378,
+      "step": 2283
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.968369078845858e-05,
+      "loss": 0.8891,
+      "step": 2284
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.968111916885255e-05,
+      "loss": 0.879,
+      "step": 2285
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.967854754924652e-05,
+      "loss": 0.6712,
+      "step": 2286
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.967597592964049e-05,
+      "loss": 0.6921,
+      "step": 2287
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.967340431003446e-05,
+      "loss": 0.8849,
+      "step": 2288
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.9670832690428434e-05,
+      "loss": 1.1611,
+      "step": 2289
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.966826107082241e-05,
+      "loss": 0.9664,
+      "step": 2290
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.966568945121638e-05,
+      "loss": 1.0549,
+      "step": 2291
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.966311783161035e-05,
+      "loss": 0.8611,
+      "step": 2292
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.966054621200432e-05,
+      "loss": 0.7444,
+      "step": 2293
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.9657974592398296e-05,
+      "loss": 1.0331,
+      "step": 2294
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.9655402972792265e-05,
+      "loss": 0.909,
+      "step": 2295
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.965283135318624e-05,
+      "loss": 0.699,
+      "step": 2296
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.965025973358021e-05,
+      "loss": 1.0411,
+      "step": 2297
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.964768811397418e-05,
+      "loss": 0.8285,
+      "step": 2298
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.964511649436816e-05,
+      "loss": 0.8022,
+      "step": 2299
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.964254487476213e-05,
+      "loss": 0.9323,
+      "step": 2300
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.96399732551561e-05,
+      "loss": 0.9563,
+      "step": 2301
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.9637401635550074e-05,
+      "loss": 1.1629,
+      "step": 2302
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.963483001594404e-05,
+      "loss": 0.9977,
+      "step": 2303
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.963225839633801e-05,
+      "loss": 0.6088,
+      "step": 2304
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.962968677673199e-05,
+      "loss": 1.0825,
+      "step": 2305
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.962711515712596e-05,
+      "loss": 0.6424,
+      "step": 2306
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.962454353751993e-05,
+      "loss": 0.7878,
+      "step": 2307
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.9621971917913905e-05,
+      "loss": 0.9356,
+      "step": 2308
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.9619400298307875e-05,
+      "loss": 0.8979,
+      "step": 2309
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.961682867870185e-05,
+      "loss": 1.0603,
+      "step": 2310
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.961425705909582e-05,
+      "loss": 0.7572,
+      "step": 2311
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.961168543948979e-05,
+      "loss": 0.9715,
+      "step": 2312
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.960911381988377e-05,
+      "loss": 0.729,
+      "step": 2313
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.960654220027774e-05,
+      "loss": 0.7639,
+      "step": 2314
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.9603970580671713e-05,
+      "loss": 0.8148,
+      "step": 2315
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.960139896106568e-05,
+      "loss": 0.9371,
+      "step": 2316
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.959882734145965e-05,
+      "loss": 0.9634,
+      "step": 2317
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.959625572185362e-05,
+      "loss": 1.3153,
+      "step": 2318
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.95936841022476e-05,
+      "loss": 0.797,
+      "step": 2319
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.959111248264157e-05,
+      "loss": 0.8077,
+      "step": 2320
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.9588540863035545e-05,
+      "loss": 0.8899,
+      "step": 2321
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.9585969243429515e-05,
+      "loss": 1.0435,
+      "step": 2322
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.9583397623823484e-05,
+      "loss": 0.7551,
+      "step": 2323
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.9580826004217454e-05,
+      "loss": 0.92,
+      "step": 2324
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.957825438461143e-05,
+      "loss": 0.7937,
+      "step": 2325
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.957568276500541e-05,
+      "loss": 0.7067,
+      "step": 2326
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.957311114539938e-05,
+      "loss": 0.9619,
+      "step": 2327
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.9570539525793346e-05,
+      "loss": 1.0095,
+      "step": 2328
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.9567967906187316e-05,
+      "loss": 0.9531,
+      "step": 2329
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.956539628658129e-05,
+      "loss": 0.7537,
+      "step": 2330
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.956282466697527e-05,
+      "loss": 0.6166,
+      "step": 2331
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.956025304736924e-05,
+      "loss": 1.0586,
+      "step": 2332
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.955768142776321e-05,
+      "loss": 1.029,
+      "step": 2333
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.955510980815718e-05,
+      "loss": 0.7485,
+      "step": 2334
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.955253818855115e-05,
+      "loss": 0.8167,
+      "step": 2335
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.9549966568945124e-05,
+      "loss": 0.982,
+      "step": 2336
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.95473949493391e-05,
+      "loss": 0.869,
+      "step": 2337
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.954482332973307e-05,
+      "loss": 0.9721,
+      "step": 2338
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.954225171012704e-05,
+      "loss": 0.7776,
+      "step": 2339
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.953968009052101e-05,
+      "loss": 1.0674,
+      "step": 2340
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.9537108470914986e-05,
+      "loss": 0.8478,
+      "step": 2341
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.9534536851308956e-05,
+      "loss": 1.1277,
+      "step": 2342
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.953196523170293e-05,
+      "loss": 0.8404,
+      "step": 2343
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.95293936120969e-05,
+      "loss": 0.8337,
+      "step": 2344
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.952682199249087e-05,
+      "loss": 0.7098,
+      "step": 2345
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.952425037288485e-05,
+      "loss": 0.8368,
+      "step": 2346
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.952167875327882e-05,
+      "loss": 0.964,
+      "step": 2347
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.951910713367279e-05,
+      "loss": 0.9321,
+      "step": 2348
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.9516535514066764e-05,
+      "loss": 0.7072,
+      "step": 2349
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.9513963894460734e-05,
+      "loss": 1.1391,
+      "step": 2350
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.95113922748547e-05,
+      "loss": 0.5666,
+      "step": 2351
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.950882065524868e-05,
+      "loss": 0.7161,
+      "step": 2352
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.950624903564265e-05,
+      "loss": 0.9319,
+      "step": 2353
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.950367741603662e-05,
+      "loss": 0.7594,
+      "step": 2354
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.9501105796430596e-05,
+      "loss": 0.6636,
+      "step": 2355
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.9498534176824565e-05,
+      "loss": 0.8881,
+      "step": 2356
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.949596255721854e-05,
+      "loss": 1.1863,
+      "step": 2357
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.949339093761251e-05,
+      "loss": 0.6647,
+      "step": 2358
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.949081931800648e-05,
+      "loss": 1.0269,
+      "step": 2359
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.948824769840045e-05,
+      "loss": 0.8818,
+      "step": 2360
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.948567607879443e-05,
+      "loss": 1.1924,
+      "step": 2361
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.9483104459188404e-05,
+      "loss": 1.0352,
+      "step": 2362
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.9480532839582373e-05,
+      "loss": 0.8544,
+      "step": 2363
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.947796121997634e-05,
+      "loss": 0.8441,
+      "step": 2364
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.947538960037031e-05,
+      "loss": 0.6779,
+      "step": 2365
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.947281798076428e-05,
+      "loss": 0.8589,
+      "step": 2366
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.947024636115826e-05,
+      "loss": 0.7183,
+      "step": 2367
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.9467674741552235e-05,
+      "loss": 0.9643,
+      "step": 2368
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.9465103121946205e-05,
+      "loss": 0.8828,
+      "step": 2369
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.9462531502340175e-05,
+      "loss": 0.756,
+      "step": 2370
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.9459959882734145e-05,
+      "loss": 1.0683,
+      "step": 2371
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.945738826312812e-05,
+      "loss": 0.8806,
+      "step": 2372
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.94548166435221e-05,
+      "loss": 0.8673,
+      "step": 2373
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.945224502391607e-05,
+      "loss": 0.6272,
+      "step": 2374
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.944967340431004e-05,
+      "loss": 1.0183,
+      "step": 2375
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.9447101784704007e-05,
+      "loss": 0.6223,
+      "step": 2376
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.944453016509798e-05,
+      "loss": 0.7326,
+      "step": 2377
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.944195854549195e-05,
+      "loss": 0.9897,
+      "step": 2378
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.943938692588593e-05,
+      "loss": 0.8582,
+      "step": 2379
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.94368153062799e-05,
+      "loss": 0.9579,
+      "step": 2380
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.943424368667387e-05,
+      "loss": 1.069,
+      "step": 2381
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.943167206706784e-05,
+      "loss": 0.8017,
+      "step": 2382
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.9429100447461815e-05,
+      "loss": 0.7469,
+      "step": 2383
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.9426528827855784e-05,
+      "loss": 1.0324,
+      "step": 2384
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.942395720824976e-05,
+      "loss": 1.1989,
+      "step": 2385
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.942138558864373e-05,
+      "loss": 0.8616,
+      "step": 2386
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.94188139690377e-05,
+      "loss": 0.9072,
+      "step": 2387
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.9416242349431677e-05,
+      "loss": 0.7542,
+      "step": 2388
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.9413670729825646e-05,
+      "loss": 0.8024,
+      "step": 2389
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.9411099110219616e-05,
+      "loss": 0.6183,
+      "step": 2390
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.940852749061359e-05,
+      "loss": 1.0273,
+      "step": 2391
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.940595587100756e-05,
+      "loss": 0.6243,
+      "step": 2392
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.940338425140154e-05,
+      "loss": 0.9211,
+      "step": 2393
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.940081263179551e-05,
+      "loss": 0.9619,
+      "step": 2394
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.939824101218948e-05,
+      "loss": 0.848,
+      "step": 2395
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.939566939258345e-05,
+      "loss": 0.6643,
+      "step": 2396
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.9393097772977424e-05,
+      "loss": 0.9292,
+      "step": 2397
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.9390526153371394e-05,
+      "loss": 0.6434,
+      "step": 2398
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.938795453376537e-05,
+      "loss": 0.7601,
+      "step": 2399
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.938538291415934e-05,
+      "loss": 0.9291,
+      "step": 2400
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.938281129455331e-05,
+      "loss": 0.687,
+      "step": 2401
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.938023967494728e-05,
+      "loss": 1.1788,
+      "step": 2402
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.9377668055341256e-05,
+      "loss": 1.1318,
+      "step": 2403
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.937509643573523e-05,
+      "loss": 0.6429,
+      "step": 2404
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.93725248161292e-05,
+      "loss": 1.3655,
+      "step": 2405
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.936995319652317e-05,
+      "loss": 0.7476,
+      "step": 2406
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.936738157691714e-05,
+      "loss": 0.952,
+      "step": 2407
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.936480995731112e-05,
+      "loss": 0.8187,
+      "step": 2408
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.9362238337705094e-05,
+      "loss": 0.7863,
+      "step": 2409
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.9359666718099064e-05,
+      "loss": 0.9655,
+      "step": 2410
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.9357095098493034e-05,
+      "loss": 0.9153,
+      "step": 2411
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.9354523478887e-05,
+      "loss": 0.5778,
+      "step": 2412
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.935195185928097e-05,
+      "loss": 0.829,
+      "step": 2413
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.934938023967495e-05,
+      "loss": 1.0102,
+      "step": 2414
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.9346808620068926e-05,
+      "loss": 0.9114,
+      "step": 2415
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.9344237000462896e-05,
+      "loss": 1.2572,
+      "step": 2416
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.9341665380856865e-05,
+      "loss": 0.774,
+      "step": 2417
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.9339093761250835e-05,
+      "loss": 0.8457,
+      "step": 2418
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.933652214164481e-05,
+      "loss": 1.0277,
+      "step": 2419
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.933395052203878e-05,
+      "loss": 0.988,
+      "step": 2420
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.933137890243276e-05,
+      "loss": 1.0549,
+      "step": 2421
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.932880728282673e-05,
+      "loss": 1.0056,
+      "step": 2422
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.93262356632207e-05,
+      "loss": 0.6482,
+      "step": 2423
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.932366404361467e-05,
+      "loss": 0.7749,
+      "step": 2424
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.932109242400864e-05,
+      "loss": 0.7502,
+      "step": 2425
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.931852080440261e-05,
+      "loss": 1.0865,
+      "step": 2426
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.931594918479659e-05,
+      "loss": 0.6734,
+      "step": 2427
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.931337756519056e-05,
+      "loss": 0.8003,
+      "step": 2428
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.931080594558453e-05,
+      "loss": 0.923,
+      "step": 2429
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.9308234325978505e-05,
+      "loss": 0.6261,
+      "step": 2430
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.9305662706372475e-05,
+      "loss": 0.6208,
+      "step": 2431
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.9303091086766444e-05,
+      "loss": 0.6965,
+      "step": 2432
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.930051946716042e-05,
+      "loss": 1.1481,
+      "step": 2433
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.929794784755439e-05,
+      "loss": 0.844,
+      "step": 2434
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.929537622794837e-05,
+      "loss": 0.8116,
+      "step": 2435
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.929280460834234e-05,
+      "loss": 0.5471,
+      "step": 2436
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.9290232988736306e-05,
+      "loss": 0.8926,
+      "step": 2437
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.9287661369130276e-05,
+      "loss": 0.7426,
+      "step": 2438
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.928508974952425e-05,
+      "loss": 0.9369,
+      "step": 2439
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.928251812991823e-05,
+      "loss": 0.9075,
+      "step": 2440
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.92799465103122e-05,
+      "loss": 0.7099,
+      "step": 2441
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.927737489070617e-05,
+      "loss": 0.9318,
+      "step": 2442
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.927480327110014e-05,
+      "loss": 0.8015,
+      "step": 2443
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.927223165149411e-05,
+      "loss": 0.7956,
+      "step": 2444
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.9269660031888084e-05,
+      "loss": 1.1894,
+      "step": 2445
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.926708841228206e-05,
+      "loss": 0.7577,
+      "step": 2446
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.926451679267603e-05,
+      "loss": 0.8166,
+      "step": 2447
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.926194517307e-05,
+      "loss": 0.8863,
+      "step": 2448
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.925937355346397e-05,
+      "loss": 1.0524,
+      "step": 2449
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.9256801933857946e-05,
+      "loss": 0.8343,
+      "step": 2450
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.925423031425192e-05,
+      "loss": 1.0771,
+      "step": 2451
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.925165869464589e-05,
+      "loss": 0.8202,
+      "step": 2452
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.924908707503986e-05,
+      "loss": 0.73,
+      "step": 2453
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.924651545543383e-05,
+      "loss": 0.7743,
+      "step": 2454
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.924394383582781e-05,
+      "loss": 0.8549,
+      "step": 2455
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.924137221622178e-05,
+      "loss": 0.9664,
+      "step": 2456
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.9238800596615754e-05,
+      "loss": 0.898,
+      "step": 2457
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.9236228977009724e-05,
+      "loss": 0.7999,
+      "step": 2458
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.9233657357403694e-05,
+      "loss": 1.024,
+      "step": 2459
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.923108573779766e-05,
+      "loss": 1.1137,
+      "step": 2460
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.922851411819164e-05,
+      "loss": 0.9178,
+      "step": 2461
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.922594249858561e-05,
+      "loss": 0.7379,
+      "step": 2462
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.9223370878979586e-05,
+      "loss": 0.7953,
+      "step": 2463
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.9220799259373556e-05,
+      "loss": 0.9689,
+      "step": 2464
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.9218227639767525e-05,
+      "loss": 1.0671,
+      "step": 2465
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.92156560201615e-05,
+      "loss": 0.9004,
+      "step": 2466
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.921308440055547e-05,
+      "loss": 0.5574,
+      "step": 2467
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.921051278094945e-05,
+      "loss": 0.6631,
+      "step": 2468
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.920794116134342e-05,
+      "loss": 0.631,
+      "step": 2469
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.920536954173739e-05,
+      "loss": 0.6706,
+      "step": 2470
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.9202797922131364e-05,
+      "loss": 0.9431,
+      "step": 2471
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.9200226302525333e-05,
+      "loss": 1.0167,
+      "step": 2472
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.91976546829193e-05,
+      "loss": 0.9088,
+      "step": 2473
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.919508306331328e-05,
+      "loss": 0.7122,
+      "step": 2474
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.919251144370725e-05,
+      "loss": 0.4981,
+      "step": 2475
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.918993982410122e-05,
+      "loss": 0.8306,
+      "step": 2476
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.9187368204495195e-05,
+      "loss": 1.2625,
+      "step": 2477
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.9184796584889165e-05,
+      "loss": 0.8153,
+      "step": 2478
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.9182224965283135e-05,
+      "loss": 0.9135,
+      "step": 2479
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.917965334567711e-05,
+      "loss": 0.6698,
+      "step": 2480
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.917708172607108e-05,
+      "loss": 0.9267,
+      "step": 2481
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.917451010646506e-05,
+      "loss": 0.7863,
+      "step": 2482
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.917193848685903e-05,
+      "loss": 0.908,
+      "step": 2483
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.9169366867253e-05,
+      "loss": 0.8165,
+      "step": 2484
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.9166795247646966e-05,
+      "loss": 0.6376,
+      "step": 2485
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.916422362804094e-05,
+      "loss": 1.1863,
+      "step": 2486
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.916165200843492e-05,
+      "loss": 0.8689,
+      "step": 2487
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.915908038882889e-05,
+      "loss": 0.8435,
+      "step": 2488
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.915650876922286e-05,
+      "loss": 0.9152,
+      "step": 2489
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.915393714961683e-05,
+      "loss": 1.0663,
+      "step": 2490
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.91513655300108e-05,
+      "loss": 0.9374,
+      "step": 2491
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.9148793910404775e-05,
+      "loss": 0.9095,
+      "step": 2492
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.914622229079875e-05,
+      "loss": 0.9082,
+      "step": 2493
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.914365067119272e-05,
+      "loss": 1.1273,
+      "step": 2494
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.914107905158669e-05,
+      "loss": 1.043,
+      "step": 2495
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.913850743198066e-05,
+      "loss": 1.0061,
+      "step": 2496
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.9135935812374637e-05,
+      "loss": 0.8508,
+      "step": 2497
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.913336419276861e-05,
+      "loss": 0.9984,
+      "step": 2498
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.913079257316258e-05,
+      "loss": 1.1956,
+      "step": 2499
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.912822095355655e-05,
+      "loss": 0.9404,
+      "step": 2500
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.912564933395052e-05,
+      "loss": 0.8184,
+      "step": 2501
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.91230777143445e-05,
+      "loss": 0.7957,
+      "step": 2502
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.912050609473847e-05,
+      "loss": 0.7035,
+      "step": 2503
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.9117934475132445e-05,
+      "loss": 0.9079,
+      "step": 2504
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.9115362855526414e-05,
+      "loss": 0.8265,
+      "step": 2505
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.9112791235920384e-05,
+      "loss": 1.0543,
+      "step": 2506
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.9110219616314354e-05,
+      "loss": 0.7301,
+      "step": 2507
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.910764799670833e-05,
+      "loss": 0.8064,
+      "step": 2508
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.91050763771023e-05,
+      "loss": 0.9431,
+      "step": 2509
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.9102504757496276e-05,
+      "loss": 0.8617,
+      "step": 2510
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.9099933137890246e-05,
+      "loss": 0.7812,
+      "step": 2511
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.9097361518284216e-05,
+      "loss": 1.0409,
+      "step": 2512
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.909478989867819e-05,
+      "loss": 0.7931,
+      "step": 2513
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.909221827907216e-05,
+      "loss": 0.8554,
+      "step": 2514
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.908964665946613e-05,
+      "loss": 1.2688,
+      "step": 2515
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.908707503986011e-05,
+      "loss": 0.7905,
+      "step": 2516
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.908450342025408e-05,
+      "loss": 1.1988,
+      "step": 2517
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.9081931800648054e-05,
+      "loss": 1.2002,
+      "step": 2518
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.9079360181042024e-05,
+      "loss": 0.629,
+      "step": 2519
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.9076788561435994e-05,
+      "loss": 0.692,
+      "step": 2520
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.907421694182996e-05,
+      "loss": 0.8785,
+      "step": 2521
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.907164532222394e-05,
+      "loss": 1.3078,
+      "step": 2522
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.906907370261791e-05,
+      "loss": 0.6595,
+      "step": 2523
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.9066502083011886e-05,
+      "loss": 0.5816,
+      "step": 2524
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.9063930463405856e-05,
+      "loss": 1.2973,
+      "step": 2525
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.9061358843799825e-05,
+      "loss": 0.9169,
+      "step": 2526
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.9058787224193795e-05,
+      "loss": 0.7188,
+      "step": 2527
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.905621560458777e-05,
+      "loss": 0.9828,
+      "step": 2528
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.905364398498175e-05,
+      "loss": 1.1268,
+      "step": 2529
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.905107236537572e-05,
+      "loss": 0.6319,
+      "step": 2530
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.904850074576969e-05,
+      "loss": 0.8766,
+      "step": 2531
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.904592912616366e-05,
+      "loss": 0.6324,
+      "step": 2532
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.904335750655763e-05,
+      "loss": 0.6864,
+      "step": 2533
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.904078588695161e-05,
+      "loss": 1.3065,
+      "step": 2534
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.903821426734558e-05,
+      "loss": 0.7872,
+      "step": 2535
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.903564264773955e-05,
+      "loss": 0.767,
+      "step": 2536
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.903307102813352e-05,
+      "loss": 1.0452,
+      "step": 2537
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.903049940852749e-05,
+      "loss": 1.0385,
+      "step": 2538
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.9027927788921465e-05,
+      "loss": 0.7678,
+      "step": 2539
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.902535616931544e-05,
+      "loss": 0.8295,
+      "step": 2540
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.902278454970941e-05,
+      "loss": 0.4264,
+      "step": 2541
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.902021293010338e-05,
+      "loss": 0.7944,
+      "step": 2542
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.901764131049735e-05,
+      "loss": 0.8943,
+      "step": 2543
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.901506969089133e-05,
+      "loss": 0.5577,
+      "step": 2544
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.90124980712853e-05,
+      "loss": 1.0034,
+      "step": 2545
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.900992645167927e-05,
+      "loss": 0.8649,
+      "step": 2546
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.900735483207324e-05,
+      "loss": 0.7897,
+      "step": 2547
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.900478321246721e-05,
+      "loss": 1.3837,
+      "step": 2548
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.900221159286119e-05,
+      "loss": 1.1523,
+      "step": 2549
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.899963997325516e-05,
+      "loss": 0.8179,
+      "step": 2550
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.899706835364913e-05,
+      "loss": 0.7739,
+      "step": 2551
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.8994496734043105e-05,
+      "loss": 0.8655,
+      "step": 2552
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.8991925114437074e-05,
+      "loss": 0.6645,
+      "step": 2553
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.8989353494831044e-05,
+      "loss": 0.8682,
+      "step": 2554
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.898678187522502e-05,
+      "loss": 1.3156,
+      "step": 2555
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.898421025561899e-05,
+      "loss": 0.4426,
+      "step": 2556
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.898163863601296e-05,
+      "loss": 0.6578,
+      "step": 2557
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.8979067016406936e-05,
+      "loss": 0.662,
+      "step": 2558
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.8976495396800906e-05,
+      "loss": 0.727,
+      "step": 2559
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.897392377719488e-05,
+      "loss": 1.1496,
+      "step": 2560
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.897135215758885e-05,
+      "loss": 0.9427,
+      "step": 2561
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.896878053798282e-05,
+      "loss": 0.8342,
+      "step": 2562
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.896620891837679e-05,
+      "loss": 0.7438,
+      "step": 2563
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.896363729877077e-05,
+      "loss": 0.8546,
+      "step": 2564
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.8961065679164745e-05,
+      "loss": 0.6606,
+      "step": 2565
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.8958494059558714e-05,
+      "loss": 0.7859,
+      "step": 2566
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.8955922439952684e-05,
+      "loss": 1.2951,
+      "step": 2567
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.8953350820346654e-05,
+      "loss": 0.5518,
+      "step": 2568
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.895077920074062e-05,
+      "loss": 0.6216,
+      "step": 2569
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.89482075811346e-05,
+      "loss": 0.9249,
+      "step": 2570
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.8945635961528576e-05,
+      "loss": 0.9146,
+      "step": 2571
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.8943064341922546e-05,
+      "loss": 0.8362,
+      "step": 2572
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.8940492722316516e-05,
+      "loss": 1.3738,
+      "step": 2573
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.8937921102710485e-05,
+      "loss": 0.7536,
+      "step": 2574
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.893534948310446e-05,
+      "loss": 0.8483,
+      "step": 2575
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.893277786349844e-05,
+      "loss": 1.2182,
+      "step": 2576
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.893020624389241e-05,
+      "loss": 0.9343,
+      "step": 2577
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.892763462428638e-05,
+      "loss": 0.8876,
+      "step": 2578
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.892506300468035e-05,
+      "loss": 0.842,
+      "step": 2579
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.8922491385074324e-05,
+      "loss": 0.8163,
+      "step": 2580
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.8919919765468293e-05,
+      "loss": 0.7876,
+      "step": 2581
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.891734814586227e-05,
+      "loss": 0.9033,
+      "step": 2582
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.891477652625624e-05,
+      "loss": 0.7062,
+      "step": 2583
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.891220490665021e-05,
+      "loss": 1.0261,
+      "step": 2584
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.890963328704418e-05,
+      "loss": 0.7745,
+      "step": 2585
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.8907061667438155e-05,
+      "loss": 0.7174,
+      "step": 2586
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.8904490047832125e-05,
+      "loss": 0.9006,
+      "step": 2587
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.89019184282261e-05,
+      "loss": 0.9641,
+      "step": 2588
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.889934680862007e-05,
+      "loss": 1.0112,
+      "step": 2589
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.889677518901404e-05,
+      "loss": 0.9938,
+      "step": 2590
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.889420356940802e-05,
+      "loss": 0.9781,
+      "step": 2591
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.889163194980199e-05,
+      "loss": 0.9243,
+      "step": 2592
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.888906033019596e-05,
+      "loss": 0.8345,
+      "step": 2593
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.888648871058993e-05,
+      "loss": 0.9092,
+      "step": 2594
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.88839170909839e-05,
+      "loss": 0.8266,
+      "step": 2595
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.888134547137788e-05,
+      "loss": 0.816,
+      "step": 2596
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.887877385177185e-05,
+      "loss": 0.9073,
+      "step": 2597
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.887620223216582e-05,
+      "loss": 0.779,
+      "step": 2598
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.887363061255979e-05,
+      "loss": 1.1314,
+      "step": 2599
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.8871058992953765e-05,
+      "loss": 1.1036,
+      "step": 2600
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.8868487373347735e-05,
+      "loss": 1.128,
+      "step": 2601
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.886591575374171e-05,
+      "loss": 0.8775,
+      "step": 2602
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.886334413413568e-05,
+      "loss": 0.8311,
+      "step": 2603
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.886077251452965e-05,
+      "loss": 0.89,
+      "step": 2604
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.885820089492362e-05,
+      "loss": 0.8902,
+      "step": 2605
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.8855629275317597e-05,
+      "loss": 0.8132,
+      "step": 2606
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.885305765571157e-05,
+      "loss": 0.7466,
+      "step": 2607
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.885048603610554e-05,
+      "loss": 0.8366,
+      "step": 2608
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.884791441649951e-05,
+      "loss": 1.0427,
+      "step": 2609
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.884534279689348e-05,
+      "loss": 0.6978,
+      "step": 2610
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.884277117728746e-05,
+      "loss": 0.9462,
+      "step": 2611
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.8840199557681435e-05,
+      "loss": 0.9242,
+      "step": 2612
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.8837627938075405e-05,
+      "loss": 0.626,
+      "step": 2613
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.8835056318469374e-05,
+      "loss": 1.1129,
+      "step": 2614
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.8832484698863344e-05,
+      "loss": 1.0202,
+      "step": 2615
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.8829913079257314e-05,
+      "loss": 0.926,
+      "step": 2616
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.882734145965129e-05,
+      "loss": 0.7996,
+      "step": 2617
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.882476984004527e-05,
+      "loss": 0.9122,
+      "step": 2618
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.8822198220439236e-05,
+      "loss": 0.6106,
+      "step": 2619
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.8819626600833206e-05,
+      "loss": 0.8206,
+      "step": 2620
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.8817054981227176e-05,
+      "loss": 0.9509,
+      "step": 2621
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.881448336162115e-05,
+      "loss": 1.1009,
+      "step": 2622
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.881191174201513e-05,
+      "loss": 1.1395,
+      "step": 2623
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.88093401224091e-05,
+      "loss": 1.1237,
+      "step": 2624
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.880676850280307e-05,
+      "loss": 0.6676,
+      "step": 2625
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.880419688319704e-05,
+      "loss": 0.5851,
+      "step": 2626
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.8801625263591014e-05,
+      "loss": 0.7445,
+      "step": 2627
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.8799053643984984e-05,
+      "loss": 0.8984,
+      "step": 2628
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.879648202437896e-05,
+      "loss": 0.9797,
+      "step": 2629
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.879391040477293e-05,
+      "loss": 0.8985,
+      "step": 2630
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.87913387851669e-05,
+      "loss": 0.7691,
+      "step": 2631
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.878876716556087e-05,
+      "loss": 0.9341,
+      "step": 2632
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.8786195545954846e-05,
+      "loss": 1.1951,
+      "step": 2633
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.8783623926348815e-05,
+      "loss": 0.5567,
+      "step": 2634
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.878105230674279e-05,
+      "loss": 0.6352,
+      "step": 2635
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.877848068713676e-05,
+      "loss": 0.6446,
+      "step": 2636
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.877590906753073e-05,
+      "loss": 0.951,
+      "step": 2637
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.877333744792471e-05,
+      "loss": 0.9971,
+      "step": 2638
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.877076582831868e-05,
+      "loss": 0.7104,
+      "step": 2639
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.876819420871265e-05,
+      "loss": 0.7623,
+      "step": 2640
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.8765622589106624e-05,
+      "loss": 0.8996,
+      "step": 2641
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.876305096950059e-05,
+      "loss": 0.8676,
+      "step": 2642
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.876047934989457e-05,
+      "loss": 0.7027,
+      "step": 2643
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.875790773028854e-05,
+      "loss": 0.6978,
+      "step": 2644
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.875533611068251e-05,
+      "loss": 0.8779,
+      "step": 2645
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.875276449107648e-05,
+      "loss": 0.6353,
+      "step": 2646
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.8750192871470455e-05,
+      "loss": 0.9992,
+      "step": 2647
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.8747621251864425e-05,
+      "loss": 0.6381,
+      "step": 2648
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.87450496322584e-05,
+      "loss": 0.8251,
+      "step": 2649
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.874247801265237e-05,
+      "loss": 1.0041,
+      "step": 2650
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.873990639304634e-05,
+      "loss": 0.8703,
+      "step": 2651
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.873733477344031e-05,
+      "loss": 0.7328,
+      "step": 2652
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.873476315383429e-05,
+      "loss": 1.0198,
+      "step": 2653
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.8732191534228263e-05,
+      "loss": 0.9388,
+      "step": 2654
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.872961991462223e-05,
+      "loss": 0.6428,
+      "step": 2655
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.87270482950162e-05,
+      "loss": 0.7733,
+      "step": 2656
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.872447667541017e-05,
+      "loss": 0.9359,
+      "step": 2657
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.872190505580415e-05,
+      "loss": 0.9002,
+      "step": 2658
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.8719333436198125e-05,
+      "loss": 0.745,
+      "step": 2659
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.8716761816592095e-05,
+      "loss": 0.7725,
+      "step": 2660
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.8714190196986065e-05,
+      "loss": 0.8438,
+      "step": 2661
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.8711618577380034e-05,
+      "loss": 0.8826,
+      "step": 2662
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.8709046957774004e-05,
+      "loss": 1.1154,
+      "step": 2663
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.870647533816798e-05,
+      "loss": 0.8681,
+      "step": 2664
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.870390371856196e-05,
+      "loss": 0.9018,
+      "step": 2665
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.870133209895593e-05,
+      "loss": 0.8755,
+      "step": 2666
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.8698760479349896e-05,
+      "loss": 0.7588,
+      "step": 2667
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.8696188859743866e-05,
+      "loss": 0.8923,
+      "step": 2668
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.869361724013784e-05,
+      "loss": 0.9798,
+      "step": 2669
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.869104562053181e-05,
+      "loss": 1.2395,
+      "step": 2670
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.868847400092579e-05,
+      "loss": 1.2737,
+      "step": 2671
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.868590238131976e-05,
+      "loss": 0.5702,
+      "step": 2672
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.868333076171373e-05,
+      "loss": 1.0486,
+      "step": 2673
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.8680759142107705e-05,
+      "loss": 0.6877,
+      "step": 2674
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.8678187522501674e-05,
+      "loss": 0.7795,
+      "step": 2675
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.8675615902895644e-05,
+      "loss": 0.8782,
+      "step": 2676
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.867304428328962e-05,
+      "loss": 1.2542,
+      "step": 2677
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.867047266368359e-05,
+      "loss": 0.6472,
+      "step": 2678
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.866790104407756e-05,
+      "loss": 1.2358,
+      "step": 2679
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.8665329424471536e-05,
+      "loss": 0.7286,
+      "step": 2680
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.8662757804865506e-05,
+      "loss": 1.0599,
+      "step": 2681
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.8660186185259476e-05,
+      "loss": 0.6665,
+      "step": 2682
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.865761456565345e-05,
+      "loss": 0.9547,
+      "step": 2683
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.865504294604742e-05,
+      "loss": 0.867,
+      "step": 2684
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.86524713264414e-05,
+      "loss": 1.0525,
+      "step": 2685
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.864989970683537e-05,
+      "loss": 0.7568,
+      "step": 2686
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.864732808722934e-05,
+      "loss": 0.7299,
+      "step": 2687
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.864475646762331e-05,
+      "loss": 0.9237,
+      "step": 2688
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.8642184848017284e-05,
+      "loss": 0.8088,
+      "step": 2689
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.863961322841126e-05,
+      "loss": 0.9193,
+      "step": 2690
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.863704160880523e-05,
+      "loss": 0.8381,
+      "step": 2691
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.86344699891992e-05,
+      "loss": 0.878,
+      "step": 2692
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.863189836959317e-05,
+      "loss": 0.8807,
+      "step": 2693
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.862932674998714e-05,
+      "loss": 0.8209,
+      "step": 2694
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.8626755130381115e-05,
+      "loss": 0.6393,
+      "step": 2695
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.862418351077509e-05,
+      "loss": 0.8871,
+      "step": 2696
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.862161189116906e-05,
+      "loss": 0.7428,
+      "step": 2697
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.861904027156303e-05,
+      "loss": 0.7296,
+      "step": 2698
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.8616468651957e-05,
+      "loss": 1.0986,
+      "step": 2699
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.861389703235098e-05,
+      "loss": 0.9181,
+      "step": 2700
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.8611325412744954e-05,
+      "loss": 1.0157,
+      "step": 2701
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.8608753793138924e-05,
+      "loss": 0.8477,
+      "step": 2702
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.860618217353289e-05,
+      "loss": 0.7429,
+      "step": 2703
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.860361055392686e-05,
+      "loss": 1.0385,
+      "step": 2704
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.860103893432084e-05,
+      "loss": 0.8135,
+      "step": 2705
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.859846731471481e-05,
+      "loss": 0.7179,
+      "step": 2706
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.8595895695108785e-05,
+      "loss": 0.7212,
+      "step": 2707
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.8593324075502755e-05,
+      "loss": 0.7902,
+      "step": 2708
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.8590752455896725e-05,
+      "loss": 0.7564,
+      "step": 2709
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.8588180836290695e-05,
+      "loss": 1.1838,
+      "step": 2710
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.858560921668467e-05,
+      "loss": 0.9768,
+      "step": 2711
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.858303759707864e-05,
+      "loss": 1.2467,
+      "step": 2712
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.858046597747262e-05,
+      "loss": 0.5441,
+      "step": 2713
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.857789435786659e-05,
+      "loss": 0.8105,
+      "step": 2714
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.8575322738260557e-05,
+      "loss": 1.0346,
+      "step": 2715
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.857275111865453e-05,
+      "loss": 0.7875,
+      "step": 2716
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.85701794990485e-05,
+      "loss": 0.877,
+      "step": 2717
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.856760787944247e-05,
+      "loss": 0.6963,
+      "step": 2718
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.856503625983645e-05,
+      "loss": 0.9395,
+      "step": 2719
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.856246464023042e-05,
+      "loss": 0.8271,
+      "step": 2720
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.8559893020624395e-05,
+      "loss": 0.7942,
+      "step": 2721
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.8557321401018365e-05,
+      "loss": 1.0877,
+      "step": 2722
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.8554749781412334e-05,
+      "loss": 0.8642,
+      "step": 2723
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.8552178161806304e-05,
+      "loss": 0.6109,
+      "step": 2724
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.854960654220028e-05,
+      "loss": 0.6706,
+      "step": 2725
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.854703492259425e-05,
+      "loss": 0.9219,
+      "step": 2726
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.854446330298823e-05,
+      "loss": 0.9116,
+      "step": 2727
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.8541891683382196e-05,
+      "loss": 0.9928,
+      "step": 2728
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.8539320063776166e-05,
+      "loss": 0.788,
+      "step": 2729
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.8536748444170136e-05,
+      "loss": 0.8907,
+      "step": 2730
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.853417682456411e-05,
+      "loss": 1.0112,
+      "step": 2731
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.853160520495809e-05,
+      "loss": 0.7608,
+      "step": 2732
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.852903358535206e-05,
+      "loss": 0.9798,
+      "step": 2733
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.852646196574603e-05,
+      "loss": 0.7145,
+      "step": 2734
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.852389034614e-05,
+      "loss": 1.0036,
+      "step": 2735
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.8521318726533974e-05,
+      "loss": 0.7981,
+      "step": 2736
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.851874710692795e-05,
+      "loss": 1.0427,
+      "step": 2737
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.851617548732192e-05,
+      "loss": 0.9893,
+      "step": 2738
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.851360386771589e-05,
+      "loss": 1.0535,
+      "step": 2739
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.851103224810986e-05,
+      "loss": 1.0156,
+      "step": 2740
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.850846062850383e-05,
+      "loss": 0.9757,
+      "step": 2741
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.8505889008897806e-05,
+      "loss": 0.7767,
+      "step": 2742
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.850331738929178e-05,
+      "loss": 0.5861,
+      "step": 2743
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.850074576968575e-05,
+      "loss": 0.9367,
+      "step": 2744
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.849817415007972e-05,
+      "loss": 0.9096,
+      "step": 2745
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.849560253047369e-05,
+      "loss": 0.7895,
+      "step": 2746
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.849303091086767e-05,
+      "loss": 0.7149,
+      "step": 2747
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.849045929126164e-05,
+      "loss": 1.1059,
+      "step": 2748
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.8487887671655614e-05,
+      "loss": 0.8566,
+      "step": 2749
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.8485316052049584e-05,
+      "loss": 0.9207,
+      "step": 2750
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.848274443244355e-05,
+      "loss": 0.7722,
+      "step": 2751
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.848017281283753e-05,
+      "loss": 0.7985,
+      "step": 2752
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.84776011932315e-05,
+      "loss": 0.8311,
+      "step": 2753
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.847502957362547e-05,
+      "loss": 1.0212,
+      "step": 2754
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.8472457954019446e-05,
+      "loss": 0.7347,
+      "step": 2755
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.8469886334413415e-05,
+      "loss": 0.5658,
+      "step": 2756
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.8467314714807385e-05,
+      "loss": 0.7543,
+      "step": 2757
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.846474309520136e-05,
+      "loss": 0.9107,
+      "step": 2758
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.846217147559533e-05,
+      "loss": 0.7127,
+      "step": 2759
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.84595998559893e-05,
+      "loss": 1.2458,
+      "step": 2760
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.845702823638328e-05,
+      "loss": 0.6368,
+      "step": 2761
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.845445661677725e-05,
+      "loss": 0.9414,
+      "step": 2762
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.845188499717122e-05,
+      "loss": 0.9203,
+      "step": 2763
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.844931337756519e-05,
+      "loss": 0.771,
+      "step": 2764
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.844674175795916e-05,
+      "loss": 1.2855,
+      "step": 2765
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.844417013835314e-05,
+      "loss": 0.789,
+      "step": 2766
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.844159851874711e-05,
+      "loss": 0.7607,
+      "step": 2767
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.8439026899141085e-05,
+      "loss": 0.6539,
+      "step": 2768
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.8436455279535055e-05,
+      "loss": 0.9608,
+      "step": 2769
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.8433883659929025e-05,
+      "loss": 1.0911,
+      "step": 2770
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.8431312040322994e-05,
+      "loss": 0.9755,
+      "step": 2771
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.842874042071697e-05,
+      "loss": 0.8187,
+      "step": 2772
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.842616880111094e-05,
+      "loss": 0.9819,
+      "step": 2773
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.842359718150492e-05,
+      "loss": 0.643,
+      "step": 2774
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.842102556189889e-05,
+      "loss": 0.9235,
+      "step": 2775
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.8418453942292856e-05,
+      "loss": 0.7022,
+      "step": 2776
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.8415882322686826e-05,
+      "loss": 0.7728,
+      "step": 2777
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.84133107030808e-05,
+      "loss": 0.8668,
+      "step": 2778
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.841073908347478e-05,
+      "loss": 0.8858,
+      "step": 2779
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.840816746386875e-05,
+      "loss": 0.7283,
+      "step": 2780
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.840559584426272e-05,
+      "loss": 0.7792,
+      "step": 2781
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.840302422465669e-05,
+      "loss": 1.2677,
+      "step": 2782
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.8400452605050665e-05,
+      "loss": 1.1654,
+      "step": 2783
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.839788098544464e-05,
+      "loss": 0.7973,
+      "step": 2784
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.839530936583861e-05,
+      "loss": 0.6206,
+      "step": 2785
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.839273774623258e-05,
+      "loss": 0.9048,
+      "step": 2786
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.839016612662655e-05,
+      "loss": 0.9232,
+      "step": 2787
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.838759450702052e-05,
+      "loss": 0.8415,
+      "step": 2788
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.8385022887414496e-05,
+      "loss": 0.9537,
+      "step": 2789
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.838245126780847e-05,
+      "loss": 0.9717,
+      "step": 2790
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.837987964820244e-05,
+      "loss": 0.7867,
+      "step": 2791
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.837730802859641e-05,
+      "loss": 0.996,
+      "step": 2792
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.837473640899038e-05,
+      "loss": 1.0363,
+      "step": 2793
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.837216478938436e-05,
+      "loss": 0.8592,
+      "step": 2794
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.836959316977833e-05,
+      "loss": 0.8476,
+      "step": 2795
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.8367021550172304e-05,
+      "loss": 1.0624,
+      "step": 2796
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.8364449930566274e-05,
+      "loss": 0.6093,
+      "step": 2797
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.8361878310960244e-05,
+      "loss": 0.8889,
+      "step": 2798
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.835930669135422e-05,
+      "loss": 1.0001,
+      "step": 2799
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.835673507174819e-05,
+      "loss": 0.7324,
+      "step": 2800
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.835416345214216e-05,
+      "loss": 0.8438,
+      "step": 2801
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.8351591832536136e-05,
+      "loss": 0.7098,
+      "step": 2802
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.8349020212930106e-05,
+      "loss": 1.187,
+      "step": 2803
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.8346448593324075e-05,
+      "loss": 0.7473,
+      "step": 2804
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.834387697371805e-05,
+      "loss": 1.0399,
+      "step": 2805
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.834130535411202e-05,
+      "loss": 0.7818,
+      "step": 2806
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.833873373450599e-05,
+      "loss": 1.0829,
+      "step": 2807
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.833616211489997e-05,
+      "loss": 0.7357,
+      "step": 2808
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.833359049529394e-05,
+      "loss": 0.8393,
+      "step": 2809
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.8331018875687914e-05,
+      "loss": 0.7558,
+      "step": 2810
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.8328447256081883e-05,
+      "loss": 0.9279,
+      "step": 2811
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.832587563647585e-05,
+      "loss": 0.8232,
+      "step": 2812
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.832330401686982e-05,
+      "loss": 0.8993,
+      "step": 2813
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.83207323972638e-05,
+      "loss": 0.8912,
+      "step": 2814
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.8318160777657776e-05,
+      "loss": 1.0265,
+      "step": 2815
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.8315589158051745e-05,
+      "loss": 0.8956,
+      "step": 2816
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.8313017538445715e-05,
+      "loss": 0.798,
+      "step": 2817
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.8310445918839685e-05,
+      "loss": 0.4249,
+      "step": 2818
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.8307874299233655e-05,
+      "loss": 0.869,
+      "step": 2819
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.830530267962763e-05,
+      "loss": 0.7998,
+      "step": 2820
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.830273106002161e-05,
+      "loss": 0.9926,
+      "step": 2821
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.830015944041558e-05,
+      "loss": 0.7838,
+      "step": 2822
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.829758782080955e-05,
+      "loss": 0.6699,
+      "step": 2823
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.8295016201203516e-05,
+      "loss": 0.8737,
+      "step": 2824
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.829244458159749e-05,
+      "loss": 0.734,
+      "step": 2825
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.828987296199147e-05,
+      "loss": 0.8385,
+      "step": 2826
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.828730134238544e-05,
+      "loss": 0.5912,
+      "step": 2827
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.828472972277941e-05,
+      "loss": 1.0294,
+      "step": 2828
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.828215810317338e-05,
+      "loss": 0.7408,
+      "step": 2829
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.8279586483567355e-05,
+      "loss": 1.088,
+      "step": 2830
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.8277014863961325e-05,
+      "loss": 0.5423,
+      "step": 2831
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.82744432443553e-05,
+      "loss": 0.4677,
+      "step": 2832
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.827187162474927e-05,
+      "loss": 0.7799,
+      "step": 2833
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.826930000514324e-05,
+      "loss": 0.8973,
+      "step": 2834
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.826672838553721e-05,
+      "loss": 0.7367,
+      "step": 2835
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.8264156765931187e-05,
+      "loss": 0.773,
+      "step": 2836
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.8261585146325156e-05,
+      "loss": 0.6968,
+      "step": 2837
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.825901352671913e-05,
+      "loss": 0.7265,
+      "step": 2838
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.82564419071131e-05,
+      "loss": 0.6964,
+      "step": 2839
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.825387028750707e-05,
+      "loss": 0.676,
+      "step": 2840
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.825129866790105e-05,
+      "loss": 1.1054,
+      "step": 2841
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.824872704829502e-05,
+      "loss": 0.9552,
+      "step": 2842
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.824615542868899e-05,
+      "loss": 0.7131,
+      "step": 2843
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.8243583809082964e-05,
+      "loss": 0.8291,
+      "step": 2844
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.8241012189476934e-05,
+      "loss": 0.9705,
+      "step": 2845
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.823844056987091e-05,
+      "loss": 0.8979,
+      "step": 2846
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.823586895026488e-05,
+      "loss": 0.9373,
+      "step": 2847
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.823329733065885e-05,
+      "loss": 0.8516,
+      "step": 2848
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.823072571105282e-05,
+      "loss": 1.1275,
+      "step": 2849
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.8228154091446796e-05,
+      "loss": 0.9465,
+      "step": 2850
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.8225582471840766e-05,
+      "loss": 1.1504,
+      "step": 2851
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.822301085223474e-05,
+      "loss": 1.025,
+      "step": 2852
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.822043923262871e-05,
+      "loss": 0.7328,
+      "step": 2853
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.821786761302268e-05,
+      "loss": 1.0598,
+      "step": 2854
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.821529599341665e-05,
+      "loss": 0.9787,
+      "step": 2855
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.821272437381063e-05,
+      "loss": 0.9571,
+      "step": 2856
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.8210152754204604e-05,
+      "loss": 0.8026,
+      "step": 2857
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.8207581134598574e-05,
+      "loss": 0.5625,
+      "step": 2858
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.8205009514992544e-05,
+      "loss": 1.2715,
+      "step": 2859
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.820243789538651e-05,
+      "loss": 0.7798,
+      "step": 2860
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.819986627578049e-05,
+      "loss": 0.8628,
+      "step": 2861
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.8197294656174466e-05,
+      "loss": 1.0151,
+      "step": 2862
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.8194723036568436e-05,
+      "loss": 0.6467,
+      "step": 2863
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.8192151416962406e-05,
+      "loss": 0.7955,
+      "step": 2864
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.8189579797356375e-05,
+      "loss": 0.5663,
+      "step": 2865
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.8187008177750345e-05,
+      "loss": 0.7844,
+      "step": 2866
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.818443655814432e-05,
+      "loss": 0.7361,
+      "step": 2867
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.81818649385383e-05,
+      "loss": 1.3293,
+      "step": 2868
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.817929331893227e-05,
+      "loss": 0.8981,
+      "step": 2869
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.817672169932624e-05,
+      "loss": 0.9664,
+      "step": 2870
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.817415007972021e-05,
+      "loss": 0.8779,
+      "step": 2871
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.817157846011418e-05,
+      "loss": 1.1789,
+      "step": 2872
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.816900684050815e-05,
+      "loss": 0.8427,
+      "step": 2873
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.816643522090213e-05,
+      "loss": 1.0025,
+      "step": 2874
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.81638636012961e-05,
+      "loss": 1.078,
+      "step": 2875
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.816129198169007e-05,
+      "loss": 0.6936,
+      "step": 2876
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.8158720362084045e-05,
+      "loss": 1.0371,
+      "step": 2877
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.8156148742478015e-05,
+      "loss": 0.858,
+      "step": 2878
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.8153577122871985e-05,
+      "loss": 1.1222,
+      "step": 2879
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.815100550326596e-05,
+      "loss": 1.1318,
+      "step": 2880
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.814843388365993e-05,
+      "loss": 0.8578,
+      "step": 2881
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.81458622640539e-05,
+      "loss": 0.9588,
+      "step": 2882
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.814329064444788e-05,
+      "loss": 0.931,
+      "step": 2883
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.814071902484185e-05,
+      "loss": 0.9292,
+      "step": 2884
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.8138147405235816e-05,
+      "loss": 0.8011,
+      "step": 2885
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.813557578562979e-05,
+      "loss": 0.7383,
+      "step": 2886
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.813300416602376e-05,
+      "loss": 0.8243,
+      "step": 2887
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.813043254641774e-05,
+      "loss": 0.7817,
+      "step": 2888
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.812786092681171e-05,
+      "loss": 0.9982,
+      "step": 2889
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.812528930720568e-05,
+      "loss": 1.0045,
+      "step": 2890
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.812271768759965e-05,
+      "loss": 0.6534,
+      "step": 2891
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.8120146067993624e-05,
+      "loss": 1.134,
+      "step": 2892
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.81175744483876e-05,
+      "loss": 0.8748,
+      "step": 2893
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.811500282878157e-05,
+      "loss": 0.4834,
+      "step": 2894
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.811243120917554e-05,
+      "loss": 0.8539,
+      "step": 2895
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.810985958956951e-05,
+      "loss": 0.69,
+      "step": 2896
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.810728796996348e-05,
+      "loss": 0.8117,
+      "step": 2897
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.8104716350357456e-05,
+      "loss": 0.6773,
+      "step": 2898
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.810214473075143e-05,
+      "loss": 0.7276,
+      "step": 2899
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.80995731111454e-05,
+      "loss": 1.1227,
+      "step": 2900
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.809700149153937e-05,
+      "loss": 1.0176,
+      "step": 2901
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.809442987193334e-05,
+      "loss": 0.8393,
+      "step": 2902
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.809185825232732e-05,
+      "loss": 1.0941,
+      "step": 2903
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.8089286632721295e-05,
+      "loss": 0.8752,
+      "step": 2904
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.8086715013115264e-05,
+      "loss": 1.2238,
+      "step": 2905
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.8084143393509234e-05,
+      "loss": 0.7346,
+      "step": 2906
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.8081571773903204e-05,
+      "loss": 0.9708,
+      "step": 2907
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.807900015429718e-05,
+      "loss": 1.0436,
+      "step": 2908
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.807642853469115e-05,
+      "loss": 1.0328,
+      "step": 2909
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.8073856915085126e-05,
+      "loss": 1.0571,
+      "step": 2910
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.8071285295479096e-05,
+      "loss": 0.7967,
+      "step": 2911
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.8068713675873066e-05,
+      "loss": 1.1157,
+      "step": 2912
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.8066142056267035e-05,
+      "loss": 0.5473,
+      "step": 2913
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.806357043666101e-05,
+      "loss": 0.7542,
+      "step": 2914
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.806099881705499e-05,
+      "loss": 0.9597,
+      "step": 2915
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.805842719744896e-05,
+      "loss": 0.889,
+      "step": 2916
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.805585557784293e-05,
+      "loss": 0.9052,
+      "step": 2917
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.80532839582369e-05,
+      "loss": 0.8511,
+      "step": 2918
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.8050712338630874e-05,
+      "loss": 0.8826,
+      "step": 2919
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.8048140719024843e-05,
+      "loss": 0.6676,
+      "step": 2920
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.804556909941882e-05,
+      "loss": 1.0081,
+      "step": 2921
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.804299747981279e-05,
+      "loss": 0.9022,
+      "step": 2922
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.804042586020676e-05,
+      "loss": 1.0429,
+      "step": 2923
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.8037854240600736e-05,
+      "loss": 0.7916,
+      "step": 2924
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.8035282620994705e-05,
+      "loss": 0.7673,
+      "step": 2925
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.8032711001388675e-05,
+      "loss": 0.9802,
+      "step": 2926
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.803013938178265e-05,
+      "loss": 0.8021,
+      "step": 2927
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.802756776217662e-05,
+      "loss": 1.0344,
+      "step": 2928
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.802499614257059e-05,
+      "loss": 0.8253,
+      "step": 2929
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.802242452296457e-05,
+      "loss": 0.7143,
+      "step": 2930
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.801985290335854e-05,
+      "loss": 0.6718,
+      "step": 2931
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.801728128375251e-05,
+      "loss": 1.0097,
+      "step": 2932
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.801470966414648e-05,
+      "loss": 0.6874,
+      "step": 2933
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.801213804454045e-05,
+      "loss": 0.67,
+      "step": 2934
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.800956642493443e-05,
+      "loss": 0.7631,
+      "step": 2935
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.80069948053284e-05,
+      "loss": 0.7812,
+      "step": 2936
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.800442318572237e-05,
+      "loss": 0.6556,
+      "step": 2937
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.800185156611634e-05,
+      "loss": 0.5682,
+      "step": 2938
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.7999279946510315e-05,
+      "loss": 0.8263,
+      "step": 2939
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.799670832690429e-05,
+      "loss": 0.6538,
+      "step": 2940
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.799413670729826e-05,
+      "loss": 1.1098,
+      "step": 2941
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.799156508769223e-05,
+      "loss": 0.7692,
+      "step": 2942
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.79889934680862e-05,
+      "loss": 0.7299,
+      "step": 2943
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.798642184848017e-05,
+      "loss": 1.3693,
+      "step": 2944
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.7983850228874147e-05,
+      "loss": 0.8473,
+      "step": 2945
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.798127860926812e-05,
+      "loss": 1.2464,
+      "step": 2946
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.797870698966209e-05,
+      "loss": 0.9442,
+      "step": 2947
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.797613537005606e-05,
+      "loss": 0.6881,
+      "step": 2948
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.797356375045003e-05,
+      "loss": 0.7244,
+      "step": 2949
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.797099213084401e-05,
+      "loss": 1.0966,
+      "step": 2950
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.7968420511237985e-05,
+      "loss": 0.9539,
+      "step": 2951
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.7965848891631955e-05,
+      "loss": 1.3539,
+      "step": 2952
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.7963277272025924e-05,
+      "loss": 0.9195,
+      "step": 2953
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.7960705652419894e-05,
+      "loss": 0.558,
+      "step": 2954
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.795813403281387e-05,
+      "loss": 0.8177,
+      "step": 2955
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.795556241320784e-05,
+      "loss": 1.0669,
+      "step": 2956
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.795299079360182e-05,
+      "loss": 0.6434,
+      "step": 2957
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.7950419173995786e-05,
+      "loss": 0.7733,
+      "step": 2958
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.7947847554389756e-05,
+      "loss": 0.6732,
+      "step": 2959
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.7945275934783726e-05,
+      "loss": 0.6655,
+      "step": 2960
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.79427043151777e-05,
+      "loss": 0.8267,
+      "step": 2961
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.794013269557167e-05,
+      "loss": 0.9454,
+      "step": 2962
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.793756107596565e-05,
+      "loss": 0.9587,
+      "step": 2963
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.793498945635962e-05,
+      "loss": 0.9911,
+      "step": 2964
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.793241783675359e-05,
+      "loss": 1.1777,
+      "step": 2965
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.7929846217147564e-05,
+      "loss": 0.9916,
+      "step": 2966
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.7927274597541534e-05,
+      "loss": 0.8884,
+      "step": 2967
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.7924702977935504e-05,
+      "loss": 1.0653,
+      "step": 2968
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.792213135832948e-05,
+      "loss": 1.0167,
+      "step": 2969
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.791955973872345e-05,
+      "loss": 0.7671,
+      "step": 2970
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.7916988119117426e-05,
+      "loss": 1.1024,
+      "step": 2971
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.7914416499511396e-05,
+      "loss": 0.8537,
+      "step": 2972
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.7911844879905366e-05,
+      "loss": 0.9407,
+      "step": 2973
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.7909273260299335e-05,
+      "loss": 1.1542,
+      "step": 2974
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.790670164069331e-05,
+      "loss": 0.731,
+      "step": 2975
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.790413002108728e-05,
+      "loss": 0.7874,
+      "step": 2976
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.790155840148126e-05,
+      "loss": 0.9861,
+      "step": 2977
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.789898678187523e-05,
+      "loss": 0.9198,
+      "step": 2978
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.78964151622692e-05,
+      "loss": 0.6823,
+      "step": 2979
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.789384354266317e-05,
+      "loss": 0.8364,
+      "step": 2980
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.789127192305714e-05,
+      "loss": 1.0369,
+      "step": 2981
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.788870030345112e-05,
+      "loss": 0.8747,
+      "step": 2982
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.788612868384509e-05,
+      "loss": 0.7996,
+      "step": 2983
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.788355706423906e-05,
+      "loss": 0.5516,
+      "step": 2984
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.788098544463303e-05,
+      "loss": 1.0385,
+      "step": 2985
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.7878413825027005e-05,
+      "loss": 0.9922,
+      "step": 2986
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.787584220542098e-05,
+      "loss": 1.0568,
+      "step": 2987
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.787327058581495e-05,
+      "loss": 0.8091,
+      "step": 2988
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.787069896620892e-05,
+      "loss": 1.2625,
+      "step": 2989
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.786812734660289e-05,
+      "loss": 0.6308,
+      "step": 2990
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.786555572699686e-05,
+      "loss": 1.0018,
+      "step": 2991
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.786298410739084e-05,
+      "loss": 0.6592,
+      "step": 2992
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.7860412487784813e-05,
+      "loss": 0.9955,
+      "step": 2993
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.785784086817878e-05,
+      "loss": 0.7106,
+      "step": 2994
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.785526924857275e-05,
+      "loss": 0.9812,
+      "step": 2995
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.785269762896672e-05,
+      "loss": 0.6999,
+      "step": 2996
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.78501260093607e-05,
+      "loss": 1.1082,
+      "step": 2997
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.784755438975467e-05,
+      "loss": 0.8957,
+      "step": 2998
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.7844982770148645e-05,
+      "loss": 0.7857,
+      "step": 2999
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.7842411150542615e-05,
+      "loss": 0.7746,
+      "step": 3000
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.7839839530936584e-05,
+      "loss": 0.8408,
+      "step": 3001
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.783726791133056e-05,
+      "loss": 0.9023,
+      "step": 3002
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.783469629172453e-05,
+      "loss": 0.9287,
+      "step": 3003
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.78321246721185e-05,
+      "loss": 0.9876,
+      "step": 3004
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.782955305251248e-05,
+      "loss": 1.1,
+      "step": 3005
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.7826981432906446e-05,
+      "loss": 0.5552,
+      "step": 3006
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.7824409813300416e-05,
+      "loss": 0.7537,
+      "step": 3007
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.782183819369439e-05,
+      "loss": 0.7959,
+      "step": 3008
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.781926657408836e-05,
+      "loss": 0.7353,
+      "step": 3009
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.781669495448233e-05,
+      "loss": 0.8743,
+      "step": 3010
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.781412333487631e-05,
+      "loss": 0.7019,
+      "step": 3011
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.781155171527028e-05,
+      "loss": 0.705,
+      "step": 3012
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.7808980095664255e-05,
+      "loss": 0.7326,
+      "step": 3013
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.7806408476058224e-05,
+      "loss": 0.6494,
+      "step": 3014
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.7803836856452194e-05,
+      "loss": 0.8348,
+      "step": 3015
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.7801265236846164e-05,
+      "loss": 1.1908,
+      "step": 3016
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.779869361724014e-05,
+      "loss": 1.0746,
+      "step": 3017
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.7796121997634117e-05,
+      "loss": 0.8823,
+      "step": 3018
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.7793550378028086e-05,
+      "loss": 0.785,
+      "step": 3019
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.7790978758422056e-05,
+      "loss": 0.9381,
+      "step": 3020
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.7788407138816026e-05,
+      "loss": 0.917,
+      "step": 3021
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.7785835519209995e-05,
+      "loss": 1.1071,
+      "step": 3022
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.778326389960397e-05,
+      "loss": 0.8407,
+      "step": 3023
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.778069227999795e-05,
+      "loss": 0.9666,
+      "step": 3024
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.777812066039192e-05,
+      "loss": 0.8647,
+      "step": 3025
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.777554904078589e-05,
+      "loss": 0.7703,
+      "step": 3026
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.777297742117986e-05,
+      "loss": 0.6609,
+      "step": 3027
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.7770405801573834e-05,
+      "loss": 1.2376,
+      "step": 3028
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.776783418196781e-05,
+      "loss": 0.8189,
+      "step": 3029
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.776526256236178e-05,
+      "loss": 0.9032,
+      "step": 3030
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.776269094275575e-05,
+      "loss": 0.7354,
+      "step": 3031
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.776011932314972e-05,
+      "loss": 0.9019,
+      "step": 3032
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.7757547703543696e-05,
+      "loss": 0.7756,
+      "step": 3033
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.7754976083937665e-05,
+      "loss": 0.9249,
+      "step": 3034
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.775240446433164e-05,
+      "loss": 0.9562,
+      "step": 3035
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.774983284472561e-05,
+      "loss": 0.681,
+      "step": 3036
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.774726122511958e-05,
+      "loss": 0.975,
+      "step": 3037
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.774468960551355e-05,
+      "loss": 1.2568,
+      "step": 3038
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.774211798590753e-05,
+      "loss": 1.0539,
+      "step": 3039
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.77395463663015e-05,
+      "loss": 0.8021,
+      "step": 3040
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.7736974746695474e-05,
+      "loss": 0.6432,
+      "step": 3041
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.773440312708944e-05,
+      "loss": 1.0026,
+      "step": 3042
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.773183150748341e-05,
+      "loss": 0.8386,
+      "step": 3043
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.772925988787739e-05,
+      "loss": 0.8295,
+      "step": 3044
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.772668826827136e-05,
+      "loss": 0.924,
+      "step": 3045
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.772411664866533e-05,
+      "loss": 0.7471,
+      "step": 3046
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.7721545029059305e-05,
+      "loss": 0.8973,
+      "step": 3047
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.7718973409453275e-05,
+      "loss": 0.8385,
+      "step": 3048
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.771640178984725e-05,
+      "loss": 0.7198,
+      "step": 3049
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.771383017024122e-05,
+      "loss": 0.8452,
+      "step": 3050
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.771125855063519e-05,
+      "loss": 0.9229,
+      "step": 3051
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.770868693102916e-05,
+      "loss": 0.7618,
+      "step": 3052
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.770611531142314e-05,
+      "loss": 0.8618,
+      "step": 3053
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.7703543691817107e-05,
+      "loss": 1.1153,
+      "step": 3054
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.770097207221108e-05,
+      "loss": 1.0336,
+      "step": 3055
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.769840045260505e-05,
+      "loss": 0.8092,
+      "step": 3056
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.769582883299902e-05,
+      "loss": 0.5337,
+      "step": 3057
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.769325721339299e-05,
+      "loss": 0.9675,
+      "step": 3058
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.769068559378697e-05,
+      "loss": 0.9228,
+      "step": 3059
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.7688113974180945e-05,
+      "loss": 0.7923,
+      "step": 3060
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.7685542354574915e-05,
+      "loss": 1.0137,
+      "step": 3061
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.7682970734968884e-05,
+      "loss": 1.058,
+      "step": 3062
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.7680399115362854e-05,
+      "loss": 0.8404,
+      "step": 3063
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.767782749575683e-05,
+      "loss": 0.6632,
+      "step": 3064
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.767525587615081e-05,
+      "loss": 0.954,
+      "step": 3065
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.767268425654478e-05,
+      "loss": 1.1748,
+      "step": 3066
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.7670112636938746e-05,
+      "loss": 1.2298,
+      "step": 3067
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.7667541017332716e-05,
+      "loss": 0.9251,
+      "step": 3068
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.7664969397726686e-05,
+      "loss": 0.8135,
+      "step": 3069
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.766239777812066e-05,
+      "loss": 0.919,
+      "step": 3070
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.765982615851464e-05,
+      "loss": 0.9498,
+      "step": 3071
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.765725453890861e-05,
+      "loss": 1.0721,
+      "step": 3072
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.765468291930258e-05,
+      "loss": 0.9284,
+      "step": 3073
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.765211129969655e-05,
+      "loss": 0.5592,
+      "step": 3074
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.7649539680090524e-05,
+      "loss": 0.7092,
+      "step": 3075
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.76469680604845e-05,
+      "loss": 0.882,
+      "step": 3076
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.764439644087847e-05,
+      "loss": 0.6151,
+      "step": 3077
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.764182482127244e-05,
+      "loss": 0.8644,
+      "step": 3078
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.763925320166641e-05,
+      "loss": 0.8044,
+      "step": 3079
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.7636681582060386e-05,
+      "loss": 0.9625,
+      "step": 3080
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.7634109962454356e-05,
+      "loss": 0.6181,
+      "step": 3081
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.763153834284833e-05,
+      "loss": 0.8563,
+      "step": 3082
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.76289667232423e-05,
+      "loss": 0.8151,
+      "step": 3083
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.762639510363627e-05,
+      "loss": 1.0335,
+      "step": 3084
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.762382348403024e-05,
+      "loss": 0.7044,
+      "step": 3085
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.762125186442422e-05,
+      "loss": 0.7601,
+      "step": 3086
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.761868024481819e-05,
+      "loss": 1.0194,
+      "step": 3087
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.7616108625212164e-05,
+      "loss": 0.6613,
+      "step": 3088
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.7613537005606134e-05,
+      "loss": 1.0427,
+      "step": 3089
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.76109653860001e-05,
+      "loss": 0.9009,
+      "step": 3090
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.760839376639408e-05,
+      "loss": 1.0467,
+      "step": 3091
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.760582214678805e-05,
+      "loss": 0.9339,
+      "step": 3092
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.760325052718202e-05,
+      "loss": 0.7409,
+      "step": 3093
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.7600678907575996e-05,
+      "loss": 0.9572,
+      "step": 3094
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.7598107287969965e-05,
+      "loss": 0.9425,
+      "step": 3095
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.759553566836394e-05,
+      "loss": 0.9702,
+      "step": 3096
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.759296404875791e-05,
+      "loss": 1.0429,
+      "step": 3097
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.759039242915188e-05,
+      "loss": 0.8259,
+      "step": 3098
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.758782080954585e-05,
+      "loss": 0.9719,
+      "step": 3099
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.758524918993983e-05,
+      "loss": 0.6358,
+      "step": 3100
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.75826775703338e-05,
+      "loss": 0.6472,
+      "step": 3101
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.7580105950727773e-05,
+      "loss": 1.0786,
+      "step": 3102
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.757753433112174e-05,
+      "loss": 0.8184,
+      "step": 3103
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.757496271151571e-05,
+      "loss": 0.8929,
+      "step": 3104
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.757239109190968e-05,
+      "loss": 0.8277,
+      "step": 3105
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.756981947230366e-05,
+      "loss": 0.9024,
+      "step": 3106
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.7567247852697635e-05,
+      "loss": 0.9715,
+      "step": 3107
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.7564676233091605e-05,
+      "loss": 0.8202,
+      "step": 3108
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.7562104613485575e-05,
+      "loss": 0.7643,
+      "step": 3109
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.7559532993879544e-05,
+      "loss": 0.8721,
+      "step": 3110
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.755696137427352e-05,
+      "loss": 0.9107,
+      "step": 3111
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.75543897546675e-05,
+      "loss": 0.7287,
+      "step": 3112
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.755181813506147e-05,
+      "loss": 0.6506,
+      "step": 3113
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.754924651545544e-05,
+      "loss": 0.7141,
+      "step": 3114
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.7546674895849406e-05,
+      "loss": 0.7548,
+      "step": 3115
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.7544103276243376e-05,
+      "loss": 0.8485,
+      "step": 3116
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.754153165663735e-05,
+      "loss": 0.6481,
+      "step": 3117
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.753896003703133e-05,
+      "loss": 0.6818,
+      "step": 3118
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.75363884174253e-05,
+      "loss": 0.8376,
+      "step": 3119
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.753381679781927e-05,
+      "loss": 0.9802,
+      "step": 3120
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.753124517821324e-05,
+      "loss": 0.8753,
+      "step": 3121
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.7528673558607215e-05,
+      "loss": 0.8686,
+      "step": 3122
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.7526101939001184e-05,
+      "loss": 0.8482,
+      "step": 3123
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.752353031939516e-05,
+      "loss": 0.4782,
+      "step": 3124
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.752095869978913e-05,
+      "loss": 0.9473,
+      "step": 3125
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.75183870801831e-05,
+      "loss": 1.0302,
+      "step": 3126
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.7515815460577077e-05,
+      "loss": 0.7562,
+      "step": 3127
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.7513243840971046e-05,
+      "loss": 0.6966,
+      "step": 3128
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.7510672221365016e-05,
+      "loss": 1.0237,
+      "step": 3129
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.750810060175899e-05,
+      "loss": 0.915,
+      "step": 3130
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.750552898215296e-05,
+      "loss": 1.0115,
+      "step": 3131
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.750295736254693e-05,
+      "loss": 0.7971,
+      "step": 3132
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.750038574294091e-05,
+      "loss": 0.8124,
+      "step": 3133
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.749781412333488e-05,
+      "loss": 0.8775,
+      "step": 3134
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.749524250372885e-05,
+      "loss": 0.7249,
+      "step": 3135
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.7492670884122824e-05,
+      "loss": 0.6914,
+      "step": 3136
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.7490099264516794e-05,
+      "loss": 0.5218,
+      "step": 3137
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.748752764491077e-05,
+      "loss": 0.6372,
+      "step": 3138
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.748495602530474e-05,
+      "loss": 1.0116,
+      "step": 3139
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.748238440569871e-05,
+      "loss": 1.0123,
+      "step": 3140
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.747981278609268e-05,
+      "loss": 0.6644,
+      "step": 3141
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.7477241166486656e-05,
+      "loss": 0.5848,
+      "step": 3142
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.747466954688063e-05,
+      "loss": 0.8162,
+      "step": 3143
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.74720979272746e-05,
+      "loss": 0.9134,
+      "step": 3144
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.746952630766857e-05,
+      "loss": 0.6765,
+      "step": 3145
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.746695468806254e-05,
+      "loss": 0.9888,
+      "step": 3146
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.746438306845651e-05,
+      "loss": 0.7992,
+      "step": 3147
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.746181144885049e-05,
+      "loss": 0.7271,
+      "step": 3148
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.7459239829244464e-05,
+      "loss": 0.9129,
+      "step": 3149
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.7456668209638433e-05,
+      "loss": 0.8443,
+      "step": 3150
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.74540965900324e-05,
+      "loss": 0.7049,
+      "step": 3151
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.745152497042637e-05,
+      "loss": 1.0835,
+      "step": 3152
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.744895335082035e-05,
+      "loss": 0.5834,
+      "step": 3153
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.7446381731214326e-05,
+      "loss": 0.4676,
+      "step": 3154
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.7443810111608295e-05,
+      "loss": 0.981,
+      "step": 3155
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.7441238492002265e-05,
+      "loss": 0.8113,
+      "step": 3156
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.7438666872396235e-05,
+      "loss": 0.9471,
+      "step": 3157
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.743609525279021e-05,
+      "loss": 0.7006,
+      "step": 3158
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.743352363318418e-05,
+      "loss": 0.7902,
+      "step": 3159
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.743095201357816e-05,
+      "loss": 0.9929,
+      "step": 3160
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.742838039397213e-05,
+      "loss": 0.8241,
+      "step": 3161
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.74258087743661e-05,
+      "loss": 0.7248,
+      "step": 3162
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.7423237154760066e-05,
+      "loss": 1.1115,
+      "step": 3163
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.742066553515404e-05,
+      "loss": 0.4419,
+      "step": 3164
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.741809391554801e-05,
+      "loss": 1.3097,
+      "step": 3165
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.741552229594199e-05,
+      "loss": 0.9349,
+      "step": 3166
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.741295067633596e-05,
+      "loss": 0.6989,
+      "step": 3167
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.741037905672993e-05,
+      "loss": 0.5131,
+      "step": 3168
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.7407807437123905e-05,
+      "loss": 1.0133,
+      "step": 3169
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.7405235817517875e-05,
+      "loss": 0.9435,
+      "step": 3170
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.7402664197911844e-05,
+      "loss": 0.7034,
+      "step": 3171
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.740009257830582e-05,
+      "loss": 0.5427,
+      "step": 3172
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.739752095869979e-05,
+      "loss": 1.2157,
+      "step": 3173
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.739494933909377e-05,
+      "loss": 1.0643,
+      "step": 3174
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.7392377719487737e-05,
+      "loss": 1.1271,
+      "step": 3175
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.7389806099881706e-05,
+      "loss": 0.8544,
+      "step": 3176
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.7387234480275676e-05,
+      "loss": 0.6589,
+      "step": 3177
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.738466286066965e-05,
+      "loss": 1.0406,
+      "step": 3178
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.738209124106362e-05,
+      "loss": 0.9915,
+      "step": 3179
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.73795196214576e-05,
+      "loss": 1.1348,
+      "step": 3180
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.737694800185157e-05,
+      "loss": 0.9428,
+      "step": 3181
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.737437638224554e-05,
+      "loss": 0.767,
+      "step": 3182
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.737180476263951e-05,
+      "loss": 0.9039,
+      "step": 3183
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.7369233143033484e-05,
+      "loss": 0.6667,
+      "step": 3184
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.736666152342746e-05,
+      "loss": 0.8455,
+      "step": 3185
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.736408990382143e-05,
+      "loss": 0.8681,
+      "step": 3186
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.73615182842154e-05,
+      "loss": 0.7616,
+      "step": 3187
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.735894666460937e-05,
+      "loss": 0.6646,
+      "step": 3188
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.7356375045003346e-05,
+      "loss": 0.9818,
+      "step": 3189
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.735380342539732e-05,
+      "loss": 1.2137,
+      "step": 3190
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.735123180579129e-05,
+      "loss": 0.8768,
+      "step": 3191
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.734866018618526e-05,
+      "loss": 0.9081,
+      "step": 3192
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.734608856657923e-05,
+      "loss": 0.9666,
+      "step": 3193
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.73435169469732e-05,
+      "loss": 1.1052,
+      "step": 3194
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.734094532736718e-05,
+      "loss": 0.7501,
+      "step": 3195
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.7338373707761154e-05,
+      "loss": 0.5011,
+      "step": 3196
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.7335802088155124e-05,
+      "loss": 0.7915,
+      "step": 3197
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.7333230468549094e-05,
+      "loss": 0.8221,
+      "step": 3198
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.733065884894306e-05,
+      "loss": 1.0774,
+      "step": 3199
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.732808722933704e-05,
+      "loss": 0.9777,
+      "step": 3200
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.732551560973101e-05,
+      "loss": 0.795,
+      "step": 3201
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.7322943990124986e-05,
+      "loss": 1.0307,
+      "step": 3202
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.7320372370518956e-05,
+      "loss": 1.1358,
+      "step": 3203
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.7317800750912925e-05,
+      "loss": 0.964,
+      "step": 3204
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.73152291313069e-05,
+      "loss": 0.9942,
+      "step": 3205
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.731265751170087e-05,
+      "loss": 0.9469,
+      "step": 3206
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.731008589209484e-05,
+      "loss": 0.8653,
+      "step": 3207
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.730751427248882e-05,
+      "loss": 0.8503,
+      "step": 3208
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.730494265288279e-05,
+      "loss": 1.0118,
+      "step": 3209
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.730237103327676e-05,
+      "loss": 0.7444,
+      "step": 3210
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.729979941367073e-05,
+      "loss": 0.8375,
+      "step": 3211
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.72972277940647e-05,
+      "loss": 1.2671,
+      "step": 3212
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.729465617445867e-05,
+      "loss": 0.714,
+      "step": 3213
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.729208455485265e-05,
+      "loss": 1.0976,
+      "step": 3214
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.728951293524662e-05,
+      "loss": 0.7125,
+      "step": 3215
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.7286941315640595e-05,
+      "loss": 0.8854,
+      "step": 3216
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.7284369696034565e-05,
+      "loss": 0.6984,
+      "step": 3217
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.7281798076428535e-05,
+      "loss": 1.0892,
+      "step": 3218
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.727922645682251e-05,
+      "loss": 0.9962,
+      "step": 3219
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.727665483721648e-05,
+      "loss": 0.9972,
+      "step": 3220
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.727408321761046e-05,
+      "loss": 0.8701,
+      "step": 3221
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.727151159800443e-05,
+      "loss": 0.7645,
+      "step": 3222
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.72689399783984e-05,
+      "loss": 0.8768,
+      "step": 3223
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.7266368358792366e-05,
+      "loss": 0.7582,
+      "step": 3224
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.726379673918634e-05,
+      "loss": 0.9085,
+      "step": 3225
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.726122511958031e-05,
+      "loss": 0.8249,
+      "step": 3226
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.725865349997429e-05,
+      "loss": 0.5406,
+      "step": 3227
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.725608188036826e-05,
+      "loss": 0.9879,
+      "step": 3228
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.725351026076223e-05,
+      "loss": 1.0083,
+      "step": 3229
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.72509386411562e-05,
+      "loss": 0.7951,
+      "step": 3230
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.7248367021550175e-05,
+      "loss": 0.6422,
+      "step": 3231
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.724579540194415e-05,
+      "loss": 1.1022,
+      "step": 3232
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.724322378233812e-05,
+      "loss": 0.8768,
+      "step": 3233
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.724065216273209e-05,
+      "loss": 0.6672,
+      "step": 3234
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.723808054312606e-05,
+      "loss": 0.6541,
+      "step": 3235
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.7235508923520036e-05,
+      "loss": 0.7421,
+      "step": 3236
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.723293730391401e-05,
+      "loss": 0.7492,
+      "step": 3237
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.723036568430798e-05,
+      "loss": 0.4166,
+      "step": 3238
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.722779406470195e-05,
+      "loss": 0.654,
+      "step": 3239
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.722522244509592e-05,
+      "loss": 0.9588,
+      "step": 3240
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.722265082548989e-05,
+      "loss": 0.6866,
+      "step": 3241
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.722007920588387e-05,
+      "loss": 1.0704,
+      "step": 3242
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.7217507586277845e-05,
+      "loss": 0.5835,
+      "step": 3243
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.7214935966671814e-05,
+      "loss": 0.7586,
+      "step": 3244
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.7212364347065784e-05,
+      "loss": 0.849,
+      "step": 3245
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.7209792727459754e-05,
+      "loss": 1.065,
+      "step": 3246
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.720722110785373e-05,
+      "loss": 0.9251,
+      "step": 3247
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.72046494882477e-05,
+      "loss": 0.8909,
+      "step": 3248
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.7202077868641676e-05,
+      "loss": 0.6544,
+      "step": 3249
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.7199506249035646e-05,
+      "loss": 0.7752,
+      "step": 3250
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.7196934629429616e-05,
+      "loss": 0.9937,
+      "step": 3251
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.719436300982359e-05,
+      "loss": 0.6725,
+      "step": 3252
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.719179139021756e-05,
+      "loss": 0.6734,
+      "step": 3253
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.718921977061153e-05,
+      "loss": 0.987,
+      "step": 3254
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.718664815100551e-05,
+      "loss": 0.9305,
+      "step": 3255
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.718407653139948e-05,
+      "loss": 0.6989,
+      "step": 3256
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.718150491179345e-05,
+      "loss": 0.608,
+      "step": 3257
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.7178933292187424e-05,
+      "loss": 0.6002,
+      "step": 3258
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.7176361672581393e-05,
+      "loss": 0.6644,
+      "step": 3259
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.717379005297536e-05,
+      "loss": 0.9069,
+      "step": 3260
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.717121843336934e-05,
+      "loss": 0.6379,
+      "step": 3261
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.716864681376331e-05,
+      "loss": 0.9708,
+      "step": 3262
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.7166075194157286e-05,
+      "loss": 1.1447,
+      "step": 3263
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.7163503574551255e-05,
+      "loss": 0.7807,
+      "step": 3264
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.7160931954945225e-05,
+      "loss": 0.8057,
+      "step": 3265
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.7158360335339195e-05,
+      "loss": 0.9508,
+      "step": 3266
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.715578871573317e-05,
+      "loss": 0.6357,
+      "step": 3267
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.715321709612715e-05,
+      "loss": 0.9633,
+      "step": 3268
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.715064547652112e-05,
+      "loss": 0.9683,
+      "step": 3269
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.714807385691509e-05,
+      "loss": 1.1331,
+      "step": 3270
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.714550223730906e-05,
+      "loss": 0.655,
+      "step": 3271
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.7142930617703026e-05,
+      "loss": 0.8315,
+      "step": 3272
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.7140358998097e-05,
+      "loss": 0.8447,
+      "step": 3273
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.713778737849098e-05,
+      "loss": 0.8611,
+      "step": 3274
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.713521575888495e-05,
+      "loss": 0.9729,
+      "step": 3275
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.713264413927892e-05,
+      "loss": 0.6132,
+      "step": 3276
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.713007251967289e-05,
+      "loss": 0.9018,
+      "step": 3277
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.7127500900066865e-05,
+      "loss": 1.0378,
+      "step": 3278
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.712492928046084e-05,
+      "loss": 0.7636,
+      "step": 3279
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.712235766085481e-05,
+      "loss": 0.6292,
+      "step": 3280
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.711978604124878e-05,
+      "loss": 1.1229,
+      "step": 3281
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.711721442164275e-05,
+      "loss": 0.8709,
+      "step": 3282
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.711464280203673e-05,
+      "loss": 0.9342,
+      "step": 3283
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.7112071182430697e-05,
+      "loss": 0.5598,
+      "step": 3284
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.710949956282467e-05,
+      "loss": 0.9875,
+      "step": 3285
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.710692794321864e-05,
+      "loss": 0.7527,
+      "step": 3286
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.710435632361261e-05,
+      "loss": 0.5526,
+      "step": 3287
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.710178470400658e-05,
+      "loss": 0.7188,
+      "step": 3288
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.709921308440056e-05,
+      "loss": 1.1787,
+      "step": 3289
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.709664146479453e-05,
+      "loss": 0.675,
+      "step": 3290
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.7094069845188505e-05,
+      "loss": 0.8468,
+      "step": 3291
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.7091498225582474e-05,
+      "loss": 0.8649,
+      "step": 3292
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.7088926605976444e-05,
+      "loss": 0.6567,
+      "step": 3293
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.708635498637042e-05,
+      "loss": 0.8095,
+      "step": 3294
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.708378336676439e-05,
+      "loss": 0.8037,
+      "step": 3295
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.708121174715836e-05,
+      "loss": 0.7191,
+      "step": 3296
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.7078640127552336e-05,
+      "loss": 0.8287,
+      "step": 3297
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.7076068507946306e-05,
+      "loss": 0.8121,
+      "step": 3298
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.707349688834028e-05,
+      "loss": 0.6364,
+      "step": 3299
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.707092526873425e-05,
+      "loss": 0.7288,
+      "step": 3300
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.706835364912822e-05,
+      "loss": 1.1463,
+      "step": 3301
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.706578202952219e-05,
+      "loss": 1.0156,
+      "step": 3302
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.706321040991617e-05,
+      "loss": 0.7596,
+      "step": 3303
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.706063879031014e-05,
+      "loss": 0.8153,
+      "step": 3304
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.7058067170704114e-05,
+      "loss": 0.6892,
+      "step": 3305
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.7055495551098084e-05,
+      "loss": 0.8331,
+      "step": 3306
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.7052923931492054e-05,
+      "loss": 0.8199,
+      "step": 3307
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.705035231188602e-05,
+      "loss": 0.7935,
+      "step": 3308
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.704778069228e-05,
+      "loss": 0.6766,
+      "step": 3309
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.7045209072673976e-05,
+      "loss": 0.9262,
+      "step": 3310
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.7042637453067946e-05,
+      "loss": 0.8384,
+      "step": 3311
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.7040065833461916e-05,
+      "loss": 0.9859,
+      "step": 3312
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.7037494213855885e-05,
+      "loss": 1.0783,
+      "step": 3313
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.703492259424986e-05,
+      "loss": 0.8244,
+      "step": 3314
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.703235097464384e-05,
+      "loss": 0.8807,
+      "step": 3315
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.702977935503781e-05,
+      "loss": 0.7189,
+      "step": 3316
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.702720773543178e-05,
+      "loss": 0.9351,
+      "step": 3317
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.702463611582575e-05,
+      "loss": 1.0372,
+      "step": 3318
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.702206449621972e-05,
+      "loss": 0.9577,
+      "step": 3319
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.701949287661369e-05,
+      "loss": 1.1514,
+      "step": 3320
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.701692125700767e-05,
+      "loss": 0.9428,
+      "step": 3321
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.701434963740164e-05,
+      "loss": 0.8282,
+      "step": 3322
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.701177801779561e-05,
+      "loss": 0.6283,
+      "step": 3323
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.700920639818958e-05,
+      "loss": 0.724,
+      "step": 3324
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.7006634778583555e-05,
+      "loss": 0.7653,
+      "step": 3325
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.7004063158977525e-05,
+      "loss": 0.8022,
+      "step": 3326
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.70014915393715e-05,
+      "loss": 0.7419,
+      "step": 3327
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.699891991976547e-05,
+      "loss": 0.8018,
+      "step": 3328
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.699634830015944e-05,
+      "loss": 0.9683,
+      "step": 3329
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.699377668055342e-05,
+      "loss": 0.8382,
+      "step": 3330
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.699120506094739e-05,
+      "loss": 0.779,
+      "step": 3331
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.698863344134136e-05,
+      "loss": 0.73,
+      "step": 3332
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.698606182173533e-05,
+      "loss": 0.6001,
+      "step": 3333
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.69834902021293e-05,
+      "loss": 0.7214,
+      "step": 3334
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.698091858252327e-05,
+      "loss": 1.0965,
+      "step": 3335
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.697834696291725e-05,
+      "loss": 0.8574,
+      "step": 3336
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.697577534331122e-05,
+      "loss": 0.6275,
+      "step": 3337
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.697320372370519e-05,
+      "loss": 0.8248,
+      "step": 3338
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.6970632104099165e-05,
+      "loss": 1.041,
+      "step": 3339
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.6968060484493134e-05,
+      "loss": 0.9478,
+      "step": 3340
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.696548886488711e-05,
+      "loss": 0.5798,
+      "step": 3341
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.696291724528108e-05,
+      "loss": 0.9801,
+      "step": 3342
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.696034562567505e-05,
+      "loss": 0.9748,
+      "step": 3343
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.695777400606902e-05,
+      "loss": 0.8435,
+      "step": 3344
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.6955202386462996e-05,
+      "loss": 0.8613,
+      "step": 3345
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.695263076685697e-05,
+      "loss": 1.2192,
+      "step": 3346
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.695005914725094e-05,
+      "loss": 0.9807,
+      "step": 3347
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.694748752764491e-05,
+      "loss": 0.6411,
+      "step": 3348
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.694491590803888e-05,
+      "loss": 0.8144,
+      "step": 3349
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.694234428843285e-05,
+      "loss": 0.7241,
+      "step": 3350
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.693977266882683e-05,
+      "loss": 0.8616,
+      "step": 3351
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.6937201049220805e-05,
+      "loss": 0.664,
+      "step": 3352
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.6934629429614774e-05,
+      "loss": 0.8312,
+      "step": 3353
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.6932057810008744e-05,
+      "loss": 0.6818,
+      "step": 3354
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.6929486190402714e-05,
+      "loss": 0.9556,
+      "step": 3355
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.692691457079669e-05,
+      "loss": 0.8449,
+      "step": 3356
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.6924342951190667e-05,
+      "loss": 1.2527,
+      "step": 3357
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.6921771331584636e-05,
+      "loss": 1.0393,
+      "step": 3358
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.6919199711978606e-05,
+      "loss": 0.7504,
+      "step": 3359
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.6916628092372576e-05,
+      "loss": 0.867,
+      "step": 3360
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.691405647276655e-05,
+      "loss": 1.1254,
+      "step": 3361
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.691148485316052e-05,
+      "loss": 0.9982,
+      "step": 3362
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.69089132335545e-05,
+      "loss": 1.0219,
+      "step": 3363
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.690634161394847e-05,
+      "loss": 0.6895,
+      "step": 3364
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.690376999434244e-05,
+      "loss": 0.6301,
+      "step": 3365
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.690119837473641e-05,
+      "loss": 0.8761,
+      "step": 3366
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.6898626755130384e-05,
+      "loss": 0.7615,
+      "step": 3367
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.689605513552436e-05,
+      "loss": 0.8376,
+      "step": 3368
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.689348351591833e-05,
+      "loss": 1.0938,
+      "step": 3369
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.68909118963123e-05,
+      "loss": 0.8687,
+      "step": 3370
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.688834027670627e-05,
+      "loss": 0.8209,
+      "step": 3371
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.6885768657100246e-05,
+      "loss": 0.982,
+      "step": 3372
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.6883197037494215e-05,
+      "loss": 0.9636,
+      "step": 3373
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.688062541788819e-05,
+      "loss": 1.0944,
+      "step": 3374
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.687805379828216e-05,
+      "loss": 0.8214,
+      "step": 3375
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.687548217867613e-05,
+      "loss": 0.859,
+      "step": 3376
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.687291055907011e-05,
+      "loss": 0.7558,
+      "step": 3377
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.687033893946408e-05,
+      "loss": 0.6489,
+      "step": 3378
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.686776731985805e-05,
+      "loss": 0.6595,
+      "step": 3379
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.6865195700252024e-05,
+      "loss": 1.0054,
+      "step": 3380
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.686262408064599e-05,
+      "loss": 0.9864,
+      "step": 3381
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.686005246103996e-05,
+      "loss": 0.724,
+      "step": 3382
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.685748084143394e-05,
+      "loss": 0.6147,
+      "step": 3383
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.685490922182791e-05,
+      "loss": 0.9314,
+      "step": 3384
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.685233760222188e-05,
+      "loss": 0.7806,
+      "step": 3385
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.6849765982615855e-05,
+      "loss": 0.7383,
+      "step": 3386
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.6847194363009825e-05,
+      "loss": 0.8795,
+      "step": 3387
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.68446227434038e-05,
+      "loss": 0.5692,
+      "step": 3388
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.684205112379777e-05,
+      "loss": 0.4844,
+      "step": 3389
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.683947950419174e-05,
+      "loss": 0.6806,
+      "step": 3390
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.683690788458571e-05,
+      "loss": 1.025,
+      "step": 3391
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.683433626497969e-05,
+      "loss": 0.8899,
+      "step": 3392
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.683176464537366e-05,
+      "loss": 0.7436,
+      "step": 3393
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.682919302576763e-05,
+      "loss": 0.8485,
+      "step": 3394
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.68266214061616e-05,
+      "loss": 0.8155,
+      "step": 3395
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.682404978655557e-05,
+      "loss": 1.2401,
+      "step": 3396
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.682147816694954e-05,
+      "loss": 1.2958,
+      "step": 3397
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.681890654734352e-05,
+      "loss": 0.5919,
+      "step": 3398
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.6816334927737495e-05,
+      "loss": 0.9442,
+      "step": 3399
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.6813763308131465e-05,
+      "loss": 0.6642,
+      "step": 3400
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.6811191688525434e-05,
+      "loss": 1.0191,
+      "step": 3401
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.6808620068919404e-05,
+      "loss": 0.9033,
+      "step": 3402
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.680604844931338e-05,
+      "loss": 1.3374,
+      "step": 3403
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.680347682970736e-05,
+      "loss": 0.9063,
+      "step": 3404
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.680090521010133e-05,
+      "loss": 0.6973,
+      "step": 3405
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.6798333590495296e-05,
+      "loss": 0.9177,
+      "step": 3406
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.6795761970889266e-05,
+      "loss": 1.0482,
+      "step": 3407
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.679319035128324e-05,
+      "loss": 0.8615,
+      "step": 3408
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.679061873167721e-05,
+      "loss": 0.7808,
+      "step": 3409
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.678804711207119e-05,
+      "loss": 0.7526,
+      "step": 3410
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.678547549246516e-05,
+      "loss": 0.8227,
+      "step": 3411
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.678290387285913e-05,
+      "loss": 0.7469,
+      "step": 3412
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.67803322532531e-05,
+      "loss": 0.63,
+      "step": 3413
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.6777760633647074e-05,
+      "loss": 0.8621,
+      "step": 3414
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.6775189014041044e-05,
+      "loss": 1.0286,
+      "step": 3415
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.677261739443502e-05,
+      "loss": 0.9707,
+      "step": 3416
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.677004577482899e-05,
+      "loss": 0.7873,
+      "step": 3417
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.676747415522296e-05,
+      "loss": 0.8824,
+      "step": 3418
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.6764902535616936e-05,
+      "loss": 0.6476,
+      "step": 3419
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.6762330916010906e-05,
+      "loss": 0.857,
+      "step": 3420
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.6759759296404875e-05,
+      "loss": 0.7257,
+      "step": 3421
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.675718767679885e-05,
+      "loss": 0.7629,
+      "step": 3422
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.675461605719282e-05,
+      "loss": 0.838,
+      "step": 3423
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.67520444375868e-05,
+      "loss": 0.7738,
+      "step": 3424
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.674947281798077e-05,
+      "loss": 0.8871,
+      "step": 3425
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.674690119837474e-05,
+      "loss": 1.0029,
+      "step": 3426
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.674432957876871e-05,
+      "loss": 0.7009,
+      "step": 3427
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.6741757959162684e-05,
+      "loss": 1.0848,
+      "step": 3428
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.673918633955665e-05,
+      "loss": 0.8087,
+      "step": 3429
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.673661471995063e-05,
+      "loss": 0.9887,
+      "step": 3430
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.67340431003446e-05,
+      "loss": 0.7862,
+      "step": 3431
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.673147148073857e-05,
+      "loss": 0.7237,
+      "step": 3432
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.672889986113254e-05,
+      "loss": 0.8316,
+      "step": 3433
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.6726328241526515e-05,
+      "loss": 1.2757,
+      "step": 3434
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.672375662192049e-05,
+      "loss": 0.6694,
+      "step": 3435
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.672118500231446e-05,
+      "loss": 0.9176,
+      "step": 3436
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.671861338270843e-05,
+      "loss": 0.754,
+      "step": 3437
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.67160417631024e-05,
+      "loss": 1.077,
+      "step": 3438
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.671347014349638e-05,
+      "loss": 0.7408,
+      "step": 3439
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.6710898523890354e-05,
+      "loss": 0.9094,
+      "step": 3440
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.6708326904284323e-05,
+      "loss": 0.6177,
+      "step": 3441
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.670575528467829e-05,
+      "loss": 0.5448,
+      "step": 3442
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.670318366507226e-05,
+      "loss": 0.6352,
+      "step": 3443
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.670061204546623e-05,
+      "loss": 0.9475,
+      "step": 3444
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.669804042586021e-05,
+      "loss": 1.2757,
+      "step": 3445
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.6695468806254185e-05,
+      "loss": 0.8763,
+      "step": 3446
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.6692897186648155e-05,
+      "loss": 0.4111,
+      "step": 3447
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.6690325567042125e-05,
+      "loss": 1.3656,
+      "step": 3448
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.6687753947436094e-05,
+      "loss": 0.7456,
+      "step": 3449
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.668518232783007e-05,
+      "loss": 1.028,
+      "step": 3450
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.668261070822404e-05,
+      "loss": 0.7856,
+      "step": 3451
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.668003908861802e-05,
+      "loss": 0.786,
+      "step": 3452
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.667746746901199e-05,
+      "loss": 0.8606,
+      "step": 3453
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.6674895849405956e-05,
+      "loss": 0.7972,
+      "step": 3454
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.667232422979993e-05,
+      "loss": 0.8083,
+      "step": 3455
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.66697526101939e-05,
+      "loss": 0.5508,
+      "step": 3456
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.666718099058787e-05,
+      "loss": 0.9835,
+      "step": 3457
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.666460937098185e-05,
+      "loss": 1.172,
+      "step": 3458
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.666203775137582e-05,
+      "loss": 0.9927,
+      "step": 3459
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.665946613176979e-05,
+      "loss": 0.8646,
+      "step": 3460
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.6656894512163765e-05,
+      "loss": 0.6452,
+      "step": 3461
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.6654322892557734e-05,
+      "loss": 0.9597,
+      "step": 3462
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.6651751272951704e-05,
+      "loss": 1.1055,
+      "step": 3463
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.664917965334568e-05,
+      "loss": 0.8336,
+      "step": 3464
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.664660803373965e-05,
+      "loss": 0.8938,
+      "step": 3465
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.6644036414133627e-05,
+      "loss": 0.8097,
+      "step": 3466
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.6641464794527596e-05,
+      "loss": 0.9548,
+      "step": 3467
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.6638893174921566e-05,
+      "loss": 1.0752,
+      "step": 3468
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.6636321555315536e-05,
+      "loss": 1.1351,
+      "step": 3469
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.663374993570951e-05,
+      "loss": 0.8273,
+      "step": 3470
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.663117831610349e-05,
+      "loss": 0.8661,
+      "step": 3471
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.662860669649746e-05,
+      "loss": 0.8103,
+      "step": 3472
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.662603507689143e-05,
+      "loss": 1.0246,
+      "step": 3473
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.66234634572854e-05,
+      "loss": 1.0211,
+      "step": 3474
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.662089183767937e-05,
+      "loss": 1.0897,
+      "step": 3475
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.6618320218073344e-05,
+      "loss": 1.0771,
+      "step": 3476
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.661574859846732e-05,
+      "loss": 0.9875,
+      "step": 3477
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.661317697886129e-05,
+      "loss": 0.8296,
+      "step": 3478
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.661060535925526e-05,
+      "loss": 0.8311,
+      "step": 3479
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.660803373964923e-05,
+      "loss": 0.7412,
+      "step": 3480
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.6605462120043206e-05,
+      "loss": 0.7522,
+      "step": 3481
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.660289050043718e-05,
+      "loss": 0.7232,
+      "step": 3482
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.660031888083115e-05,
+      "loss": 0.9065,
+      "step": 3483
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.659774726122512e-05,
+      "loss": 0.603,
+      "step": 3484
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.659517564161909e-05,
+      "loss": 1.085,
+      "step": 3485
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.659260402201307e-05,
+      "loss": 0.9821,
+      "step": 3486
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.659003240240704e-05,
+      "loss": 0.8783,
+      "step": 3487
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.6587460782801014e-05,
+      "loss": 0.6712,
+      "step": 3488
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.6584889163194983e-05,
+      "loss": 0.6865,
+      "step": 3489
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.658231754358895e-05,
+      "loss": 0.6227,
+      "step": 3490
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.657974592398292e-05,
+      "loss": 0.764,
+      "step": 3491
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.65771743043769e-05,
+      "loss": 1.081,
+      "step": 3492
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.657460268477087e-05,
+      "loss": 0.996,
+      "step": 3493
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.6572031065164845e-05,
+      "loss": 0.6996,
+      "step": 3494
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.6569459445558815e-05,
+      "loss": 0.4726,
+      "step": 3495
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.6566887825952785e-05,
+      "loss": 0.9568,
+      "step": 3496
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.656431620634676e-05,
+      "loss": 0.7841,
+      "step": 3497
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.656174458674073e-05,
+      "loss": 0.7173,
+      "step": 3498
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.65591729671347e-05,
+      "loss": 0.9625,
+      "step": 3499
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.655660134752868e-05,
+      "loss": 1.0709,
+      "step": 3500
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.655402972792265e-05,
+      "loss": 0.7922,
+      "step": 3501
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.655145810831662e-05,
+      "loss": 0.6088,
+      "step": 3502
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.654888648871059e-05,
+      "loss": 0.7433,
+      "step": 3503
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.654631486910456e-05,
+      "loss": 0.6224,
+      "step": 3504
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.654374324949853e-05,
+      "loss": 0.9159,
+      "step": 3505
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.654117162989251e-05,
+      "loss": 1.0374,
+      "step": 3506
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.653860001028648e-05,
+      "loss": 0.8574,
+      "step": 3507
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.6536028390680455e-05,
+      "loss": 0.7382,
+      "step": 3508
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.6533456771074425e-05,
+      "loss": 0.8716,
+      "step": 3509
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.6530885151468394e-05,
+      "loss": 0.5832,
+      "step": 3510
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.6528313531862364e-05,
+      "loss": 0.7681,
+      "step": 3511
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.652574191225634e-05,
+      "loss": 0.7876,
+      "step": 3512
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.652317029265032e-05,
+      "loss": 0.6325,
+      "step": 3513
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.652059867304429e-05,
+      "loss": 0.4495,
+      "step": 3514
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.6518027053438256e-05,
+      "loss": 0.8871,
+      "step": 3515
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.6515455433832226e-05,
+      "loss": 0.6476,
+      "step": 3516
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.65128838142262e-05,
+      "loss": 1.0115,
+      "step": 3517
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.651031219462018e-05,
+      "loss": 1.7392,
+      "step": 3518
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.650774057501415e-05,
+      "loss": 0.5831,
+      "step": 3519
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.650516895540812e-05,
+      "loss": 0.7618,
+      "step": 3520
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.650259733580209e-05,
+      "loss": 0.9569,
+      "step": 3521
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.650002571619606e-05,
+      "loss": 1.1305,
+      "step": 3522
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.6497454096590034e-05,
+      "loss": 1.076,
+      "step": 3523
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.649488247698401e-05,
+      "loss": 0.69,
+      "step": 3524
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.649231085737798e-05,
+      "loss": 0.9689,
+      "step": 3525
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.648973923777195e-05,
+      "loss": 0.4693,
+      "step": 3526
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.648716761816592e-05,
+      "loss": 0.7194,
+      "step": 3527
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.6484595998559896e-05,
+      "loss": 0.7388,
+      "step": 3528
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.648202437895387e-05,
+      "loss": 0.8047,
+      "step": 3529
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.647945275934784e-05,
+      "loss": 1.0319,
+      "step": 3530
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.647688113974181e-05,
+      "loss": 1.0302,
+      "step": 3531
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.647430952013578e-05,
+      "loss": 0.83,
+      "step": 3532
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.647173790052976e-05,
+      "loss": 0.7822,
+      "step": 3533
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.646916628092373e-05,
+      "loss": 0.8607,
+      "step": 3534
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.6466594661317704e-05,
+      "loss": 0.8221,
+      "step": 3535
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.6464023041711674e-05,
+      "loss": 0.6181,
+      "step": 3536
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.6461451422105644e-05,
+      "loss": 0.8729,
+      "step": 3537
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.645887980249961e-05,
+      "loss": 0.4733,
+      "step": 3538
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.645630818289359e-05,
+      "loss": 0.9912,
+      "step": 3539
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.645373656328756e-05,
+      "loss": 0.5487,
+      "step": 3540
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.6451164943681536e-05,
+      "loss": 0.9782,
+      "step": 3541
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.6448593324075506e-05,
+      "loss": 0.8216,
+      "step": 3542
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.6446021704469475e-05,
+      "loss": 0.7269,
+      "step": 3543
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.644345008486345e-05,
+      "loss": 0.7046,
+      "step": 3544
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.644087846525742e-05,
+      "loss": 0.6286,
+      "step": 3545
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.643830684565139e-05,
+      "loss": 1.0079,
+      "step": 3546
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.643573522604537e-05,
+      "loss": 1.0076,
+      "step": 3547
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.643316360643934e-05,
+      "loss": 0.9445,
+      "step": 3548
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.6430591986833314e-05,
+      "loss": 1.0851,
+      "step": 3549
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.642802036722728e-05,
+      "loss": 0.9191,
+      "step": 3550
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.642544874762125e-05,
+      "loss": 0.7209,
+      "step": 3551
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.642287712801522e-05,
+      "loss": 0.5559,
+      "step": 3552
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.64203055084092e-05,
+      "loss": 0.7824,
+      "step": 3553
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.641773388880317e-05,
+      "loss": 0.6262,
+      "step": 3554
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.6415162269197145e-05,
+      "loss": 0.7978,
+      "step": 3555
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.6412590649591115e-05,
+      "loss": 0.7783,
+      "step": 3556
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.6410019029985085e-05,
+      "loss": 0.8141,
+      "step": 3557
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.6407447410379054e-05,
+      "loss": 0.75,
+      "step": 3558
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.640487579077303e-05,
+      "loss": 0.9651,
+      "step": 3559
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.640230417116701e-05,
+      "loss": 0.8059,
+      "step": 3560
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.639973255156098e-05,
+      "loss": 1.0251,
+      "step": 3561
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.639716093195495e-05,
+      "loss": 0.594,
+      "step": 3562
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.6394589312348916e-05,
+      "loss": 1.0147,
+      "step": 3563
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.639201769274289e-05,
+      "loss": 0.5805,
+      "step": 3564
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.638944607313687e-05,
+      "loss": 0.5782,
+      "step": 3565
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.638687445353084e-05,
+      "loss": 0.7842,
+      "step": 3566
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.638430283392481e-05,
+      "loss": 0.6727,
+      "step": 3567
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.638173121431878e-05,
+      "loss": 0.8097,
+      "step": 3568
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.637915959471275e-05,
+      "loss": 0.667,
+      "step": 3569
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.6376587975106725e-05,
+      "loss": 0.8159,
+      "step": 3570
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.63740163555007e-05,
+      "loss": 0.7709,
+      "step": 3571
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.637144473589467e-05,
+      "loss": 0.8423,
+      "step": 3572
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.636887311628864e-05,
+      "loss": 0.9216,
+      "step": 3573
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.636630149668261e-05,
+      "loss": 0.8883,
+      "step": 3574
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.6363729877076586e-05,
+      "loss": 0.9007,
+      "step": 3575
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.6361158257470556e-05,
+      "loss": 0.9653,
+      "step": 3576
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.635858663786453e-05,
+      "loss": 0.8156,
+      "step": 3577
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.63560150182585e-05,
+      "loss": 0.6841,
+      "step": 3578
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.635344339865247e-05,
+      "loss": 0.8847,
+      "step": 3579
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.635087177904645e-05,
+      "loss": 0.6383,
+      "step": 3580
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.634830015944042e-05,
+      "loss": 0.7798,
+      "step": 3581
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.634572853983439e-05,
+      "loss": 1.0539,
+      "step": 3582
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.6343156920228364e-05,
+      "loss": 0.6795,
+      "step": 3583
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.6340585300622334e-05,
+      "loss": 0.9689,
+      "step": 3584
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.6338013681016304e-05,
+      "loss": 0.6062,
+      "step": 3585
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.633544206141028e-05,
+      "loss": 0.6811,
+      "step": 3586
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.633287044180425e-05,
+      "loss": 0.7361,
+      "step": 3587
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.633029882219822e-05,
+      "loss": 1.1389,
+      "step": 3588
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.6327727202592196e-05,
+      "loss": 0.5964,
+      "step": 3589
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.6325155582986166e-05,
+      "loss": 1.0505,
+      "step": 3590
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.632258396338014e-05,
+      "loss": 0.8449,
+      "step": 3591
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.632001234377411e-05,
+      "loss": 0.9827,
+      "step": 3592
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.631744072416808e-05,
+      "loss": 0.706,
+      "step": 3593
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.631486910456205e-05,
+      "loss": 0.8675,
+      "step": 3594
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.631229748495603e-05,
+      "loss": 1.0397,
+      "step": 3595
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.6309725865350004e-05,
+      "loss": 0.9395,
+      "step": 3596
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.6307154245743974e-05,
+      "loss": 0.8591,
+      "step": 3597
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.6304582626137943e-05,
+      "loss": 1.2611,
+      "step": 3598
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.630201100653191e-05,
+      "loss": 0.93,
+      "step": 3599
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.629943938692588e-05,
+      "loss": 0.8498,
+      "step": 3600
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.629686776731986e-05,
+      "loss": 0.7084,
+      "step": 3601
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.6294296147713836e-05,
+      "loss": 0.9906,
+      "step": 3602
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.6291724528107805e-05,
+      "loss": 1.0034,
+      "step": 3603
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.6289152908501775e-05,
+      "loss": 0.7656,
+      "step": 3604
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.6286581288895745e-05,
+      "loss": 0.9217,
+      "step": 3605
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.628400966928972e-05,
+      "loss": 0.733,
+      "step": 3606
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.62814380496837e-05,
+      "loss": 0.8391,
+      "step": 3607
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.627886643007767e-05,
+      "loss": 0.8635,
+      "step": 3608
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.627629481047164e-05,
+      "loss": 0.7805,
+      "step": 3609
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.627372319086561e-05,
+      "loss": 0.8701,
+      "step": 3610
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.627115157125958e-05,
+      "loss": 0.9653,
+      "step": 3611
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.626857995165355e-05,
+      "loss": 0.6654,
+      "step": 3612
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.626600833204753e-05,
+      "loss": 0.6828,
+      "step": 3613
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.62634367124415e-05,
+      "loss": 0.8664,
+      "step": 3614
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.626086509283547e-05,
+      "loss": 0.9259,
+      "step": 3615
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.625829347322944e-05,
+      "loss": 0.726,
+      "step": 3616
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.6255721853623415e-05,
+      "loss": 0.7176,
+      "step": 3617
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.6253150234017385e-05,
+      "loss": 1.0275,
+      "step": 3618
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.625057861441136e-05,
+      "loss": 0.6198,
+      "step": 3619
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.624800699480533e-05,
+      "loss": 0.6317,
+      "step": 3620
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.62454353751993e-05,
+      "loss": 0.757,
+      "step": 3621
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.624286375559328e-05,
+      "loss": 0.9053,
+      "step": 3622
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.6240292135987247e-05,
+      "loss": 0.6804,
+      "step": 3623
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.6237720516381216e-05,
+      "loss": 0.8646,
+      "step": 3624
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.623514889677519e-05,
+      "loss": 0.8609,
+      "step": 3625
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.623257727716916e-05,
+      "loss": 0.9957,
+      "step": 3626
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.623000565756314e-05,
+      "loss": 0.7925,
+      "step": 3627
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.622743403795711e-05,
+      "loss": 0.7194,
+      "step": 3628
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.622486241835108e-05,
+      "loss": 0.8401,
+      "step": 3629
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.622229079874505e-05,
+      "loss": 0.6777,
+      "step": 3630
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.6219719179139024e-05,
+      "loss": 1.0048,
+      "step": 3631
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.6217147559532994e-05,
+      "loss": 0.721,
+      "step": 3632
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.621457593992697e-05,
+      "loss": 1.0098,
+      "step": 3633
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.621200432032094e-05,
+      "loss": 0.6817,
+      "step": 3634
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.620943270071491e-05,
+      "loss": 0.861,
+      "step": 3635
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.620686108110888e-05,
+      "loss": 0.8162,
+      "step": 3636
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.6204289461502856e-05,
+      "loss": 0.9541,
+      "step": 3637
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.620171784189683e-05,
+      "loss": 0.8869,
+      "step": 3638
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.61991462222908e-05,
+      "loss": 0.7848,
+      "step": 3639
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.619657460268477e-05,
+      "loss": 0.9521,
+      "step": 3640
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.619400298307874e-05,
+      "loss": 1.196,
+      "step": 3641
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.619143136347272e-05,
+      "loss": 0.955,
+      "step": 3642
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.6188859743866695e-05,
+      "loss": 0.6932,
+      "step": 3643
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.6186288124260664e-05,
+      "loss": 0.6982,
+      "step": 3644
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.6183716504654634e-05,
+      "loss": 0.5999,
+      "step": 3645
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.6181144885048604e-05,
+      "loss": 0.7284,
+      "step": 3646
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.617857326544257e-05,
+      "loss": 1.1305,
+      "step": 3647
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.617600164583655e-05,
+      "loss": 1.2482,
+      "step": 3648
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.6173430026230526e-05,
+      "loss": 1.0453,
+      "step": 3649
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.6170858406624496e-05,
+      "loss": 0.824,
+      "step": 3650
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.6168286787018466e-05,
+      "loss": 0.8607,
+      "step": 3651
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.6165715167412435e-05,
+      "loss": 0.9271,
+      "step": 3652
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.616314354780641e-05,
+      "loss": 0.6881,
+      "step": 3653
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.616057192820038e-05,
+      "loss": 0.7545,
+      "step": 3654
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.615800030859436e-05,
+      "loss": 0.7228,
+      "step": 3655
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.615542868898833e-05,
+      "loss": 0.4952,
+      "step": 3656
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.61528570693823e-05,
+      "loss": 1.048,
+      "step": 3657
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.6150285449776274e-05,
+      "loss": 0.9865,
+      "step": 3658
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.614771383017024e-05,
+      "loss": 0.7716,
+      "step": 3659
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.614514221056421e-05,
+      "loss": 0.7224,
+      "step": 3660
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.614257059095819e-05,
+      "loss": 0.8964,
+      "step": 3661
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.613999897135216e-05,
+      "loss": 0.699,
+      "step": 3662
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.613742735174613e-05,
+      "loss": 0.8696,
+      "step": 3663
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.6134855732140105e-05,
+      "loss": 0.9471,
+      "step": 3664
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.6132284112534075e-05,
+      "loss": 0.4052,
+      "step": 3665
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.6129712492928045e-05,
+      "loss": 0.4771,
+      "step": 3666
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.612714087332202e-05,
+      "loss": 0.7544,
+      "step": 3667
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.612456925371599e-05,
+      "loss": 0.5314,
+      "step": 3668
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.612199763410997e-05,
+      "loss": 1.3156,
+      "step": 3669
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.611942601450394e-05,
+      "loss": 0.9547,
+      "step": 3670
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.611685439489791e-05,
+      "loss": 0.9375,
+      "step": 3671
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.611428277529188e-05,
+      "loss": 0.8055,
+      "step": 3672
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.611171115568585e-05,
+      "loss": 0.6807,
+      "step": 3673
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.610913953607983e-05,
+      "loss": 0.9375,
+      "step": 3674
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.61065679164738e-05,
+      "loss": 1.053,
+      "step": 3675
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.610399629686777e-05,
+      "loss": 0.9947,
+      "step": 3676
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.610142467726174e-05,
+      "loss": 0.8209,
+      "step": 3677
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.6098853057655715e-05,
+      "loss": 0.7725,
+      "step": 3678
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.6096281438049684e-05,
+      "loss": 0.5658,
+      "step": 3679
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.609370981844366e-05,
+      "loss": 0.7815,
+      "step": 3680
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.609113819883763e-05,
+      "loss": 1.1849,
+      "step": 3681
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.60885665792316e-05,
+      "loss": 0.8112,
+      "step": 3682
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.608599495962557e-05,
+      "loss": 0.6297,
+      "step": 3683
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.6083423340019546e-05,
+      "loss": 0.8238,
+      "step": 3684
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.608085172041352e-05,
+      "loss": 1.2726,
+      "step": 3685
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.607828010080749e-05,
+      "loss": 0.7373,
+      "step": 3686
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.607570848120146e-05,
+      "loss": 0.8312,
+      "step": 3687
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.607313686159543e-05,
+      "loss": 0.7986,
+      "step": 3688
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.607056524198941e-05,
+      "loss": 0.7958,
+      "step": 3689
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.6067993622383385e-05,
+      "loss": 0.7736,
+      "step": 3690
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.6065422002777355e-05,
+      "loss": 0.9028,
+      "step": 3691
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.6062850383171324e-05,
+      "loss": 1.395,
+      "step": 3692
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.6060278763565294e-05,
+      "loss": 0.7335,
+      "step": 3693
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.6057707143959264e-05,
+      "loss": 0.7621,
+      "step": 3694
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.605513552435324e-05,
+      "loss": 1.0038,
+      "step": 3695
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.6052563904747217e-05,
+      "loss": 0.6658,
+      "step": 3696
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.6049992285141186e-05,
+      "loss": 1.0588,
+      "step": 3697
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.6047420665535156e-05,
+      "loss": 0.9238,
+      "step": 3698
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.6044849045929126e-05,
+      "loss": 0.7447,
+      "step": 3699
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.60422774263231e-05,
+      "loss": 0.5275,
+      "step": 3700
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.603970580671707e-05,
+      "loss": 0.8438,
+      "step": 3701
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.603713418711105e-05,
+      "loss": 0.8618,
+      "step": 3702
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.603456256750502e-05,
+      "loss": 0.8906,
+      "step": 3703
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.603199094789899e-05,
+      "loss": 0.8647,
+      "step": 3704
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.6029419328292964e-05,
+      "loss": 0.8005,
+      "step": 3705
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.6026847708686934e-05,
+      "loss": 0.7498,
+      "step": 3706
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.6024276089080903e-05,
+      "loss": 0.8349,
+      "step": 3707
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.602170446947488e-05,
+      "loss": 0.7072,
+      "step": 3708
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.601913284986885e-05,
+      "loss": 0.6605,
+      "step": 3709
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.601656123026282e-05,
+      "loss": 0.8579,
+      "step": 3710
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.6013989610656796e-05,
+      "loss": 0.7305,
+      "step": 3711
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.6011417991050765e-05,
+      "loss": 1.0454,
+      "step": 3712
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.6008846371444735e-05,
+      "loss": 0.8013,
+      "step": 3713
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.600627475183871e-05,
+      "loss": 0.8425,
+      "step": 3714
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.600370313223268e-05,
+      "loss": 0.6142,
+      "step": 3715
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.600113151262666e-05,
+      "loss": 0.9454,
+      "step": 3716
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.599855989302063e-05,
+      "loss": 1.17,
+      "step": 3717
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.59959882734146e-05,
+      "loss": 0.733,
+      "step": 3718
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.599341665380857e-05,
+      "loss": 0.7709,
+      "step": 3719
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.599084503420254e-05,
+      "loss": 0.9903,
+      "step": 3720
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.598827341459652e-05,
+      "loss": 0.6817,
+      "step": 3721
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.598570179499049e-05,
+      "loss": 0.7342,
+      "step": 3722
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.598313017538446e-05,
+      "loss": 0.9967,
+      "step": 3723
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.598055855577843e-05,
+      "loss": 1.0295,
+      "step": 3724
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.59779869361724e-05,
+      "loss": 0.5269,
+      "step": 3725
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.5975415316566375e-05,
+      "loss": 0.6508,
+      "step": 3726
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.597284369696035e-05,
+      "loss": 0.9557,
+      "step": 3727
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.597027207735432e-05,
+      "loss": 0.9134,
+      "step": 3728
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.596770045774829e-05,
+      "loss": 0.5641,
+      "step": 3729
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.596512883814226e-05,
+      "loss": 0.7823,
+      "step": 3730
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.596255721853624e-05,
+      "loss": 0.788,
+      "step": 3731
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.595998559893021e-05,
+      "loss": 0.6416,
+      "step": 3732
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.595741397932418e-05,
+      "loss": 0.6083,
+      "step": 3733
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.595484235971815e-05,
+      "loss": 1.0437,
+      "step": 3734
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.595227074011212e-05,
+      "loss": 0.9142,
+      "step": 3735
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.59496991205061e-05,
+      "loss": 0.7207,
+      "step": 3736
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.594712750090007e-05,
+      "loss": 0.7425,
+      "step": 3737
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.5944555881294045e-05,
+      "loss": 0.9533,
+      "step": 3738
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.5941984261688015e-05,
+      "loss": 1.113,
+      "step": 3739
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.5939412642081984e-05,
+      "loss": 1.2537,
+      "step": 3740
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.5936841022475954e-05,
+      "loss": 0.9055,
+      "step": 3741
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.593426940286993e-05,
+      "loss": 1.1412,
+      "step": 3742
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.59316977832639e-05,
+      "loss": 1.0544,
+      "step": 3743
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.592912616365788e-05,
+      "loss": 1.0521,
+      "step": 3744
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.5926554544051846e-05,
+      "loss": 0.759,
+      "step": 3745
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.5923982924445816e-05,
+      "loss": 1.1441,
+      "step": 3746
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.592141130483979e-05,
+      "loss": 0.91,
+      "step": 3747
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.591883968523376e-05,
+      "loss": 1.0668,
+      "step": 3748
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.591626806562773e-05,
+      "loss": 0.6613,
+      "step": 3749
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.591369644602171e-05,
+      "loss": 0.8226,
+      "step": 3750
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.591112482641568e-05,
+      "loss": 0.9528,
+      "step": 3751
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.5908553206809654e-05,
+      "loss": 1.008,
+      "step": 3752
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.5905981587203624e-05,
+      "loss": 1.1486,
+      "step": 3753
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.5903409967597594e-05,
+      "loss": 0.7059,
+      "step": 3754
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.5900838347991564e-05,
+      "loss": 0.6635,
+      "step": 3755
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.589826672838554e-05,
+      "loss": 1.0458,
+      "step": 3756
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.589569510877951e-05,
+      "loss": 0.607,
+      "step": 3757
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.5893123489173486e-05,
+      "loss": 0.3695,
+      "step": 3758
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.5890551869567456e-05,
+      "loss": 0.9626,
+      "step": 3759
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.5887980249961426e-05,
+      "loss": 0.8569,
+      "step": 3760
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.5885408630355395e-05,
+      "loss": 1.0143,
+      "step": 3761
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.588283701074937e-05,
+      "loss": 0.619,
+      "step": 3762
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.588026539114335e-05,
+      "loss": 0.8774,
+      "step": 3763
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.587769377153732e-05,
+      "loss": 0.6263,
+      "step": 3764
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.587512215193129e-05,
+      "loss": 0.835,
+      "step": 3765
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.587255053232526e-05,
+      "loss": 0.8145,
+      "step": 3766
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.5869978912719234e-05,
+      "loss": 0.7727,
+      "step": 3767
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.586740729311321e-05,
+      "loss": 0.9563,
+      "step": 3768
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.586483567350718e-05,
+      "loss": 0.6589,
+      "step": 3769
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.586226405390115e-05,
+      "loss": 0.6667,
+      "step": 3770
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.585969243429512e-05,
+      "loss": 1.4588,
+      "step": 3771
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.585712081468909e-05,
+      "loss": 0.8678,
+      "step": 3772
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.5854549195083065e-05,
+      "loss": 0.8824,
+      "step": 3773
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.585197757547704e-05,
+      "loss": 0.8867,
+      "step": 3774
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.584940595587101e-05,
+      "loss": 1.0288,
+      "step": 3775
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.584683433626498e-05,
+      "loss": 0.9,
+      "step": 3776
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.584426271665895e-05,
+      "loss": 0.8838,
+      "step": 3777
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.584169109705293e-05,
+      "loss": 0.8492,
+      "step": 3778
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.58391194774469e-05,
+      "loss": 0.6063,
+      "step": 3779
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.5836547857840873e-05,
+      "loss": 0.9933,
+      "step": 3780
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.583397623823484e-05,
+      "loss": 0.8435,
+      "step": 3781
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.583140461862881e-05,
+      "loss": 0.9306,
+      "step": 3782
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.582883299902279e-05,
+      "loss": 0.9152,
+      "step": 3783
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.582626137941676e-05,
+      "loss": 0.7921,
+      "step": 3784
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.582368975981073e-05,
+      "loss": 0.6793,
+      "step": 3785
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.5821118140204705e-05,
+      "loss": 0.8813,
+      "step": 3786
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.5818546520598675e-05,
+      "loss": 0.7681,
+      "step": 3787
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.5815974900992644e-05,
+      "loss": 0.716,
+      "step": 3788
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.581340328138662e-05,
+      "loss": 0.977,
+      "step": 3789
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.581083166178059e-05,
+      "loss": 0.9396,
+      "step": 3790
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.580826004217456e-05,
+      "loss": 0.7147,
+      "step": 3791
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.580568842256854e-05,
+      "loss": 0.5901,
+      "step": 3792
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.5803116802962506e-05,
+      "loss": 1.0556,
+      "step": 3793
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.580054518335648e-05,
+      "loss": 0.8023,
+      "step": 3794
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.579797356375045e-05,
+      "loss": 0.7961,
+      "step": 3795
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.579540194414442e-05,
+      "loss": 1.0289,
+      "step": 3796
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.579283032453839e-05,
+      "loss": 0.7638,
+      "step": 3797
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.579025870493237e-05,
+      "loss": 0.6895,
+      "step": 3798
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.5787687085326345e-05,
+      "loss": 0.8234,
+      "step": 3799
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.5785115465720315e-05,
+      "loss": 1.0708,
+      "step": 3800
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.5782543846114284e-05,
+      "loss": 0.9585,
+      "step": 3801
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.5779972226508254e-05,
+      "loss": 1.1587,
+      "step": 3802
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.5777400606902224e-05,
+      "loss": 1.1072,
+      "step": 3803
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.57748289872962e-05,
+      "loss": 0.8204,
+      "step": 3804
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.5772257367690177e-05,
+      "loss": 0.6508,
+      "step": 3805
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.5769685748084146e-05,
+      "loss": 0.5622,
+      "step": 3806
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.5767114128478116e-05,
+      "loss": 0.9529,
+      "step": 3807
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.5764542508872086e-05,
+      "loss": 1.2573,
+      "step": 3808
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.576197088926606e-05,
+      "loss": 0.7959,
+      "step": 3809
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.575939926966004e-05,
+      "loss": 0.9258,
+      "step": 3810
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.575682765005401e-05,
+      "loss": 1.0146,
+      "step": 3811
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.575425603044798e-05,
+      "loss": 0.9718,
+      "step": 3812
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.575168441084195e-05,
+      "loss": 0.8295,
+      "step": 3813
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.5749112791235924e-05,
+      "loss": 0.7428,
+      "step": 3814
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.5746541171629894e-05,
+      "loss": 0.5751,
+      "step": 3815
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.574396955202387e-05,
+      "loss": 0.7016,
+      "step": 3816
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.574139793241784e-05,
+      "loss": 1.063,
+      "step": 3817
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.573882631281181e-05,
+      "loss": 0.4692,
+      "step": 3818
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.573625469320578e-05,
+      "loss": 0.6934,
+      "step": 3819
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.5733683073599756e-05,
+      "loss": 0.9371,
+      "step": 3820
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.573111145399373e-05,
+      "loss": 0.7996,
+      "step": 3821
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.57285398343877e-05,
+      "loss": 0.8932,
+      "step": 3822
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.572596821478167e-05,
+      "loss": 0.8398,
+      "step": 3823
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.572339659517564e-05,
+      "loss": 1.1696,
+      "step": 3824
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.572082497556962e-05,
+      "loss": 0.8551,
+      "step": 3825
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.571825335596359e-05,
+      "loss": 0.4988,
+      "step": 3826
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.5715681736357564e-05,
+      "loss": 0.8285,
+      "step": 3827
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.5713110116751534e-05,
+      "loss": 0.456,
+      "step": 3828
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.57105384971455e-05,
+      "loss": 0.5475,
+      "step": 3829
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.570796687753948e-05,
+      "loss": 0.9207,
+      "step": 3830
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.570539525793345e-05,
+      "loss": 0.7302,
+      "step": 3831
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.570282363832742e-05,
+      "loss": 0.6914,
+      "step": 3832
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.5700252018721395e-05,
+      "loss": 0.6913,
+      "step": 3833
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.5697680399115365e-05,
+      "loss": 0.8433,
+      "step": 3834
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.5695108779509335e-05,
+      "loss": 0.8038,
+      "step": 3835
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.569253715990331e-05,
+      "loss": 0.7229,
+      "step": 3836
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.568996554029728e-05,
+      "loss": 0.742,
+      "step": 3837
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.568739392069125e-05,
+      "loss": 0.7877,
+      "step": 3838
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.568482230108523e-05,
+      "loss": 0.8797,
+      "step": 3839
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.56822506814792e-05,
+      "loss": 0.6262,
+      "step": 3840
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.567967906187317e-05,
+      "loss": 0.6577,
+      "step": 3841
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.567710744226714e-05,
+      "loss": 0.7586,
+      "step": 3842
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.567453582266111e-05,
+      "loss": 0.8551,
+      "step": 3843
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.567196420305508e-05,
+      "loss": 0.8934,
+      "step": 3844
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.566939258344906e-05,
+      "loss": 0.6817,
+      "step": 3845
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.5666820963843035e-05,
+      "loss": 1.0726,
+      "step": 3846
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.5664249344237005e-05,
+      "loss": 0.7379,
+      "step": 3847
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.5661677724630975e-05,
+      "loss": 0.7303,
+      "step": 3848
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.5659106105024944e-05,
+      "loss": 1.1777,
+      "step": 3849
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.5656534485418914e-05,
+      "loss": 0.7387,
+      "step": 3850
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.565396286581289e-05,
+      "loss": 1.0114,
+      "step": 3851
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.565139124620687e-05,
+      "loss": 0.7613,
+      "step": 3852
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.564881962660084e-05,
+      "loss": 0.9553,
+      "step": 3853
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.5646248006994806e-05,
+      "loss": 0.7072,
+      "step": 3854
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.5643676387388776e-05,
+      "loss": 1.157,
+      "step": 3855
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.564110476778275e-05,
+      "loss": 1.0897,
+      "step": 3856
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.563853314817673e-05,
+      "loss": 0.8818,
+      "step": 3857
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.56359615285707e-05,
+      "loss": 0.7782,
+      "step": 3858
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.563338990896467e-05,
+      "loss": 0.8961,
+      "step": 3859
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.563081828935864e-05,
+      "loss": 0.9255,
+      "step": 3860
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.5628246669752614e-05,
+      "loss": 0.6492,
+      "step": 3861
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.5625675050146584e-05,
+      "loss": 0.826,
+      "step": 3862
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.562310343054056e-05,
+      "loss": 0.8481,
+      "step": 3863
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.562053181093453e-05,
+      "loss": 1.1443,
+      "step": 3864
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.56179601913285e-05,
+      "loss": 0.6657,
+      "step": 3865
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.561538857172247e-05,
+      "loss": 0.5704,
+      "step": 3866
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.5612816952116446e-05,
+      "loss": 0.7871,
+      "step": 3867
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.5610245332510416e-05,
+      "loss": 0.7578,
+      "step": 3868
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.560767371290439e-05,
+      "loss": 1.1858,
+      "step": 3869
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.560510209329836e-05,
+      "loss": 0.9098,
+      "step": 3870
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.560253047369233e-05,
+      "loss": 0.8873,
+      "step": 3871
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.559995885408631e-05,
+      "loss": 1.1293,
+      "step": 3872
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.559738723448028e-05,
+      "loss": 0.7922,
+      "step": 3873
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.559481561487425e-05,
+      "loss": 1.2471,
+      "step": 3874
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.5592243995268224e-05,
+      "loss": 0.8365,
+      "step": 3875
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.5589672375662194e-05,
+      "loss": 0.9668,
+      "step": 3876
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.558710075605617e-05,
+      "loss": 0.8438,
+      "step": 3877
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.558452913645014e-05,
+      "loss": 0.8903,
+      "step": 3878
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.558195751684411e-05,
+      "loss": 0.842,
+      "step": 3879
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.557938589723808e-05,
+      "loss": 0.8056,
+      "step": 3880
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.5576814277632056e-05,
+      "loss": 0.9254,
+      "step": 3881
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.5574242658026025e-05,
+      "loss": 0.6326,
+      "step": 3882
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.557167103842e-05,
+      "loss": 0.8009,
+      "step": 3883
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.556909941881397e-05,
+      "loss": 0.9801,
+      "step": 3884
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.556652779920794e-05,
+      "loss": 0.8302,
+      "step": 3885
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.556395617960191e-05,
+      "loss": 0.8623,
+      "step": 3886
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.556138455999589e-05,
+      "loss": 0.6372,
+      "step": 3887
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.5558812940389864e-05,
+      "loss": 0.759,
+      "step": 3888
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.5556241320783833e-05,
+      "loss": 0.7602,
+      "step": 3889
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.55536697011778e-05,
+      "loss": 0.9418,
+      "step": 3890
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.555109808157177e-05,
+      "loss": 0.6497,
+      "step": 3891
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.554852646196575e-05,
+      "loss": 0.6332,
+      "step": 3892
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.5545954842359726e-05,
+      "loss": 0.6577,
+      "step": 3893
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.5543383222753695e-05,
+      "loss": 1.052,
+      "step": 3894
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.5540811603147665e-05,
+      "loss": 0.8205,
+      "step": 3895
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.5538239983541635e-05,
+      "loss": 1.027,
+      "step": 3896
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.5535668363935604e-05,
+      "loss": 1.0717,
+      "step": 3897
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.553309674432958e-05,
+      "loss": 0.8149,
+      "step": 3898
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.553052512472356e-05,
+      "loss": 0.9243,
+      "step": 3899
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.552795350511753e-05,
+      "loss": 0.6821,
+      "step": 3900
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.55253818855115e-05,
+      "loss": 0.6907,
+      "step": 3901
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.5522810265905466e-05,
+      "loss": 0.9128,
+      "step": 3902
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.552023864629944e-05,
+      "loss": 1.0659,
+      "step": 3903
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.551766702669341e-05,
+      "loss": 0.9107,
+      "step": 3904
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.551509540708739e-05,
+      "loss": 0.8212,
+      "step": 3905
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.551252378748136e-05,
+      "loss": 1.0106,
+      "step": 3906
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.550995216787533e-05,
+      "loss": 0.8608,
+      "step": 3907
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.5507380548269305e-05,
+      "loss": 0.8391,
+      "step": 3908
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.5504808928663275e-05,
+      "loss": 0.9107,
+      "step": 3909
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.5502237309057244e-05,
+      "loss": 1.1441,
+      "step": 3910
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.549966568945122e-05,
+      "loss": 0.7068,
+      "step": 3911
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.549709406984519e-05,
+      "loss": 0.9274,
+      "step": 3912
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.549452245023916e-05,
+      "loss": 0.7191,
+      "step": 3913
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.5491950830633137e-05,
+      "loss": 1.1128,
+      "step": 3914
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.5489379211027106e-05,
+      "loss": 0.7972,
+      "step": 3915
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.5486807591421076e-05,
+      "loss": 0.9641,
+      "step": 3916
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.548423597181505e-05,
+      "loss": 0.7073,
+      "step": 3917
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.548166435220902e-05,
+      "loss": 1.1318,
+      "step": 3918
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.5479092732603e-05,
+      "loss": 0.9798,
+      "step": 3919
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.547652111299697e-05,
+      "loss": 0.9873,
+      "step": 3920
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.547394949339094e-05,
+      "loss": 0.7826,
+      "step": 3921
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.547137787378491e-05,
+      "loss": 0.7332,
+      "step": 3922
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.5468806254178884e-05,
+      "loss": 0.8887,
+      "step": 3923
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.546623463457286e-05,
+      "loss": 0.8226,
+      "step": 3924
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.546366301496683e-05,
+      "loss": 0.9496,
+      "step": 3925
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.54610913953608e-05,
+      "loss": 0.7778,
+      "step": 3926
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.545851977575477e-05,
+      "loss": 0.8701,
+      "step": 3927
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.545594815614874e-05,
+      "loss": 0.8849,
+      "step": 3928
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.5453376536542716e-05,
+      "loss": 0.765,
+      "step": 3929
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.545080491693669e-05,
+      "loss": 0.7268,
+      "step": 3930
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.544823329733066e-05,
+      "loss": 0.6251,
+      "step": 3931
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.544566167772463e-05,
+      "loss": 0.7481,
+      "step": 3932
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.54430900581186e-05,
+      "loss": 1.0508,
+      "step": 3933
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.544051843851258e-05,
+      "loss": 1.0679,
+      "step": 3934
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.5437946818906554e-05,
+      "loss": 1.2478,
+      "step": 3935
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.5435375199300524e-05,
+      "loss": 0.8278,
+      "step": 3936
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.5432803579694493e-05,
+      "loss": 0.8275,
+      "step": 3937
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.543023196008846e-05,
+      "loss": 0.8183,
+      "step": 3938
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.542766034048244e-05,
+      "loss": 1.0101,
+      "step": 3939
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.542508872087641e-05,
+      "loss": 1.036,
+      "step": 3940
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.5422517101270386e-05,
+      "loss": 0.6705,
+      "step": 3941
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.5419945481664355e-05,
+      "loss": 1.0787,
+      "step": 3942
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.5417373862058325e-05,
+      "loss": 0.5713,
+      "step": 3943
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.5414802242452295e-05,
+      "loss": 0.5328,
+      "step": 3944
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.541223062284627e-05,
+      "loss": 0.6527,
+      "step": 3945
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.540965900324024e-05,
+      "loss": 0.6659,
+      "step": 3946
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.540708738363422e-05,
+      "loss": 0.6546,
+      "step": 3947
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.540451576402819e-05,
+      "loss": 0.6375,
+      "step": 3948
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.540194414442216e-05,
+      "loss": 0.9309,
+      "step": 3949
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.539937252481613e-05,
+      "loss": 0.8154,
+      "step": 3950
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.53968009052101e-05,
+      "loss": 0.8887,
+      "step": 3951
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.539422928560407e-05,
+      "loss": 0.6708,
+      "step": 3952
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.539165766599805e-05,
+      "loss": 0.707,
+      "step": 3953
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.538908604639202e-05,
+      "loss": 0.915,
+      "step": 3954
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.5386514426785995e-05,
+      "loss": 0.69,
+      "step": 3955
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.5383942807179965e-05,
+      "loss": 0.8518,
+      "step": 3956
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.5381371187573935e-05,
+      "loss": 1.1051,
+      "step": 3957
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.5378799567967904e-05,
+      "loss": 1.1731,
+      "step": 3958
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.537622794836188e-05,
+      "loss": 0.9184,
+      "step": 3959
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.537365632875585e-05,
+      "loss": 0.7811,
+      "step": 3960
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.537108470914983e-05,
+      "loss": 0.5667,
+      "step": 3961
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.5368513089543797e-05,
+      "loss": 0.4698,
+      "step": 3962
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.5365941469937766e-05,
+      "loss": 0.7103,
+      "step": 3963
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.5363369850331736e-05,
+      "loss": 1.0655,
+      "step": 3964
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.536079823072571e-05,
+      "loss": 0.9092,
+      "step": 3965
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.535822661111969e-05,
+      "loss": 1.1432,
+      "step": 3966
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.535565499151366e-05,
+      "loss": 0.8861,
+      "step": 3967
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.535308337190763e-05,
+      "loss": 0.9927,
+      "step": 3968
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.53505117523016e-05,
+      "loss": 0.8043,
+      "step": 3969
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.5347940132695574e-05,
+      "loss": 0.7792,
+      "step": 3970
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.534536851308955e-05,
+      "loss": 1.0718,
+      "step": 3971
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.534279689348352e-05,
+      "loss": 0.8532,
+      "step": 3972
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.534022527387749e-05,
+      "loss": 0.7347,
+      "step": 3973
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.533765365427146e-05,
+      "loss": 0.9532,
+      "step": 3974
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.533508203466543e-05,
+      "loss": 0.6584,
+      "step": 3975
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.5332510415059406e-05,
+      "loss": 0.9184,
+      "step": 3976
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.532993879545338e-05,
+      "loss": 0.76,
+      "step": 3977
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.532736717584735e-05,
+      "loss": 0.9632,
+      "step": 3978
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.532479555624132e-05,
+      "loss": 0.7768,
+      "step": 3979
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.532222393663529e-05,
+      "loss": 0.7752,
+      "step": 3980
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.531965231702927e-05,
+      "loss": 0.751,
+      "step": 3981
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.5317080697423245e-05,
+      "loss": 0.7235,
+      "step": 3982
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.5314509077817214e-05,
+      "loss": 0.7787,
+      "step": 3983
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.5311937458211184e-05,
+      "loss": 0.7589,
+      "step": 3984
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.5309365838605154e-05,
+      "loss": 0.9917,
+      "step": 3985
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.530679421899913e-05,
+      "loss": 0.7677,
+      "step": 3986
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.53042225993931e-05,
+      "loss": 0.9847,
+      "step": 3987
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.5301650979787076e-05,
+      "loss": 1.0682,
+      "step": 3988
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.5299079360181046e-05,
+      "loss": 0.9113,
+      "step": 3989
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.5296507740575016e-05,
+      "loss": 1.042,
+      "step": 3990
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.5293936120968985e-05,
+      "loss": 0.9627,
+      "step": 3991
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.529136450136296e-05,
+      "loss": 1.0351,
+      "step": 3992
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.528879288175693e-05,
+      "loss": 0.7962,
+      "step": 3993
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.528622126215091e-05,
+      "loss": 0.8764,
+      "step": 3994
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.528364964254488e-05,
+      "loss": 0.8236,
+      "step": 3995
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.528107802293885e-05,
+      "loss": 0.8473,
+      "step": 3996
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.5278506403332824e-05,
+      "loss": 1.0922,
+      "step": 3997
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.527593478372679e-05,
+      "loss": 0.9742,
+      "step": 3998
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.527336316412076e-05,
+      "loss": 0.8452,
+      "step": 3999
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.527079154451474e-05,
+      "loss": 0.9453,
+      "step": 4000
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.526821992490871e-05,
+      "loss": 0.8851,
+      "step": 4001
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.5265648305302686e-05,
+      "loss": 0.941,
+      "step": 4002
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.5263076685696655e-05,
+      "loss": 1.0451,
+      "step": 4003
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.5260505066090625e-05,
+      "loss": 0.9787,
+      "step": 4004
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.5257933446484595e-05,
+      "loss": 0.5285,
+      "step": 4005
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.525536182687857e-05,
+      "loss": 0.7292,
+      "step": 4006
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.525279020727254e-05,
+      "loss": 1.2662,
+      "step": 4007
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.525021858766652e-05,
+      "loss": 0.8097,
+      "step": 4008
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.524764696806049e-05,
+      "loss": 0.8518,
+      "step": 4009
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.524507534845446e-05,
+      "loss": 0.7426,
+      "step": 4010
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.5242503728848426e-05,
+      "loss": 0.5659,
+      "step": 4011
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.52399321092424e-05,
+      "loss": 0.7248,
+      "step": 4012
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.523736048963638e-05,
+      "loss": 0.7549,
+      "step": 4013
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.523478887003035e-05,
+      "loss": 0.6546,
+      "step": 4014
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.523221725042432e-05,
+      "loss": 0.7385,
+      "step": 4015
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.522964563081829e-05,
+      "loss": 1.0631,
+      "step": 4016
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.5227074011212265e-05,
+      "loss": 0.7175,
+      "step": 4017
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.522450239160624e-05,
+      "loss": 0.8014,
+      "step": 4018
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.522193077200021e-05,
+      "loss": 0.9978,
+      "step": 4019
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.521935915239418e-05,
+      "loss": 0.9855,
+      "step": 4020
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.521678753278815e-05,
+      "loss": 0.5221,
+      "step": 4021
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.521421591318212e-05,
+      "loss": 0.8453,
+      "step": 4022
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.5211644293576096e-05,
+      "loss": 0.4596,
+      "step": 4023
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.520907267397007e-05,
+      "loss": 0.6566,
+      "step": 4024
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.520650105436404e-05,
+      "loss": 1.0665,
+      "step": 4025
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.520392943475801e-05,
+      "loss": 1.0195,
+      "step": 4026
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.520135781515198e-05,
+      "loss": 0.6229,
+      "step": 4027
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.519878619554596e-05,
+      "loss": 0.763,
+      "step": 4028
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.519621457593993e-05,
+      "loss": 1.026,
+      "step": 4029
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.5193642956333905e-05,
+      "loss": 0.998,
+      "step": 4030
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.5191071336727874e-05,
+      "loss": 0.9944,
+      "step": 4031
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.5188499717121844e-05,
+      "loss": 0.7449,
+      "step": 4032
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.518592809751582e-05,
+      "loss": 0.7384,
+      "step": 4033
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.518335647790979e-05,
+      "loss": 0.7235,
+      "step": 4034
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.518078485830376e-05,
+      "loss": 0.6566,
+      "step": 4035
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.5178213238697736e-05,
+      "loss": 0.9537,
+      "step": 4036
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.5175641619091706e-05,
+      "loss": 0.6743,
+      "step": 4037
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.5173069999485676e-05,
+      "loss": 0.7409,
+      "step": 4038
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.517049837987965e-05,
+      "loss": 0.7304,
+      "step": 4039
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.516792676027362e-05,
+      "loss": 0.7441,
+      "step": 4040
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.516535514066759e-05,
+      "loss": 1.0891,
+      "step": 4041
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.516278352106157e-05,
+      "loss": 1.1734,
+      "step": 4042
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.516021190145554e-05,
+      "loss": 0.8032,
+      "step": 4043
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.5157640281849514e-05,
+      "loss": 0.9112,
+      "step": 4044
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.5155068662243484e-05,
+      "loss": 1.0063,
+      "step": 4045
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.5152497042637453e-05,
+      "loss": 0.7808,
+      "step": 4046
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.514992542303142e-05,
+      "loss": 0.7952,
+      "step": 4047
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.51473538034254e-05,
+      "loss": 0.7479,
+      "step": 4048
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.5144782183819376e-05,
+      "loss": 0.7718,
+      "step": 4049
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.5142210564213346e-05,
+      "loss": 0.9758,
+      "step": 4050
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.5139638944607315e-05,
+      "loss": 0.5379,
+      "step": 4051
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.5137067325001285e-05,
+      "loss": 0.9548,
+      "step": 4052
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.5134495705395255e-05,
+      "loss": 0.9964,
+      "step": 4053
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.513192408578923e-05,
+      "loss": 1.0271,
+      "step": 4054
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.512935246618321e-05,
+      "loss": 1.0485,
+      "step": 4055
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.512678084657718e-05,
+      "loss": 0.7989,
+      "step": 4056
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.512420922697115e-05,
+      "loss": 1.0416,
+      "step": 4057
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.512163760736512e-05,
+      "loss": 0.9441,
+      "step": 4058
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.511906598775909e-05,
+      "loss": 0.8168,
+      "step": 4059
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.511649436815307e-05,
+      "loss": 0.9516,
+      "step": 4060
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.511392274854704e-05,
+      "loss": 1.3104,
+      "step": 4061
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.511135112894101e-05,
+      "loss": 0.8603,
+      "step": 4062
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.510877950933498e-05,
+      "loss": 1.0217,
+      "step": 4063
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.5106207889728955e-05,
+      "loss": 0.7504,
+      "step": 4064
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.5103636270122925e-05,
+      "loss": 0.9513,
+      "step": 4065
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.51010646505169e-05,
+      "loss": 0.8452,
+      "step": 4066
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.509849303091087e-05,
+      "loss": 0.8139,
+      "step": 4067
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.509592141130484e-05,
+      "loss": 0.885,
+      "step": 4068
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.509334979169881e-05,
+      "loss": 0.8774,
+      "step": 4069
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.509077817209279e-05,
+      "loss": 1.042,
+      "step": 4070
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.5088206552486757e-05,
+      "loss": 0.8687,
+      "step": 4071
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.508563493288073e-05,
+      "loss": 0.443,
+      "step": 4072
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.50830633132747e-05,
+      "loss": 0.6935,
+      "step": 4073
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.508049169366867e-05,
+      "loss": 0.939,
+      "step": 4074
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.507792007406265e-05,
+      "loss": 1.2459,
+      "step": 4075
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.507534845445662e-05,
+      "loss": 0.8636,
+      "step": 4076
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.507277683485059e-05,
+      "loss": 0.8632,
+      "step": 4077
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.5070205215244565e-05,
+      "loss": 0.895,
+      "step": 4078
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.5067633595638534e-05,
+      "loss": 0.8373,
+      "step": 4079
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.506506197603251e-05,
+      "loss": 1.0182,
+      "step": 4080
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.506249035642648e-05,
+      "loss": 0.7965,
+      "step": 4081
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.505991873682045e-05,
+      "loss": 0.9825,
+      "step": 4082
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.505734711721442e-05,
+      "loss": 0.7529,
+      "step": 4083
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.5054775497608396e-05,
+      "loss": 0.7787,
+      "step": 4084
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.5052203878002366e-05,
+      "loss": 0.9384,
+      "step": 4085
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.504963225839634e-05,
+      "loss": 0.7291,
+      "step": 4086
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.504706063879031e-05,
+      "loss": 0.6516,
+      "step": 4087
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.504448901918428e-05,
+      "loss": 1.0408,
+      "step": 4088
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.504191739957825e-05,
+      "loss": 1.0099,
+      "step": 4089
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.503934577997223e-05,
+      "loss": 0.7782,
+      "step": 4090
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.5036774160366204e-05,
+      "loss": 0.7879,
+      "step": 4091
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.5034202540760174e-05,
+      "loss": 0.5109,
+      "step": 4092
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.5031630921154144e-05,
+      "loss": 1.0478,
+      "step": 4093
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.5029059301548114e-05,
+      "loss": 0.9951,
+      "step": 4094
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.502648768194209e-05,
+      "loss": 1.0933,
+      "step": 4095
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.5023916062336066e-05,
+      "loss": 0.7437,
+      "step": 4096
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.5021344442730036e-05,
+      "loss": 0.8076,
+      "step": 4097
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.5018772823124006e-05,
+      "loss": 0.8622,
+      "step": 4098
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.5016201203517976e-05,
+      "loss": 0.8442,
+      "step": 4099
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.5013629583911945e-05,
+      "loss": 0.7256,
+      "step": 4100
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.501105796430592e-05,
+      "loss": 0.9763,
+      "step": 4101
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.50084863446999e-05,
+      "loss": 0.6139,
+      "step": 4102
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.500591472509387e-05,
+      "loss": 0.9067,
+      "step": 4103
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.500334310548784e-05,
+      "loss": 0.7065,
+      "step": 4104
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.500077148588181e-05,
+      "loss": 1.1317,
+      "step": 4105
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.4998199866275784e-05,
+      "loss": 0.7995,
+      "step": 4106
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.499562824666975e-05,
+      "loss": 0.7236,
+      "step": 4107
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.499305662706373e-05,
+      "loss": 0.9881,
+      "step": 4108
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.49904850074577e-05,
+      "loss": 0.8235,
+      "step": 4109
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.498791338785167e-05,
+      "loss": 0.7196,
+      "step": 4110
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.4985341768245646e-05,
+      "loss": 0.8051,
+      "step": 4111
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.4982770148639615e-05,
+      "loss": 1.1021,
+      "step": 4112
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.4980198529033585e-05,
+      "loss": 1.0115,
+      "step": 4113
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.497762690942756e-05,
+      "loss": 0.9202,
+      "step": 4114
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.497505528982153e-05,
+      "loss": 0.9523,
+      "step": 4115
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.49724836702155e-05,
+      "loss": 0.8019,
+      "step": 4116
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.496991205060948e-05,
+      "loss": 0.989,
+      "step": 4117
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.496734043100345e-05,
+      "loss": 0.7534,
+      "step": 4118
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.496476881139742e-05,
+      "loss": 0.9235,
+      "step": 4119
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.496219719179139e-05,
+      "loss": 0.7633,
+      "step": 4120
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.495962557218536e-05,
+      "loss": 0.6846,
+      "step": 4121
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.495705395257934e-05,
+      "loss": 0.808,
+      "step": 4122
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.495448233297331e-05,
+      "loss": 0.8481,
+      "step": 4123
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.495191071336728e-05,
+      "loss": 0.5886,
+      "step": 4124
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.4949339093761255e-05,
+      "loss": 1.0877,
+      "step": 4125
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.4946767474155225e-05,
+      "loss": 1.1296,
+      "step": 4126
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.49441958545492e-05,
+      "loss": 0.8501,
+      "step": 4127
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.494162423494317e-05,
+      "loss": 0.8643,
+      "step": 4128
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.493905261533714e-05,
+      "loss": 0.9164,
+      "step": 4129
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.493648099573111e-05,
+      "loss": 1.1834,
+      "step": 4130
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.493390937612509e-05,
+      "loss": 0.8359,
+      "step": 4131
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.4931337756519056e-05,
+      "loss": 0.8675,
+      "step": 4132
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.492876613691303e-05,
+      "loss": 0.8299,
+      "step": 4133
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.4926194517307e-05,
+      "loss": 0.7583,
+      "step": 4134
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.492362289770097e-05,
+      "loss": 0.9124,
+      "step": 4135
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.492105127809494e-05,
+      "loss": 0.8696,
+      "step": 4136
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.491847965848892e-05,
+      "loss": 0.6796,
+      "step": 4137
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.4915908038882895e-05,
+      "loss": 0.9545,
+      "step": 4138
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.4913336419276865e-05,
+      "loss": 0.9116,
+      "step": 4139
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.4910764799670834e-05,
+      "loss": 0.7792,
+      "step": 4140
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.4908193180064804e-05,
+      "loss": 0.7749,
+      "step": 4141
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.490562156045878e-05,
+      "loss": 1.2985,
+      "step": 4142
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.490304994085276e-05,
+      "loss": 0.6416,
+      "step": 4143
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.4900478321246727e-05,
+      "loss": 0.9274,
+      "step": 4144
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.4897906701640696e-05,
+      "loss": 0.9206,
+      "step": 4145
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.4895335082034666e-05,
+      "loss": 0.8891,
+      "step": 4146
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.4892763462428636e-05,
+      "loss": 0.8675,
+      "step": 4147
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.489019184282261e-05,
+      "loss": 0.7515,
+      "step": 4148
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.488762022321659e-05,
+      "loss": 0.8382,
+      "step": 4149
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.488504860361056e-05,
+      "loss": 0.746,
+      "step": 4150
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.488247698400453e-05,
+      "loss": 0.9675,
+      "step": 4151
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.48799053643985e-05,
+      "loss": 1.2249,
+      "step": 4152
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.4877333744792474e-05,
+      "loss": 0.7829,
+      "step": 4153
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.4874762125186444e-05,
+      "loss": 0.9365,
+      "step": 4154
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.487219050558042e-05,
+      "loss": 0.5938,
+      "step": 4155
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.486961888597439e-05,
+      "loss": 0.7298,
+      "step": 4156
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.486704726636836e-05,
+      "loss": 0.7259,
+      "step": 4157
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.4864475646762336e-05,
+      "loss": 0.626,
+      "step": 4158
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.4861904027156306e-05,
+      "loss": 1.0672,
+      "step": 4159
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.4859332407550275e-05,
+      "loss": 1.083,
+      "step": 4160
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.485676078794425e-05,
+      "loss": 0.8712,
+      "step": 4161
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.485418916833822e-05,
+      "loss": 0.8808,
+      "step": 4162
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.485161754873219e-05,
+      "loss": 0.9751,
+      "step": 4163
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.484904592912617e-05,
+      "loss": 0.8166,
+      "step": 4164
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.484647430952014e-05,
+      "loss": 0.5924,
+      "step": 4165
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.484390268991411e-05,
+      "loss": 0.7935,
+      "step": 4166
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.4841331070308084e-05,
+      "loss": 0.6882,
+      "step": 4167
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.483875945070205e-05,
+      "loss": 0.9258,
+      "step": 4168
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.483618783109603e-05,
+      "loss": 0.9004,
+      "step": 4169
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.483361621149e-05,
+      "loss": 0.7119,
+      "step": 4170
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.483104459188397e-05,
+      "loss": 0.7595,
+      "step": 4171
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.482847297227794e-05,
+      "loss": 0.8257,
+      "step": 4172
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.4825901352671915e-05,
+      "loss": 0.4943,
+      "step": 4173
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.482332973306589e-05,
+      "loss": 0.5402,
+      "step": 4174
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.482075811345986e-05,
+      "loss": 1.0472,
+      "step": 4175
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.481818649385383e-05,
+      "loss": 0.7153,
+      "step": 4176
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.48156148742478e-05,
+      "loss": 0.9683,
+      "step": 4177
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.481304325464177e-05,
+      "loss": 0.8772,
+      "step": 4178
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.481047163503575e-05,
+      "loss": 0.7546,
+      "step": 4179
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.480790001542972e-05,
+      "loss": 0.6201,
+      "step": 4180
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.480532839582369e-05,
+      "loss": 1.0132,
+      "step": 4181
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.480275677621766e-05,
+      "loss": 0.9421,
+      "step": 4182
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.480018515661163e-05,
+      "loss": 0.7743,
+      "step": 4183
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.479761353700561e-05,
+      "loss": 0.9922,
+      "step": 4184
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.4795041917399585e-05,
+      "loss": 0.8414,
+      "step": 4185
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.4792470297793555e-05,
+      "loss": 0.94,
+      "step": 4186
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.4789898678187525e-05,
+      "loss": 0.798,
+      "step": 4187
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.4787327058581494e-05,
+      "loss": 0.9133,
+      "step": 4188
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.478475543897547e-05,
+      "loss": 0.9401,
+      "step": 4189
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.478218381936944e-05,
+      "loss": 0.9225,
+      "step": 4190
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.477961219976342e-05,
+      "loss": 1.1773,
+      "step": 4191
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.477704058015739e-05,
+      "loss": 0.6911,
+      "step": 4192
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.4774468960551356e-05,
+      "loss": 0.9322,
+      "step": 4193
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.4771897340945326e-05,
+      "loss": 0.977,
+      "step": 4194
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.47693257213393e-05,
+      "loss": 0.7889,
+      "step": 4195
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.476675410173327e-05,
+      "loss": 0.9538,
+      "step": 4196
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.476418248212725e-05,
+      "loss": 0.8697,
+      "step": 4197
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.476161086252122e-05,
+      "loss": 0.6584,
+      "step": 4198
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.475903924291519e-05,
+      "loss": 0.7689,
+      "step": 4199
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.4756467623309164e-05,
+      "loss": 0.8307,
+      "step": 4200
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.4753896003703134e-05,
+      "loss": 0.9131,
+      "step": 4201
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.4751324384097104e-05,
+      "loss": 0.9884,
+      "step": 4202
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.474875276449108e-05,
+      "loss": 0.8716,
+      "step": 4203
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.474618114488505e-05,
+      "loss": 1.171,
+      "step": 4204
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.4743609525279026e-05,
+      "loss": 0.6533,
+      "step": 4205
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.4741037905672996e-05,
+      "loss": 0.8113,
+      "step": 4206
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.4738466286066966e-05,
+      "loss": 0.8686,
+      "step": 4207
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.4735894666460935e-05,
+      "loss": 0.9524,
+      "step": 4208
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.473332304685491e-05,
+      "loss": 0.7137,
+      "step": 4209
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.473075142724888e-05,
+      "loss": 0.624,
+      "step": 4210
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.472817980764286e-05,
+      "loss": 0.7543,
+      "step": 4211
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.472560818803683e-05,
+      "loss": 0.845,
+      "step": 4212
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.47230365684308e-05,
+      "loss": 0.6285,
+      "step": 4213
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.472046494882477e-05,
+      "loss": 0.7323,
+      "step": 4214
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.4717893329218744e-05,
+      "loss": 0.6605,
+      "step": 4215
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.471532170961272e-05,
+      "loss": 1.0203,
+      "step": 4216
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.471275009000669e-05,
+      "loss": 0.8198,
+      "step": 4217
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.471017847040066e-05,
+      "loss": 0.802,
+      "step": 4218
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.470760685079463e-05,
+      "loss": 0.8902,
+      "step": 4219
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.4705035231188606e-05,
+      "loss": 0.6942,
+      "step": 4220
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.470246361158258e-05,
+      "loss": 0.7137,
+      "step": 4221
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.469989199197655e-05,
+      "loss": 0.715,
+      "step": 4222
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.469732037237052e-05,
+      "loss": 0.6716,
+      "step": 4223
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.469474875276449e-05,
+      "loss": 0.7687,
+      "step": 4224
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.469217713315846e-05,
+      "loss": 0.7374,
+      "step": 4225
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.468960551355244e-05,
+      "loss": 0.595,
+      "step": 4226
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.4687033893946414e-05,
+      "loss": 0.7034,
+      "step": 4227
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.4684462274340383e-05,
+      "loss": 0.5754,
+      "step": 4228
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.468189065473435e-05,
+      "loss": 1.1672,
+      "step": 4229
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.467931903512832e-05,
+      "loss": 0.9626,
+      "step": 4230
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.46767474155223e-05,
+      "loss": 1.344,
+      "step": 4231
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.467417579591627e-05,
+      "loss": 1.1563,
+      "step": 4232
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.4671604176310245e-05,
+      "loss": 0.9082,
+      "step": 4233
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.4669032556704215e-05,
+      "loss": 0.9697,
+      "step": 4234
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.4666460937098185e-05,
+      "loss": 0.8623,
+      "step": 4235
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.466388931749216e-05,
+      "loss": 0.7226,
+      "step": 4236
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.466131769788613e-05,
+      "loss": 0.7991,
+      "step": 4237
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.46587460782801e-05,
+      "loss": 0.8659,
+      "step": 4238
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.465617445867408e-05,
+      "loss": 0.5407,
+      "step": 4239
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.465360283906805e-05,
+      "loss": 0.6102,
+      "step": 4240
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.4651031219462016e-05,
+      "loss": 0.7865,
+      "step": 4241
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.464845959985599e-05,
+      "loss": 0.8217,
+      "step": 4242
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.464588798024996e-05,
+      "loss": 0.7543,
+      "step": 4243
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.464331636064393e-05,
+      "loss": 0.7844,
+      "step": 4244
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.464074474103791e-05,
+      "loss": 0.6756,
+      "step": 4245
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.463817312143188e-05,
+      "loss": 0.8263,
+      "step": 4246
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.4635601501825855e-05,
+      "loss": 0.8992,
+      "step": 4247
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.4633029882219825e-05,
+      "loss": 0.991,
+      "step": 4248
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.4630458262613794e-05,
+      "loss": 1.0906,
+      "step": 4249
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.4627886643007764e-05,
+      "loss": 0.6344,
+      "step": 4250
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.462531502340174e-05,
+      "loss": 0.8263,
+      "step": 4251
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.462274340379572e-05,
+      "loss": 0.8721,
+      "step": 4252
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.4620171784189687e-05,
+      "loss": 1.0222,
+      "step": 4253
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.4617600164583656e-05,
+      "loss": 0.7149,
+      "step": 4254
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.4615028544977626e-05,
+      "loss": 0.7918,
+      "step": 4255
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.4612456925371596e-05,
+      "loss": 0.8376,
+      "step": 4256
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.460988530576557e-05,
+      "loss": 0.8468,
+      "step": 4257
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.460731368615955e-05,
+      "loss": 1.0639,
+      "step": 4258
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.460474206655352e-05,
+      "loss": 0.6665,
+      "step": 4259
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.460217044694749e-05,
+      "loss": 0.9932,
+      "step": 4260
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.459959882734146e-05,
+      "loss": 0.9824,
+      "step": 4261
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.4597027207735434e-05,
+      "loss": 0.6276,
+      "step": 4262
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.459445558812941e-05,
+      "loss": 1.2211,
+      "step": 4263
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.459188396852338e-05,
+      "loss": 0.7641,
+      "step": 4264
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.458931234891735e-05,
+      "loss": 0.8541,
+      "step": 4265
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.458674072931132e-05,
+      "loss": 0.7507,
+      "step": 4266
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.4584169109705296e-05,
+      "loss": 0.8979,
+      "step": 4267
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.4581597490099266e-05,
+      "loss": 0.8764,
+      "step": 4268
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.457902587049324e-05,
+      "loss": 0.6695,
+      "step": 4269
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.457645425088721e-05,
+      "loss": 1.0222,
+      "step": 4270
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.457388263128118e-05,
+      "loss": 0.7306,
+      "step": 4271
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.457131101167515e-05,
+      "loss": 1.0283,
+      "step": 4272
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.456873939206913e-05,
+      "loss": 0.8276,
+      "step": 4273
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.4566167772463104e-05,
+      "loss": 0.8156,
+      "step": 4274
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.4563596152857074e-05,
+      "loss": 0.7029,
+      "step": 4275
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.4561024533251043e-05,
+      "loss": 0.9656,
+      "step": 4276
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.455845291364501e-05,
+      "loss": 0.6981,
+      "step": 4277
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.455588129403899e-05,
+      "loss": 0.9625,
+      "step": 4278
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.455330967443296e-05,
+      "loss": 0.8716,
+      "step": 4279
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.4550738054826936e-05,
+      "loss": 0.6671,
+      "step": 4280
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.4548166435220905e-05,
+      "loss": 0.8928,
+      "step": 4281
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.4545594815614875e-05,
+      "loss": 1.1028,
+      "step": 4282
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.454302319600885e-05,
+      "loss": 0.9675,
+      "step": 4283
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.454045157640282e-05,
+      "loss": 0.8718,
+      "step": 4284
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.453787995679679e-05,
+      "loss": 0.9096,
+      "step": 4285
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.453530833719077e-05,
+      "loss": 1.1787,
+      "step": 4286
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.453273671758474e-05,
+      "loss": 1.0195,
+      "step": 4287
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.453016509797871e-05,
+      "loss": 0.8795,
+      "step": 4288
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.452759347837268e-05,
+      "loss": 0.632,
+      "step": 4289
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.452502185876665e-05,
+      "loss": 0.9482,
+      "step": 4290
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.452245023916062e-05,
+      "loss": 1.0263,
+      "step": 4291
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.45198786195546e-05,
+      "loss": 0.6585,
+      "step": 4292
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.451730699994857e-05,
+      "loss": 0.6963,
+      "step": 4293
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.4514735380342545e-05,
+      "loss": 1.011,
+      "step": 4294
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.4512163760736515e-05,
+      "loss": 0.8475,
+      "step": 4295
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.4509592141130485e-05,
+      "loss": 1.0111,
+      "step": 4296
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.4507020521524454e-05,
+      "loss": 0.573,
+      "step": 4297
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.450444890191843e-05,
+      "loss": 0.892,
+      "step": 4298
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.450187728231241e-05,
+      "loss": 0.5617,
+      "step": 4299
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.449930566270638e-05,
+      "loss": 0.8333,
+      "step": 4300
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.449673404310035e-05,
+      "loss": 0.835,
+      "step": 4301
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.4494162423494316e-05,
+      "loss": 0.7247,
+      "step": 4302
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.4491590803888286e-05,
+      "loss": 0.8496,
+      "step": 4303
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.448901918428226e-05,
+      "loss": 0.923,
+      "step": 4304
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.448644756467624e-05,
+      "loss": 0.8545,
+      "step": 4305
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.448387594507021e-05,
+      "loss": 1.0166,
+      "step": 4306
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.448130432546418e-05,
+      "loss": 0.912,
+      "step": 4307
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.447873270585815e-05,
+      "loss": 0.8409,
+      "step": 4308
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.4476161086252124e-05,
+      "loss": 0.796,
+      "step": 4309
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.44735894666461e-05,
+      "loss": 1.1566,
+      "step": 4310
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.447101784704007e-05,
+      "loss": 0.8411,
+      "step": 4311
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.446844622743404e-05,
+      "loss": 1.2888,
+      "step": 4312
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.446587460782801e-05,
+      "loss": 0.897,
+      "step": 4313
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.4463302988221986e-05,
+      "loss": 0.8581,
+      "step": 4314
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.4460731368615956e-05,
+      "loss": 0.9078,
+      "step": 4315
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.445815974900993e-05,
+      "loss": 0.7586,
+      "step": 4316
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.44555881294039e-05,
+      "loss": 0.6666,
+      "step": 4317
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.445301650979787e-05,
+      "loss": 0.86,
+      "step": 4318
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.445044489019184e-05,
+      "loss": 0.7884,
+      "step": 4319
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.444787327058582e-05,
+      "loss": 0.8234,
+      "step": 4320
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.444530165097979e-05,
+      "loss": 0.9374,
+      "step": 4321
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.4442730031373764e-05,
+      "loss": 1.2729,
+      "step": 4322
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.4440158411767734e-05,
+      "loss": 0.7022,
+      "step": 4323
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.4437586792161704e-05,
+      "loss": 0.7225,
+      "step": 4324
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.443501517255568e-05,
+      "loss": 0.8905,
+      "step": 4325
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.443244355294965e-05,
+      "loss": 0.549,
+      "step": 4326
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.442987193334362e-05,
+      "loss": 0.738,
+      "step": 4327
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.4427300313737596e-05,
+      "loss": 0.7152,
+      "step": 4328
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.4424728694131566e-05,
+      "loss": 0.782,
+      "step": 4329
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.442215707452554e-05,
+      "loss": 0.7754,
+      "step": 4330
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.441958545491951e-05,
+      "loss": 0.8669,
+      "step": 4331
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.441701383531348e-05,
+      "loss": 0.8112,
+      "step": 4332
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.441444221570745e-05,
+      "loss": 0.7111,
+      "step": 4333
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.441187059610143e-05,
+      "loss": 0.7645,
+      "step": 4334
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.44092989764954e-05,
+      "loss": 0.8025,
+      "step": 4335
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.4406727356889374e-05,
+      "loss": 1.0204,
+      "step": 4336
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.440415573728334e-05,
+      "loss": 0.8053,
+      "step": 4337
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.440158411767731e-05,
+      "loss": 0.5323,
+      "step": 4338
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.439901249807128e-05,
+      "loss": 0.7624,
+      "step": 4339
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.439644087846526e-05,
+      "loss": 0.9057,
+      "step": 4340
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.4393869258859236e-05,
+      "loss": 0.9686,
+      "step": 4341
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.4391297639253205e-05,
+      "loss": 0.9977,
+      "step": 4342
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.4388726019647175e-05,
+      "loss": 0.6275,
+      "step": 4343
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.4386154400041145e-05,
+      "loss": 1.0984,
+      "step": 4344
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.438358278043512e-05,
+      "loss": 0.8909,
+      "step": 4345
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.43810111608291e-05,
+      "loss": 0.6072,
+      "step": 4346
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.437843954122307e-05,
+      "loss": 1.2378,
+      "step": 4347
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.437586792161704e-05,
+      "loss": 0.9742,
+      "step": 4348
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.437329630201101e-05,
+      "loss": 0.8181,
+      "step": 4349
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.4370724682404976e-05,
+      "loss": 1.0446,
+      "step": 4350
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.436815306279895e-05,
+      "loss": 0.6862,
+      "step": 4351
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.436558144319293e-05,
+      "loss": 0.5002,
+      "step": 4352
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.43630098235869e-05,
+      "loss": 1.0338,
+      "step": 4353
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.436043820398087e-05,
+      "loss": 0.7759,
+      "step": 4354
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.435786658437484e-05,
+      "loss": 0.6259,
+      "step": 4355
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.4355294964768815e-05,
+      "loss": 0.7591,
+      "step": 4356
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.4352723345162785e-05,
+      "loss": 0.8221,
+      "step": 4357
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.435015172555676e-05,
+      "loss": 1.0969,
+      "step": 4358
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.434758010595073e-05,
+      "loss": 0.6301,
+      "step": 4359
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.43450084863447e-05,
+      "loss": 0.889,
+      "step": 4360
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.434243686673868e-05,
+      "loss": 0.7604,
+      "step": 4361
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.4339865247132646e-05,
+      "loss": 0.9459,
+      "step": 4362
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.4337293627526616e-05,
+      "loss": 1.0761,
+      "step": 4363
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.433472200792059e-05,
+      "loss": 0.9195,
+      "step": 4364
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.433215038831456e-05,
+      "loss": 0.9651,
+      "step": 4365
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.432957876870853e-05,
+      "loss": 0.7998,
+      "step": 4366
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.432700714910251e-05,
+      "loss": 0.7672,
+      "step": 4367
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.432443552949648e-05,
+      "loss": 0.9741,
+      "step": 4368
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.432186390989045e-05,
+      "loss": 0.9092,
+      "step": 4369
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.4319292290284424e-05,
+      "loss": 0.8414,
+      "step": 4370
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.4316720670678394e-05,
+      "loss": 0.9235,
+      "step": 4371
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.431414905107237e-05,
+      "loss": 0.6615,
+      "step": 4372
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.431157743146634e-05,
+      "loss": 0.7695,
+      "step": 4373
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.430900581186031e-05,
+      "loss": 0.9248,
+      "step": 4374
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.430643419225428e-05,
+      "loss": 0.7384,
+      "step": 4375
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.4303862572648256e-05,
+      "loss": 0.6742,
+      "step": 4376
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.430129095304223e-05,
+      "loss": 0.6579,
+      "step": 4377
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.42987193334362e-05,
+      "loss": 1.0594,
+      "step": 4378
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.429614771383017e-05,
+      "loss": 0.7477,
+      "step": 4379
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.429357609422414e-05,
+      "loss": 0.7549,
+      "step": 4380
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.429100447461811e-05,
+      "loss": 0.8614,
+      "step": 4381
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.428843285501209e-05,
+      "loss": 1.0885,
+      "step": 4382
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.4285861235406064e-05,
+      "loss": 0.8695,
+      "step": 4383
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.4283289615800034e-05,
+      "loss": 0.3887,
+      "step": 4384
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.4280717996194003e-05,
+      "loss": 0.7129,
+      "step": 4385
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.427814637658797e-05,
+      "loss": 0.8546,
+      "step": 4386
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.427557475698195e-05,
+      "loss": 1.0004,
+      "step": 4387
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.4273003137375926e-05,
+      "loss": 1.1078,
+      "step": 4388
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.4270431517769896e-05,
+      "loss": 0.526,
+      "step": 4389
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.4267859898163865e-05,
+      "loss": 0.623,
+      "step": 4390
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.4265288278557835e-05,
+      "loss": 0.7056,
+      "step": 4391
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.426271665895181e-05,
+      "loss": 1.0163,
+      "step": 4392
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.426014503934578e-05,
+      "loss": 0.7494,
+      "step": 4393
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.425757341973976e-05,
+      "loss": 0.8462,
+      "step": 4394
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.425500180013373e-05,
+      "loss": 0.5781,
+      "step": 4395
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.42524301805277e-05,
+      "loss": 0.9763,
+      "step": 4396
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.424985856092167e-05,
+      "loss": 0.8339,
+      "step": 4397
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.424728694131564e-05,
+      "loss": 0.9009,
+      "step": 4398
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.424471532170961e-05,
+      "loss": 0.7696,
+      "step": 4399
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.424214370210359e-05,
+      "loss": 0.4207,
+      "step": 4400
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.423957208249756e-05,
+      "loss": 0.9597,
+      "step": 4401
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.423700046289153e-05,
+      "loss": 0.8373,
+      "step": 4402
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.4234428843285505e-05,
+      "loss": 0.7583,
+      "step": 4403
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.4231857223679475e-05,
+      "loss": 1.1761,
+      "step": 4404
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.4229285604073445e-05,
+      "loss": 0.8297,
+      "step": 4405
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.422671398446742e-05,
+      "loss": 0.932,
+      "step": 4406
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.422414236486139e-05,
+      "loss": 0.7836,
+      "step": 4407
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.422157074525537e-05,
+      "loss": 0.9084,
+      "step": 4408
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.421899912564934e-05,
+      "loss": 0.6383,
+      "step": 4409
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.4216427506043307e-05,
+      "loss": 1.2207,
+      "step": 4410
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.4213855886437276e-05,
+      "loss": 0.9094,
+      "step": 4411
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.421128426683125e-05,
+      "loss": 0.9859,
+      "step": 4412
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.420871264722522e-05,
+      "loss": 0.7951,
+      "step": 4413
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.42061410276192e-05,
+      "loss": 0.6305,
+      "step": 4414
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.420356940801317e-05,
+      "loss": 0.9148,
+      "step": 4415
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.420099778840714e-05,
+      "loss": 0.7763,
+      "step": 4416
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.419842616880111e-05,
+      "loss": 1.1262,
+      "step": 4417
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.4195854549195084e-05,
+      "loss": 0.6067,
+      "step": 4418
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.419328292958906e-05,
+      "loss": 1.0275,
+      "step": 4419
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.419071130998303e-05,
+      "loss": 1.1259,
+      "step": 4420
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.4188139690377e-05,
+      "loss": 0.6276,
+      "step": 4421
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.418556807077097e-05,
+      "loss": 0.7561,
+      "step": 4422
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.4182996451164946e-05,
+      "loss": 0.8312,
+      "step": 4423
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.418042483155892e-05,
+      "loss": 0.6464,
+      "step": 4424
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.417785321195289e-05,
+      "loss": 0.6718,
+      "step": 4425
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.417528159234686e-05,
+      "loss": 0.8351,
+      "step": 4426
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.417270997274083e-05,
+      "loss": 1.0905,
+      "step": 4427
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.41701383531348e-05,
+      "loss": 0.6743,
+      "step": 4428
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.416756673352878e-05,
+      "loss": 0.7169,
+      "step": 4429
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.4164995113922754e-05,
+      "loss": 0.7992,
+      "step": 4430
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.4162423494316724e-05,
+      "loss": 1.1783,
+      "step": 4431
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.4159851874710694e-05,
+      "loss": 0.8253,
+      "step": 4432
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.4157280255104664e-05,
+      "loss": 0.6459,
+      "step": 4433
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.415470863549864e-05,
+      "loss": 1.067,
+      "step": 4434
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.4152137015892616e-05,
+      "loss": 0.781,
+      "step": 4435
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.4149565396286586e-05,
+      "loss": 1.1221,
+      "step": 4436
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.4146993776680556e-05,
+      "loss": 0.7244,
+      "step": 4437
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.4144422157074526e-05,
+      "loss": 0.7805,
+      "step": 4438
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.41418505374685e-05,
+      "loss": 0.8303,
+      "step": 4439
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.413927891786247e-05,
+      "loss": 0.5994,
+      "step": 4440
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.413670729825645e-05,
+      "loss": 0.6111,
+      "step": 4441
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.413413567865042e-05,
+      "loss": 0.7499,
+      "step": 4442
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.413156405904439e-05,
+      "loss": 0.6131,
+      "step": 4443
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.412899243943836e-05,
+      "loss": 0.7261,
+      "step": 4444
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.4126420819832334e-05,
+      "loss": 1.1133,
+      "step": 4445
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.41238492002263e-05,
+      "loss": 1.0349,
+      "step": 4446
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.412127758062028e-05,
+      "loss": 0.8504,
+      "step": 4447
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.411870596101425e-05,
+      "loss": 0.7559,
+      "step": 4448
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.411613434140822e-05,
+      "loss": 0.8012,
+      "step": 4449
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.4113562721802196e-05,
+      "loss": 0.929,
+      "step": 4450
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.4110991102196165e-05,
+      "loss": 0.5196,
+      "step": 4451
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.4108419482590135e-05,
+      "loss": 1.027,
+      "step": 4452
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.410584786298411e-05,
+      "loss": 1.0235,
+      "step": 4453
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.410327624337808e-05,
+      "loss": 0.5429,
+      "step": 4454
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.410070462377206e-05,
+      "loss": 0.848,
+      "step": 4455
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.409813300416603e-05,
+      "loss": 0.7415,
+      "step": 4456
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.409556138456e-05,
+      "loss": 1.0071,
+      "step": 4457
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.409298976495397e-05,
+      "loss": 0.9015,
+      "step": 4458
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.409041814534794e-05,
+      "loss": 0.6563,
+      "step": 4459
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.408784652574191e-05,
+      "loss": 0.7293,
+      "step": 4460
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.408527490613589e-05,
+      "loss": 1.0921,
+      "step": 4461
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.408270328652986e-05,
+      "loss": 0.7004,
+      "step": 4462
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.408013166692383e-05,
+      "loss": 1.1555,
+      "step": 4463
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.40775600473178e-05,
+      "loss": 0.6866,
+      "step": 4464
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.4074988427711775e-05,
+      "loss": 0.9283,
+      "step": 4465
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.407241680810575e-05,
+      "loss": 1.0907,
+      "step": 4466
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.406984518849972e-05,
+      "loss": 1.0578,
+      "step": 4467
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.406727356889369e-05,
+      "loss": 0.901,
+      "step": 4468
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.406470194928766e-05,
+      "loss": 0.7181,
+      "step": 4469
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.406213032968164e-05,
+      "loss": 0.9943,
+      "step": 4470
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.405955871007561e-05,
+      "loss": 0.8045,
+      "step": 4471
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.405698709046958e-05,
+      "loss": 0.7316,
+      "step": 4472
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.405441547086355e-05,
+      "loss": 1.1989,
+      "step": 4473
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.405184385125752e-05,
+      "loss": 0.7254,
+      "step": 4474
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.404927223165149e-05,
+      "loss": 0.7322,
+      "step": 4475
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.404670061204547e-05,
+      "loss": 1.0573,
+      "step": 4476
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.4044128992439445e-05,
+      "loss": 1.0277,
+      "step": 4477
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.4041557372833415e-05,
+      "loss": 0.8478,
+      "step": 4478
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.4038985753227384e-05,
+      "loss": 0.6781,
+      "step": 4479
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.4036414133621354e-05,
+      "loss": 0.6139,
+      "step": 4480
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.403384251401533e-05,
+      "loss": 1.0585,
+      "step": 4481
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.40312708944093e-05,
+      "loss": 1.0455,
+      "step": 4482
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.4028699274803277e-05,
+      "loss": 0.6784,
+      "step": 4483
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.4026127655197246e-05,
+      "loss": 0.7957,
+      "step": 4484
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.4023556035591216e-05,
+      "loss": 0.9438,
+      "step": 4485
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.402098441598519e-05,
+      "loss": 0.5475,
+      "step": 4486
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.401841279637916e-05,
+      "loss": 0.4708,
+      "step": 4487
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.401584117677313e-05,
+      "loss": 0.7445,
+      "step": 4488
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.401326955716711e-05,
+      "loss": 0.8057,
+      "step": 4489
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.401069793756108e-05,
+      "loss": 0.7128,
+      "step": 4490
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.400812631795505e-05,
+      "loss": 0.5688,
+      "step": 4491
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.4005554698349024e-05,
+      "loss": 1.3391,
+      "step": 4492
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.4002983078742994e-05,
+      "loss": 0.7083,
+      "step": 4493
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.4000411459136963e-05,
+      "loss": 0.8047,
+      "step": 4494
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.399783983953094e-05,
+      "loss": 0.6221,
+      "step": 4495
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.399526821992491e-05,
+      "loss": 0.6485,
+      "step": 4496
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.3992696600318886e-05,
+      "loss": 1.0191,
+      "step": 4497
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.3990124980712856e-05,
+      "loss": 0.8381,
+      "step": 4498
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.3987553361106825e-05,
+      "loss": 0.6216,
+      "step": 4499
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.3984981741500795e-05,
+      "loss": 0.8315,
+      "step": 4500
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.398241012189477e-05,
+      "loss": 0.8185,
+      "step": 4501
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.397983850228875e-05,
+      "loss": 0.6848,
+      "step": 4502
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.397726688268272e-05,
+      "loss": 1.1155,
+      "step": 4503
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.397469526307669e-05,
+      "loss": 0.9818,
+      "step": 4504
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.397212364347066e-05,
+      "loss": 1.1591,
+      "step": 4505
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.396955202386463e-05,
+      "loss": 1.4951,
+      "step": 4506
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.39669804042586e-05,
+      "loss": 0.8053,
+      "step": 4507
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.396440878465258e-05,
+      "loss": 1.0389,
+      "step": 4508
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.396183716504655e-05,
+      "loss": 0.8479,
+      "step": 4509
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.395926554544052e-05,
+      "loss": 0.5657,
+      "step": 4510
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.395669392583449e-05,
+      "loss": 0.8879,
+      "step": 4511
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.3954122306228465e-05,
+      "loss": 0.9887,
+      "step": 4512
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.395155068662244e-05,
+      "loss": 1.0076,
+      "step": 4513
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.394897906701641e-05,
+      "loss": 0.9545,
+      "step": 4514
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.394640744741038e-05,
+      "loss": 0.7821,
+      "step": 4515
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.394383582780435e-05,
+      "loss": 0.9821,
+      "step": 4516
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.394126420819833e-05,
+      "loss": 0.994,
+      "step": 4517
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.39386925885923e-05,
+      "loss": 0.9493,
+      "step": 4518
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.393612096898627e-05,
+      "loss": 0.4966,
+      "step": 4519
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.393354934938024e-05,
+      "loss": 0.4742,
+      "step": 4520
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.393097772977421e-05,
+      "loss": 0.7502,
+      "step": 4521
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.392840611016818e-05,
+      "loss": 0.7437,
+      "step": 4522
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.392583449056216e-05,
+      "loss": 0.9222,
+      "step": 4523
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.392326287095613e-05,
+      "loss": 1.3224,
+      "step": 4524
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.3920691251350105e-05,
+      "loss": 0.7569,
+      "step": 4525
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.3918119631744075e-05,
+      "loss": 0.5721,
+      "step": 4526
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.3915548012138044e-05,
+      "loss": 0.7679,
+      "step": 4527
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.391297639253202e-05,
+      "loss": 0.6666,
+      "step": 4528
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.391040477292599e-05,
+      "loss": 0.7752,
+      "step": 4529
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.390783315331996e-05,
+      "loss": 0.8568,
+      "step": 4530
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.390526153371394e-05,
+      "loss": 0.9494,
+      "step": 4531
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.3902689914107906e-05,
+      "loss": 0.8019,
+      "step": 4532
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.390011829450188e-05,
+      "loss": 1.2116,
+      "step": 4533
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.389754667489585e-05,
+      "loss": 0.8566,
+      "step": 4534
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.389497505528982e-05,
+      "loss": 0.9468,
+      "step": 4535
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.389240343568379e-05,
+      "loss": 0.5935,
+      "step": 4536
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.388983181607777e-05,
+      "loss": 0.6032,
+      "step": 4537
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.388726019647174e-05,
+      "loss": 0.7769,
+      "step": 4538
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.3884688576865714e-05,
+      "loss": 0.9944,
+      "step": 4539
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.3882116957259684e-05,
+      "loss": 0.9279,
+      "step": 4540
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.3879545337653654e-05,
+      "loss": 0.8538,
+      "step": 4541
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.3876973718047624e-05,
+      "loss": 1.0757,
+      "step": 4542
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.38744020984416e-05,
+      "loss": 0.7031,
+      "step": 4543
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.3871830478835576e-05,
+      "loss": 0.6144,
+      "step": 4544
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.3869258859229546e-05,
+      "loss": 0.8562,
+      "step": 4545
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.3866687239623516e-05,
+      "loss": 1.1224,
+      "step": 4546
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.3864115620017486e-05,
+      "loss": 0.5822,
+      "step": 4547
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.386154400041146e-05,
+      "loss": 0.8556,
+      "step": 4548
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.385897238080544e-05,
+      "loss": 1.0544,
+      "step": 4549
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.385640076119941e-05,
+      "loss": 0.8827,
+      "step": 4550
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.385382914159338e-05,
+      "loss": 0.6473,
+      "step": 4551
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.385125752198735e-05,
+      "loss": 0.8019,
+      "step": 4552
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.384868590238132e-05,
+      "loss": 0.8351,
+      "step": 4553
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.3846114282775294e-05,
+      "loss": 0.8377,
+      "step": 4554
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.384354266316927e-05,
+      "loss": 0.7695,
+      "step": 4555
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.384097104356324e-05,
+      "loss": 0.8247,
+      "step": 4556
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.383839942395721e-05,
+      "loss": 0.8113,
+      "step": 4557
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.383582780435118e-05,
+      "loss": 0.7773,
+      "step": 4558
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.3833256184745156e-05,
+      "loss": 0.6606,
+      "step": 4559
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.3830684565139125e-05,
+      "loss": 0.9197,
+      "step": 4560
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.38281129455331e-05,
+      "loss": 0.4575,
+      "step": 4561
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.382554132592707e-05,
+      "loss": 0.9718,
+      "step": 4562
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.382296970632104e-05,
+      "loss": 0.8649,
+      "step": 4563
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.382039808671502e-05,
+      "loss": 0.932,
+      "step": 4564
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.381782646710899e-05,
+      "loss": 1.2429,
+      "step": 4565
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.381525484750296e-05,
+      "loss": 0.9971,
+      "step": 4566
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.3812683227896933e-05,
+      "loss": 0.5756,
+      "step": 4567
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.38101116082909e-05,
+      "loss": 0.7281,
+      "step": 4568
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.380753998868487e-05,
+      "loss": 0.7455,
+      "step": 4569
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.380496836907885e-05,
+      "loss": 0.9736,
+      "step": 4570
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.380239674947282e-05,
+      "loss": 1.2501,
+      "step": 4571
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.379982512986679e-05,
+      "loss": 0.7015,
+      "step": 4572
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.3797253510260765e-05,
+      "loss": 0.8134,
+      "step": 4573
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.3794681890654735e-05,
+      "loss": 1.0204,
+      "step": 4574
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.379211027104871e-05,
+      "loss": 0.9729,
+      "step": 4575
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.378953865144268e-05,
+      "loss": 0.6659,
+      "step": 4576
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.378696703183665e-05,
+      "loss": 0.9243,
+      "step": 4577
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.378439541223063e-05,
+      "loss": 1.0761,
+      "step": 4578
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.37818237926246e-05,
+      "loss": 0.8243,
+      "step": 4579
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.377925217301857e-05,
+      "loss": 1.0983,
+      "step": 4580
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.377668055341254e-05,
+      "loss": 0.9788,
+      "step": 4581
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.377410893380651e-05,
+      "loss": 0.8049,
+      "step": 4582
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.377153731420048e-05,
+      "loss": 0.8855,
+      "step": 4583
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.376896569459446e-05,
+      "loss": 0.7929,
+      "step": 4584
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.376639407498843e-05,
+      "loss": 0.6521,
+      "step": 4585
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.3763822455382405e-05,
+      "loss": 0.9829,
+      "step": 4586
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.3761250835776375e-05,
+      "loss": 0.7912,
+      "step": 4587
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.3758679216170344e-05,
+      "loss": 0.6636,
+      "step": 4588
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.3756107596564314e-05,
+      "loss": 0.7541,
+      "step": 4589
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.375353597695829e-05,
+      "loss": 0.9152,
+      "step": 4590
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.375096435735227e-05,
+      "loss": 0.8369,
+      "step": 4591
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.3748392737746237e-05,
+      "loss": 0.6022,
+      "step": 4592
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.3745821118140206e-05,
+      "loss": 0.8214,
+      "step": 4593
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.3743249498534176e-05,
+      "loss": 1.2001,
+      "step": 4594
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.374067787892815e-05,
+      "loss": 0.5486,
+      "step": 4595
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.373810625932213e-05,
+      "loss": 0.6825,
+      "step": 4596
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.37355346397161e-05,
+      "loss": 0.8958,
+      "step": 4597
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.373296302011007e-05,
+      "loss": 0.9868,
+      "step": 4598
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.373039140050404e-05,
+      "loss": 0.9574,
+      "step": 4599
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.372781978089801e-05,
+      "loss": 1.085,
+      "step": 4600
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.3725248161291984e-05,
+      "loss": 1.2117,
+      "step": 4601
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.372267654168596e-05,
+      "loss": 0.9359,
+      "step": 4602
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.372010492207993e-05,
+      "loss": 1.0667,
+      "step": 4603
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.37175333024739e-05,
+      "loss": 1.0743,
+      "step": 4604
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.371496168286787e-05,
+      "loss": 0.746,
+      "step": 4605
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.3712390063261846e-05,
+      "loss": 1.0789,
+      "step": 4606
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.3709818443655816e-05,
+      "loss": 0.8632,
+      "step": 4607
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.370724682404979e-05,
+      "loss": 0.6583,
+      "step": 4608
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.370467520444376e-05,
+      "loss": 0.7258,
+      "step": 4609
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.370210358483773e-05,
+      "loss": 0.9656,
+      "step": 4610
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.369953196523171e-05,
+      "loss": 0.7898,
+      "step": 4611
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.369696034562568e-05,
+      "loss": 0.591,
+      "step": 4612
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.369438872601965e-05,
+      "loss": 0.8303,
+      "step": 4613
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.3691817106413624e-05,
+      "loss": 0.8115,
+      "step": 4614
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.3689245486807594e-05,
+      "loss": 0.6506,
+      "step": 4615
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.368667386720156e-05,
+      "loss": 0.711,
+      "step": 4616
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.368410224759554e-05,
+      "loss": 0.7408,
+      "step": 4617
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.368153062798951e-05,
+      "loss": 1.0216,
+      "step": 4618
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.367895900838348e-05,
+      "loss": 1.0186,
+      "step": 4619
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.3676387388777455e-05,
+      "loss": 0.773,
+      "step": 4620
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.3673815769171425e-05,
+      "loss": 1.1253,
+      "step": 4621
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.36712441495654e-05,
+      "loss": 1.2471,
+      "step": 4622
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.366867252995937e-05,
+      "loss": 0.8488,
+      "step": 4623
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.366610091035334e-05,
+      "loss": 0.9588,
+      "step": 4624
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.366352929074731e-05,
+      "loss": 0.8751,
+      "step": 4625
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.366095767114129e-05,
+      "loss": 0.6622,
+      "step": 4626
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.3658386051535264e-05,
+      "loss": 0.9306,
+      "step": 4627
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.365581443192923e-05,
+      "loss": 0.7572,
+      "step": 4628
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.36532428123232e-05,
+      "loss": 0.9983,
+      "step": 4629
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.365067119271717e-05,
+      "loss": 0.6319,
+      "step": 4630
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.364809957311114e-05,
+      "loss": 0.7005,
+      "step": 4631
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.364552795350512e-05,
+      "loss": 0.6689,
+      "step": 4632
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.3642956333899095e-05,
+      "loss": 1.0304,
+      "step": 4633
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.3640384714293065e-05,
+      "loss": 0.7396,
+      "step": 4634
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.3637813094687035e-05,
+      "loss": 0.9702,
+      "step": 4635
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.3635241475081004e-05,
+      "loss": 0.776,
+      "step": 4636
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.363266985547498e-05,
+      "loss": 0.7661,
+      "step": 4637
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.363009823586896e-05,
+      "loss": 0.6672,
+      "step": 4638
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.362752661626293e-05,
+      "loss": 0.9767,
+      "step": 4639
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.36249549966569e-05,
+      "loss": 1.0287,
+      "step": 4640
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.3622383377050866e-05,
+      "loss": 0.793,
+      "step": 4641
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.361981175744484e-05,
+      "loss": 0.781,
+      "step": 4642
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.361724013783881e-05,
+      "loss": 1.1781,
+      "step": 4643
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.361466851823279e-05,
+      "loss": 0.8479,
+      "step": 4644
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.361209689862676e-05,
+      "loss": 0.6545,
+      "step": 4645
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.360952527902073e-05,
+      "loss": 1.0087,
+      "step": 4646
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.36069536594147e-05,
+      "loss": 0.7514,
+      "step": 4647
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.3604382039808674e-05,
+      "loss": 0.7536,
+      "step": 4648
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.3601810420202644e-05,
+      "loss": 1.1013,
+      "step": 4649
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.359923880059662e-05,
+      "loss": 0.8599,
+      "step": 4650
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.359666718099059e-05,
+      "loss": 0.8601,
+      "step": 4651
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.359409556138456e-05,
+      "loss": 0.8202,
+      "step": 4652
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.3591523941778536e-05,
+      "loss": 1.118,
+      "step": 4653
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.3588952322172506e-05,
+      "loss": 0.7584,
+      "step": 4654
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.3586380702566476e-05,
+      "loss": 1.012,
+      "step": 4655
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.358380908296045e-05,
+      "loss": 0.7716,
+      "step": 4656
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.358123746335442e-05,
+      "loss": 0.774,
+      "step": 4657
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.35786658437484e-05,
+      "loss": 0.6544,
+      "step": 4658
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.357609422414237e-05,
+      "loss": 1.2008,
+      "step": 4659
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.357352260453634e-05,
+      "loss": 0.9327,
+      "step": 4660
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.357095098493031e-05,
+      "loss": 1.1809,
+      "step": 4661
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.3568379365324284e-05,
+      "loss": 0.8166,
+      "step": 4662
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.3565807745718254e-05,
+      "loss": 0.8939,
+      "step": 4663
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.356323612611223e-05,
+      "loss": 0.7749,
+      "step": 4664
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.35606645065062e-05,
+      "loss": 0.9726,
+      "step": 4665
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.355809288690017e-05,
+      "loss": 0.8535,
+      "step": 4666
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.355552126729414e-05,
+      "loss": 0.8572,
+      "step": 4667
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.3552949647688116e-05,
+      "loss": 0.8647,
+      "step": 4668
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.355037802808209e-05,
+      "loss": 0.6631,
+      "step": 4669
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.354780640847606e-05,
+      "loss": 1.0933,
+      "step": 4670
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.354523478887003e-05,
+      "loss": 0.8225,
+      "step": 4671
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.3542663169264e-05,
+      "loss": 0.8247,
+      "step": 4672
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.354009154965798e-05,
+      "loss": 0.7069,
+      "step": 4673
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.3537519930051954e-05,
+      "loss": 1.0311,
+      "step": 4674
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.3534948310445924e-05,
+      "loss": 1.0325,
+      "step": 4675
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.3532376690839893e-05,
+      "loss": 0.5685,
+      "step": 4676
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.352980507123386e-05,
+      "loss": 0.9847,
+      "step": 4677
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.352723345162783e-05,
+      "loss": 0.8038,
+      "step": 4678
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.352466183202181e-05,
+      "loss": 0.5158,
+      "step": 4679
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.3522090212415786e-05,
+      "loss": 0.7609,
+      "step": 4680
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.3519518592809755e-05,
+      "loss": 0.9621,
+      "step": 4681
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.3516946973203725e-05,
+      "loss": 0.6545,
+      "step": 4682
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.3514375353597695e-05,
+      "loss": 0.909,
+      "step": 4683
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.351180373399167e-05,
+      "loss": 1.0882,
+      "step": 4684
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.350923211438564e-05,
+      "loss": 0.9106,
+      "step": 4685
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.350666049477962e-05,
+      "loss": 0.6777,
+      "step": 4686
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.350408887517359e-05,
+      "loss": 0.6477,
+      "step": 4687
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.350151725556756e-05,
+      "loss": 1.156,
+      "step": 4688
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.349894563596153e-05,
+      "loss": 1.0582,
+      "step": 4689
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.34963740163555e-05,
+      "loss": 1.1904,
+      "step": 4690
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.349380239674947e-05,
+      "loss": 1.0234,
+      "step": 4691
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.349123077714345e-05,
+      "loss": 1.085,
+      "step": 4692
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.348865915753742e-05,
+      "loss": 0.8509,
+      "step": 4693
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.348608753793139e-05,
+      "loss": 0.5518,
+      "step": 4694
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.3483515918325365e-05,
+      "loss": 0.8127,
+      "step": 4695
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.3480944298719335e-05,
+      "loss": 0.8127,
+      "step": 4696
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.3478372679113304e-05,
+      "loss": 1.0969,
+      "step": 4697
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.347580105950728e-05,
+      "loss": 1.0723,
+      "step": 4698
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.347322943990125e-05,
+      "loss": 0.613,
+      "step": 4699
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.347065782029523e-05,
+      "loss": 0.9273,
+      "step": 4700
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.3468086200689197e-05,
+      "loss": 0.8637,
+      "step": 4701
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.3465514581083166e-05,
+      "loss": 0.8236,
+      "step": 4702
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.3462942961477136e-05,
+      "loss": 0.7503,
+      "step": 4703
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.346037134187111e-05,
+      "loss": 1.0664,
+      "step": 4704
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.345779972226509e-05,
+      "loss": 0.7644,
+      "step": 4705
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.345522810265906e-05,
+      "loss": 0.9575,
+      "step": 4706
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.345265648305303e-05,
+      "loss": 0.9103,
+      "step": 4707
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.3450084863447e-05,
+      "loss": 0.5974,
+      "step": 4708
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.344751324384097e-05,
+      "loss": 0.9585,
+      "step": 4709
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.3444941624234944e-05,
+      "loss": 0.6195,
+      "step": 4710
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.344237000462892e-05,
+      "loss": 1.1053,
+      "step": 4711
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.343979838502289e-05,
+      "loss": 0.6563,
+      "step": 4712
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.343722676541686e-05,
+      "loss": 0.759,
+      "step": 4713
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.343465514581083e-05,
+      "loss": 0.9327,
+      "step": 4714
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.3432083526204806e-05,
+      "loss": 1.0852,
+      "step": 4715
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.342951190659878e-05,
+      "loss": 0.765,
+      "step": 4716
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.342694028699275e-05,
+      "loss": 1.1961,
+      "step": 4717
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.342436866738672e-05,
+      "loss": 0.622,
+      "step": 4718
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.342179704778069e-05,
+      "loss": 0.8209,
+      "step": 4719
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.341922542817467e-05,
+      "loss": 0.9229,
+      "step": 4720
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.341665380856864e-05,
+      "loss": 0.8131,
+      "step": 4721
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.3414082188962614e-05,
+      "loss": 0.5858,
+      "step": 4722
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.3411510569356584e-05,
+      "loss": 0.6808,
+      "step": 4723
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.3408938949750553e-05,
+      "loss": 0.9074,
+      "step": 4724
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.340636733014452e-05,
+      "loss": 1.0503,
+      "step": 4725
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.34037957105385e-05,
+      "loss": 0.9171,
+      "step": 4726
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.3401224090932476e-05,
+      "loss": 0.7644,
+      "step": 4727
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.3398652471326446e-05,
+      "loss": 1.2947,
+      "step": 4728
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.3396080851720415e-05,
+      "loss": 0.8835,
+      "step": 4729
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.3393509232114385e-05,
+      "loss": 0.9765,
+      "step": 4730
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.339093761250836e-05,
+      "loss": 1.1806,
+      "step": 4731
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.338836599290233e-05,
+      "loss": 0.8823,
+      "step": 4732
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.338579437329631e-05,
+      "loss": 0.8581,
+      "step": 4733
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.338322275369028e-05,
+      "loss": 0.5932,
+      "step": 4734
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.338065113408425e-05,
+      "loss": 0.9904,
+      "step": 4735
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.3378079514478224e-05,
+      "loss": 0.7489,
+      "step": 4736
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.337550789487219e-05,
+      "loss": 0.7544,
+      "step": 4737
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.337293627526616e-05,
+      "loss": 0.8399,
+      "step": 4738
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.337036465566014e-05,
+      "loss": 1.1458,
+      "step": 4739
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.336779303605411e-05,
+      "loss": 0.5544,
+      "step": 4740
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.336522141644808e-05,
+      "loss": 0.9012,
+      "step": 4741
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.3362649796842055e-05,
+      "loss": 0.7154,
+      "step": 4742
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.3360078177236025e-05,
+      "loss": 0.5619,
+      "step": 4743
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.3357506557629995e-05,
+      "loss": 0.9242,
+      "step": 4744
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.335493493802397e-05,
+      "loss": 0.9338,
+      "step": 4745
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.335236331841794e-05,
+      "loss": 0.7285,
+      "step": 4746
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.334979169881192e-05,
+      "loss": 1.0019,
+      "step": 4747
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.334722007920589e-05,
+      "loss": 0.5327,
+      "step": 4748
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.3344648459599857e-05,
+      "loss": 0.6556,
+      "step": 4749
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.3342076839993826e-05,
+      "loss": 1.1299,
+      "step": 4750
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.33395052203878e-05,
+      "loss": 0.915,
+      "step": 4751
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.333693360078178e-05,
+      "loss": 0.7699,
+      "step": 4752
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.333436198117575e-05,
+      "loss": 0.8006,
+      "step": 4753
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.333179036156972e-05,
+      "loss": 0.8174,
+      "step": 4754
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.332921874196369e-05,
+      "loss": 0.6865,
+      "step": 4755
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.332664712235766e-05,
+      "loss": 0.8756,
+      "step": 4756
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.3324075502751634e-05,
+      "loss": 0.9743,
+      "step": 4757
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.332150388314561e-05,
+      "loss": 0.6843,
+      "step": 4758
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.331893226353958e-05,
+      "loss": 1.1494,
+      "step": 4759
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.331636064393355e-05,
+      "loss": 0.9041,
+      "step": 4760
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.331378902432752e-05,
+      "loss": 1.5582,
+      "step": 4761
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.3311217404721496e-05,
+      "loss": 0.7374,
+      "step": 4762
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.330864578511547e-05,
+      "loss": 1.3966,
+      "step": 4763
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.330607416550944e-05,
+      "loss": 1.0978,
+      "step": 4764
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.330350254590341e-05,
+      "loss": 1.003,
+      "step": 4765
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.330093092629738e-05,
+      "loss": 1.143,
+      "step": 4766
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.329835930669136e-05,
+      "loss": 1.0514,
+      "step": 4767
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.329578768708533e-05,
+      "loss": 0.9483,
+      "step": 4768
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.3293216067479305e-05,
+      "loss": 0.759,
+      "step": 4769
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.3290644447873274e-05,
+      "loss": 1.0075,
+      "step": 4770
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.3288072828267244e-05,
+      "loss": 0.7997,
+      "step": 4771
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.3285501208661214e-05,
+      "loss": 0.8096,
+      "step": 4772
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.328292958905519e-05,
+      "loss": 0.767,
+      "step": 4773
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.328035796944916e-05,
+      "loss": 0.7538,
+      "step": 4774
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.3277786349843136e-05,
+      "loss": 0.8086,
+      "step": 4775
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.3275214730237106e-05,
+      "loss": 0.9767,
+      "step": 4776
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.3272643110631076e-05,
+      "loss": 0.993,
+      "step": 4777
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.327007149102505e-05,
+      "loss": 0.7886,
+      "step": 4778
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.326749987141902e-05,
+      "loss": 1.1054,
+      "step": 4779
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.326492825181299e-05,
+      "loss": 0.7816,
+      "step": 4780
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.326235663220697e-05,
+      "loss": 1.0889,
+      "step": 4781
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.325978501260094e-05,
+      "loss": 0.8947,
+      "step": 4782
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.3257213392994914e-05,
+      "loss": 0.7645,
+      "step": 4783
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.3254641773388884e-05,
+      "loss": 0.7792,
+      "step": 4784
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.325207015378285e-05,
+      "loss": 1.196,
+      "step": 4785
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.324949853417682e-05,
+      "loss": 0.562,
+      "step": 4786
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.32469269145708e-05,
+      "loss": 0.9319,
+      "step": 4787
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.324435529496477e-05,
+      "loss": 0.6352,
+      "step": 4788
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.3241783675358746e-05,
+      "loss": 0.8229,
+      "step": 4789
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.3239212055752715e-05,
+      "loss": 0.6587,
+      "step": 4790
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.3236640436146685e-05,
+      "loss": 0.8322,
+      "step": 4791
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.3234068816540655e-05,
+      "loss": 1.035,
+      "step": 4792
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.323149719693463e-05,
+      "loss": 0.857,
+      "step": 4793
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.322892557732861e-05,
+      "loss": 0.7067,
+      "step": 4794
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.322635395772258e-05,
+      "loss": 0.6464,
+      "step": 4795
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.322378233811655e-05,
+      "loss": 0.8034,
+      "step": 4796
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.322121071851052e-05,
+      "loss": 0.7195,
+      "step": 4797
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.321863909890449e-05,
+      "loss": 1.0419,
+      "step": 4798
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.321606747929847e-05,
+      "loss": 0.9632,
+      "step": 4799
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.321349585969244e-05,
+      "loss": 0.7533,
+      "step": 4800
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.321092424008641e-05,
+      "loss": 0.8134,
+      "step": 4801
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.320835262048038e-05,
+      "loss": 0.755,
+      "step": 4802
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.320578100087435e-05,
+      "loss": 1.1381,
+      "step": 4803
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.3203209381268325e-05,
+      "loss": 0.6339,
+      "step": 4804
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.32006377616623e-05,
+      "loss": 1.1571,
+      "step": 4805
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.319806614205627e-05,
+      "loss": 1.1061,
+      "step": 4806
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.319549452245024e-05,
+      "loss": 0.6953,
+      "step": 4807
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.319292290284421e-05,
+      "loss": 0.6381,
+      "step": 4808
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.319035128323819e-05,
+      "loss": 0.783,
+      "step": 4809
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.3187779663632156e-05,
+      "loss": 0.8622,
+      "step": 4810
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.318520804402613e-05,
+      "loss": 0.7473,
+      "step": 4811
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.31826364244201e-05,
+      "loss": 0.9821,
+      "step": 4812
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.318006480481407e-05,
+      "loss": 0.7689,
+      "step": 4813
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.317749318520805e-05,
+      "loss": 0.7992,
+      "step": 4814
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.317492156560202e-05,
+      "loss": 0.7148,
+      "step": 4815
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.317234994599599e-05,
+      "loss": 0.8094,
+      "step": 4816
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.3169778326389965e-05,
+      "loss": 0.887,
+      "step": 4817
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.3167206706783934e-05,
+      "loss": 0.9044,
+      "step": 4818
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.3164635087177904e-05,
+      "loss": 1.2921,
+      "step": 4819
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.316206346757188e-05,
+      "loss": 0.7441,
+      "step": 4820
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.315949184796585e-05,
+      "loss": 0.6635,
+      "step": 4821
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.315692022835982e-05,
+      "loss": 0.5859,
+      "step": 4822
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.3154348608753796e-05,
+      "loss": 0.9747,
+      "step": 4823
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.3151776989147766e-05,
+      "loss": 0.5153,
+      "step": 4824
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.314920536954174e-05,
+      "loss": 0.8836,
+      "step": 4825
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.314663374993571e-05,
+      "loss": 0.643,
+      "step": 4826
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.314406213032968e-05,
+      "loss": 0.7577,
+      "step": 4827
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.314149051072365e-05,
+      "loss": 0.8599,
+      "step": 4828
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.313891889111763e-05,
+      "loss": 1.0054,
+      "step": 4829
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.3136347271511604e-05,
+      "loss": 0.8664,
+      "step": 4830
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.3133775651905574e-05,
+      "loss": 0.9624,
+      "step": 4831
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.3131204032299544e-05,
+      "loss": 1.1103,
+      "step": 4832
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.3128632412693513e-05,
+      "loss": 0.6861,
+      "step": 4833
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.312606079308748e-05,
+      "loss": 0.9432,
+      "step": 4834
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.312348917348146e-05,
+      "loss": 0.9588,
+      "step": 4835
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.3120917553875436e-05,
+      "loss": 0.7351,
+      "step": 4836
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.3118345934269406e-05,
+      "loss": 0.7937,
+      "step": 4837
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.3115774314663375e-05,
+      "loss": 0.9878,
+      "step": 4838
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.3113202695057345e-05,
+      "loss": 0.5713,
+      "step": 4839
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.311063107545132e-05,
+      "loss": 1.0154,
+      "step": 4840
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.31080594558453e-05,
+      "loss": 0.7256,
+      "step": 4841
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.310548783623927e-05,
+      "loss": 1.0459,
+      "step": 4842
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.310291621663324e-05,
+      "loss": 0.8114,
+      "step": 4843
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.310034459702721e-05,
+      "loss": 0.6318,
+      "step": 4844
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.3097772977421184e-05,
+      "loss": 0.8712,
+      "step": 4845
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.309520135781515e-05,
+      "loss": 0.7554,
+      "step": 4846
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.309262973820913e-05,
+      "loss": 0.7491,
+      "step": 4847
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.30900581186031e-05,
+      "loss": 0.7857,
+      "step": 4848
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.308748649899707e-05,
+      "loss": 0.8389,
+      "step": 4849
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.308491487939104e-05,
+      "loss": 1.0075,
+      "step": 4850
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.3082343259785015e-05,
+      "loss": 0.649,
+      "step": 4851
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.3079771640178985e-05,
+      "loss": 0.6804,
+      "step": 4852
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.307720002057296e-05,
+      "loss": 0.8055,
+      "step": 4853
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.307462840096693e-05,
+      "loss": 0.9147,
+      "step": 4854
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.30720567813609e-05,
+      "loss": 0.8802,
+      "step": 4855
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.306948516175488e-05,
+      "loss": 0.7992,
+      "step": 4856
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.306691354214885e-05,
+      "loss": 1.096,
+      "step": 4857
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.3064341922542817e-05,
+      "loss": 0.531,
+      "step": 4858
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.306177030293679e-05,
+      "loss": 0.9512,
+      "step": 4859
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.305919868333076e-05,
+      "loss": 0.8913,
+      "step": 4860
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.305662706372474e-05,
+      "loss": 0.914,
+      "step": 4861
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.305405544411871e-05,
+      "loss": 0.7839,
+      "step": 4862
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.305148382451268e-05,
+      "loss": 1.0169,
+      "step": 4863
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.304891220490665e-05,
+      "loss": 1.1402,
+      "step": 4864
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.3046340585300625e-05,
+      "loss": 0.7193,
+      "step": 4865
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.3043768965694594e-05,
+      "loss": 0.8081,
+      "step": 4866
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.304119734608857e-05,
+      "loss": 0.9633,
+      "step": 4867
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.303862572648254e-05,
+      "loss": 0.678,
+      "step": 4868
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.303605410687651e-05,
+      "loss": 0.7371,
+      "step": 4869
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.303348248727048e-05,
+      "loss": 0.8449,
+      "step": 4870
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.3030910867664456e-05,
+      "loss": 0.7092,
+      "step": 4871
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.302833924805843e-05,
+      "loss": 0.8917,
+      "step": 4872
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.30257676284524e-05,
+      "loss": 0.768,
+      "step": 4873
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.302319600884637e-05,
+      "loss": 0.7008,
+      "step": 4874
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.302062438924034e-05,
+      "loss": 0.3742,
+      "step": 4875
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.301805276963432e-05,
+      "loss": 0.9459,
+      "step": 4876
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.3015481150028295e-05,
+      "loss": 1.1248,
+      "step": 4877
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.3012909530422264e-05,
+      "loss": 0.8317,
+      "step": 4878
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.3010337910816234e-05,
+      "loss": 0.8731,
+      "step": 4879
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.3007766291210204e-05,
+      "loss": 0.7036,
+      "step": 4880
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.3005194671604174e-05,
+      "loss": 0.9973,
+      "step": 4881
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.300262305199815e-05,
+      "loss": 0.9124,
+      "step": 4882
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.3000051432392126e-05,
+      "loss": 0.7597,
+      "step": 4883
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.2997479812786096e-05,
+      "loss": 1.0597,
+      "step": 4884
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.2994908193180066e-05,
+      "loss": 1.1293,
+      "step": 4885
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.2992336573574036e-05,
+      "loss": 0.6219,
+      "step": 4886
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.298976495396801e-05,
+      "loss": 1.0001,
+      "step": 4887
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.298719333436199e-05,
+      "loss": 0.5933,
+      "step": 4888
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.298462171475596e-05,
+      "loss": 1.0829,
+      "step": 4889
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.298205009514993e-05,
+      "loss": 0.6255,
+      "step": 4890
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.29794784755439e-05,
+      "loss": 0.9474,
+      "step": 4891
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.2976906855937874e-05,
+      "loss": 0.5436,
+      "step": 4892
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.2974335236331844e-05,
+      "loss": 1.2185,
+      "step": 4893
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.297176361672582e-05,
+      "loss": 1.0252,
+      "step": 4894
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.296919199711979e-05,
+      "loss": 0.7892,
+      "step": 4895
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.296662037751376e-05,
+      "loss": 0.7772,
+      "step": 4896
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.296404875790773e-05,
+      "loss": 0.7225,
+      "step": 4897
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.2961477138301706e-05,
+      "loss": 0.9762,
+      "step": 4898
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.2958905518695675e-05,
+      "loss": 0.624,
+      "step": 4899
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.295633389908965e-05,
+      "loss": 0.955,
+      "step": 4900
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.295376227948362e-05,
+      "loss": 1.0457,
+      "step": 4901
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.295119065987759e-05,
+      "loss": 1.1285,
+      "step": 4902
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.294861904027157e-05,
+      "loss": 0.6443,
+      "step": 4903
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.294604742066554e-05,
+      "loss": 0.9774,
+      "step": 4904
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.294347580105951e-05,
+      "loss": 0.7793,
+      "step": 4905
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.2940904181453483e-05,
+      "loss": 0.8826,
+      "step": 4906
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.293833256184745e-05,
+      "loss": 0.7364,
+      "step": 4907
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.293576094224143e-05,
+      "loss": 0.8049,
+      "step": 4908
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.29331893226354e-05,
+      "loss": 0.8483,
+      "step": 4909
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.293061770302937e-05,
+      "loss": 0.7103,
+      "step": 4910
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.292804608342334e-05,
+      "loss": 0.6646,
+      "step": 4911
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.2925474463817315e-05,
+      "loss": 1.0306,
+      "step": 4912
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.2922902844211285e-05,
+      "loss": 0.7969,
+      "step": 4913
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.292033122460526e-05,
+      "loss": 1.2313,
+      "step": 4914
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.291775960499923e-05,
+      "loss": 0.9893,
+      "step": 4915
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.29151879853932e-05,
+      "loss": 1.1654,
+      "step": 4916
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.291261636578717e-05,
+      "loss": 0.818,
+      "step": 4917
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.291004474618115e-05,
+      "loss": 0.8514,
+      "step": 4918
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.290747312657512e-05,
+      "loss": 0.9773,
+      "step": 4919
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.290490150696909e-05,
+      "loss": 0.7318,
+      "step": 4920
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.290232988736306e-05,
+      "loss": 1.0403,
+      "step": 4921
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.289975826775703e-05,
+      "loss": 0.7867,
+      "step": 4922
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.289718664815101e-05,
+      "loss": 0.7763,
+      "step": 4923
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.2894615028544985e-05,
+      "loss": 0.658,
+      "step": 4924
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.2892043408938955e-05,
+      "loss": 0.8785,
+      "step": 4925
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.2889471789332925e-05,
+      "loss": 0.7897,
+      "step": 4926
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.2886900169726894e-05,
+      "loss": 0.596,
+      "step": 4927
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.2884328550120864e-05,
+      "loss": 0.8236,
+      "step": 4928
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.288175693051484e-05,
+      "loss": 1.5105,
+      "step": 4929
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.287918531090882e-05,
+      "loss": 0.9308,
+      "step": 4930
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.2876613691302787e-05,
+      "loss": 0.8019,
+      "step": 4931
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.2874042071696756e-05,
+      "loss": 0.9446,
+      "step": 4932
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.2871470452090726e-05,
+      "loss": 0.4842,
+      "step": 4933
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.28688988324847e-05,
+      "loss": 0.5943,
+      "step": 4934
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.286632721287867e-05,
+      "loss": 0.9974,
+      "step": 4935
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.286375559327265e-05,
+      "loss": 0.9227,
+      "step": 4936
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.286118397366662e-05,
+      "loss": 0.7042,
+      "step": 4937
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.285861235406059e-05,
+      "loss": 0.7554,
+      "step": 4938
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.2856040734454564e-05,
+      "loss": 0.7068,
+      "step": 4939
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.2853469114848534e-05,
+      "loss": 0.7848,
+      "step": 4940
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.2850897495242504e-05,
+      "loss": 0.6456,
+      "step": 4941
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.284832587563648e-05,
+      "loss": 0.4823,
+      "step": 4942
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.284575425603045e-05,
+      "loss": 1.0774,
+      "step": 4943
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.284318263642442e-05,
+      "loss": 1.0105,
+      "step": 4944
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.2840611016818396e-05,
+      "loss": 0.8061,
+      "step": 4945
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.2838039397212366e-05,
+      "loss": 1.0841,
+      "step": 4946
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.2835467777606335e-05,
+      "loss": 1.0534,
+      "step": 4947
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.283289615800031e-05,
+      "loss": 0.9626,
+      "step": 4948
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.283032453839428e-05,
+      "loss": 0.7686,
+      "step": 4949
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.282775291878826e-05,
+      "loss": 0.865,
+      "step": 4950
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.282518129918223e-05,
+      "loss": 0.8318,
+      "step": 4951
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.28226096795762e-05,
+      "loss": 1.0604,
+      "step": 4952
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.282003805997017e-05,
+      "loss": 0.8605,
+      "step": 4953
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.2817466440364144e-05,
+      "loss": 0.5801,
+      "step": 4954
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.281489482075812e-05,
+      "loss": 0.7763,
+      "step": 4955
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.281232320115209e-05,
+      "loss": 0.9541,
+      "step": 4956
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.280975158154606e-05,
+      "loss": 1.0571,
+      "step": 4957
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.280717996194003e-05,
+      "loss": 1.2085,
+      "step": 4958
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.2804608342334e-05,
+      "loss": 1.0983,
+      "step": 4959
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.2802036722727975e-05,
+      "loss": 0.9332,
+      "step": 4960
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.279946510312195e-05,
+      "loss": 0.7838,
+      "step": 4961
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.279689348351592e-05,
+      "loss": 0.7059,
+      "step": 4962
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.279432186390989e-05,
+      "loss": 0.7467,
+      "step": 4963
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.279175024430386e-05,
+      "loss": 0.5627,
+      "step": 4964
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.278917862469784e-05,
+      "loss": 0.6761,
+      "step": 4965
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.2786607005091814e-05,
+      "loss": 0.7206,
+      "step": 4966
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.278403538548578e-05,
+      "loss": 0.5395,
+      "step": 4967
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.278146376587975e-05,
+      "loss": 1.0552,
+      "step": 4968
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.277889214627372e-05,
+      "loss": 0.8255,
+      "step": 4969
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.27763205266677e-05,
+      "loss": 0.7494,
+      "step": 4970
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.277374890706167e-05,
+      "loss": 1.3741,
+      "step": 4971
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.2771177287455645e-05,
+      "loss": 0.9024,
+      "step": 4972
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.2768605667849615e-05,
+      "loss": 0.6447,
+      "step": 4973
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.2766034048243585e-05,
+      "loss": 1.1977,
+      "step": 4974
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.2763462428637554e-05,
+      "loss": 0.8067,
+      "step": 4975
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.276089080903153e-05,
+      "loss": 1.0322,
+      "step": 4976
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.27583191894255e-05,
+      "loss": 0.953,
+      "step": 4977
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.275574756981948e-05,
+      "loss": 0.6818,
+      "step": 4978
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.275317595021345e-05,
+      "loss": 0.8666,
+      "step": 4979
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.2750604330607416e-05,
+      "loss": 0.7787,
+      "step": 4980
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.274803271100139e-05,
+      "loss": 0.7777,
+      "step": 4981
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.274546109139536e-05,
+      "loss": 0.743,
+      "step": 4982
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.274288947178933e-05,
+      "loss": 1.0133,
+      "step": 4983
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.274031785218331e-05,
+      "loss": 0.6511,
+      "step": 4984
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.273774623257728e-05,
+      "loss": 0.5733,
+      "step": 4985
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.2735174612971255e-05,
+      "loss": 1.2485,
+      "step": 4986
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.2732602993365224e-05,
+      "loss": 1.0978,
+      "step": 4987
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.2730031373759194e-05,
+      "loss": 0.9192,
+      "step": 4988
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.2727459754153164e-05,
+      "loss": 0.607,
+      "step": 4989
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.272488813454714e-05,
+      "loss": 0.7158,
+      "step": 4990
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.272231651494111e-05,
+      "loss": 0.9692,
+      "step": 4991
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.2719744895335086e-05,
+      "loss": 0.6385,
+      "step": 4992
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.2717173275729056e-05,
+      "loss": 0.6905,
+      "step": 4993
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.2714601656123026e-05,
+      "loss": 1.2274,
+      "step": 4994
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.2712030036516995e-05,
+      "loss": 0.5234,
+      "step": 4995
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.270945841691097e-05,
+      "loss": 0.619,
+      "step": 4996
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.270688679730495e-05,
+      "loss": 0.9073,
+      "step": 4997
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.270431517769892e-05,
+      "loss": 0.7052,
+      "step": 4998
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.270174355809289e-05,
+      "loss": 0.6946,
+      "step": 4999
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.269917193848686e-05,
+      "loss": 0.6143,
+      "step": 5000
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.2696600318880834e-05,
+      "loss": 0.9104,
+      "step": 5001
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.269402869927481e-05,
+      "loss": 1.2714,
+      "step": 5002
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.269145707966878e-05,
+      "loss": 0.7689,
+      "step": 5003
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.268888546006275e-05,
+      "loss": 1.0621,
+      "step": 5004
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.268631384045672e-05,
+      "loss": 0.6253,
+      "step": 5005
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.268374222085069e-05,
+      "loss": 1.0835,
+      "step": 5006
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.2681170601244666e-05,
+      "loss": 0.7651,
+      "step": 5007
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.267859898163864e-05,
+      "loss": 0.8697,
+      "step": 5008
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.267602736203261e-05,
+      "loss": 0.8738,
+      "step": 5009
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.267345574242658e-05,
+      "loss": 0.9107,
+      "step": 5010
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.267088412282055e-05,
+      "loss": 1.0186,
+      "step": 5011
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.266831250321453e-05,
+      "loss": 0.8992,
+      "step": 5012
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.26657408836085e-05,
+      "loss": 1.1805,
+      "step": 5013
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.2663169264002474e-05,
+      "loss": 1.0471,
+      "step": 5014
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.2660597644396443e-05,
+      "loss": 0.8632,
+      "step": 5015
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.265802602479041e-05,
+      "loss": 0.907,
+      "step": 5016
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.265545440518439e-05,
+      "loss": 0.8051,
+      "step": 5017
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.265288278557836e-05,
+      "loss": 0.5994,
+      "step": 5018
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.265031116597233e-05,
+      "loss": 0.8859,
+      "step": 5019
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.2647739546366305e-05,
+      "loss": 0.931,
+      "step": 5020
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.2645167926760275e-05,
+      "loss": 0.9333,
+      "step": 5021
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.2642596307154245e-05,
+      "loss": 0.7933,
+      "step": 5022
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.264002468754822e-05,
+      "loss": 0.7781,
+      "step": 5023
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.263745306794219e-05,
+      "loss": 0.726,
+      "step": 5024
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.263488144833617e-05,
+      "loss": 0.6418,
+      "step": 5025
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.263230982873014e-05,
+      "loss": 0.8109,
+      "step": 5026
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.262973820912411e-05,
+      "loss": 0.6847,
+      "step": 5027
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.262716658951808e-05,
+      "loss": 0.7398,
+      "step": 5028
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.262459496991205e-05,
+      "loss": 1.0283,
+      "step": 5029
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.262202335030602e-05,
+      "loss": 0.5505,
+      "step": 5030
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.26194517307e-05,
+      "loss": 1.006,
+      "step": 5031
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.261688011109397e-05,
+      "loss": 0.9986,
+      "step": 5032
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.2614308491487945e-05,
+      "loss": 0.8745,
+      "step": 5033
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.2611736871881915e-05,
+      "loss": 0.7415,
+      "step": 5034
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.2609165252275885e-05,
+      "loss": 0.6858,
+      "step": 5035
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.2606593632669854e-05,
+      "loss": 0.8462,
+      "step": 5036
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.260402201306383e-05,
+      "loss": 0.539,
+      "step": 5037
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.26014503934578e-05,
+      "loss": 0.7671,
+      "step": 5038
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.259887877385178e-05,
+      "loss": 0.8412,
+      "step": 5039
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.2596307154245747e-05,
+      "loss": 0.6767,
+      "step": 5040
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.2593735534639716e-05,
+      "loss": 0.914,
+      "step": 5041
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.2591163915033686e-05,
+      "loss": 1.0446,
+      "step": 5042
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.258859229542766e-05,
+      "loss": 1.0033,
+      "step": 5043
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.258602067582164e-05,
+      "loss": 0.9329,
+      "step": 5044
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.258344905621561e-05,
+      "loss": 0.7915,
+      "step": 5045
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.258087743660958e-05,
+      "loss": 0.8125,
+      "step": 5046
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.257830581700355e-05,
+      "loss": 0.936,
+      "step": 5047
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.2575734197397524e-05,
+      "loss": 0.8296,
+      "step": 5048
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.25731625777915e-05,
+      "loss": 1.1929,
+      "step": 5049
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.257059095818547e-05,
+      "loss": 1.1102,
+      "step": 5050
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.256801933857944e-05,
+      "loss": 1.0068,
+      "step": 5051
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.256544771897341e-05,
+      "loss": 0.8915,
+      "step": 5052
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.256287609936738e-05,
+      "loss": 0.6537,
+      "step": 5053
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.2560304479761356e-05,
+      "loss": 0.7096,
+      "step": 5054
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.255773286015533e-05,
+      "loss": 0.8194,
+      "step": 5055
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.25551612405493e-05,
+      "loss": 0.7664,
+      "step": 5056
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.255258962094327e-05,
+      "loss": 0.972,
+      "step": 5057
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.255001800133724e-05,
+      "loss": 0.8784,
+      "step": 5058
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.254744638173122e-05,
+      "loss": 1.003,
+      "step": 5059
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.254487476212519e-05,
+      "loss": 0.6015,
+      "step": 5060
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.2542303142519164e-05,
+      "loss": 0.9742,
+      "step": 5061
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.2539731522913134e-05,
+      "loss": 1.1103,
+      "step": 5062
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.2537159903307103e-05,
+      "loss": 0.5888,
+      "step": 5063
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.253458828370108e-05,
+      "loss": 0.8666,
+      "step": 5064
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.253201666409505e-05,
+      "loss": 0.9915,
+      "step": 5065
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.252944504448902e-05,
+      "loss": 1.0591,
+      "step": 5066
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.2526873424882996e-05,
+      "loss": 0.6731,
+      "step": 5067
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.2524301805276965e-05,
+      "loss": 0.8652,
+      "step": 5068
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.2521730185670935e-05,
+      "loss": 0.9154,
+      "step": 5069
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.251915856606491e-05,
+      "loss": 0.5882,
+      "step": 5070
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.251658694645888e-05,
+      "loss": 0.9193,
+      "step": 5071
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.251401532685285e-05,
+      "loss": 0.8846,
+      "step": 5072
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.251144370724683e-05,
+      "loss": 0.8979,
+      "step": 5073
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.25088720876408e-05,
+      "loss": 0.8919,
+      "step": 5074
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.2506300468034774e-05,
+      "loss": 1.5619,
+      "step": 5075
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.250372884842874e-05,
+      "loss": 0.7491,
+      "step": 5076
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.250115722882271e-05,
+      "loss": 0.9741,
+      "step": 5077
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.249858560921668e-05,
+      "loss": 0.8674,
+      "step": 5078
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.249601398961066e-05,
+      "loss": 0.8498,
+      "step": 5079
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.2493442370004636e-05,
+      "loss": 0.6615,
+      "step": 5080
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.2490870750398605e-05,
+      "loss": 1.131,
+      "step": 5081
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.2488299130792575e-05,
+      "loss": 0.8905,
+      "step": 5082
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.2485727511186545e-05,
+      "loss": 0.6475,
+      "step": 5083
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.2483155891580514e-05,
+      "loss": 0.7502,
+      "step": 5084
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.248058427197449e-05,
+      "loss": 0.5269,
+      "step": 5085
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.247801265236847e-05,
+      "loss": 1.0661,
+      "step": 5086
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.247544103276244e-05,
+      "loss": 0.5331,
+      "step": 5087
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.2472869413156407e-05,
+      "loss": 0.5547,
+      "step": 5088
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.2470297793550376e-05,
+      "loss": 0.8934,
+      "step": 5089
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.246772617394435e-05,
+      "loss": 0.927,
+      "step": 5090
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.246515455433833e-05,
+      "loss": 0.8146,
+      "step": 5091
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.24625829347323e-05,
+      "loss": 1.0193,
+      "step": 5092
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.246001131512627e-05,
+      "loss": 0.8225,
+      "step": 5093
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.245743969552024e-05,
+      "loss": 0.6721,
+      "step": 5094
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.2454868075914215e-05,
+      "loss": 0.8028,
+      "step": 5095
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.2452296456308184e-05,
+      "loss": 0.6272,
+      "step": 5096
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.244972483670216e-05,
+      "loss": 0.7716,
+      "step": 5097
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.244715321709613e-05,
+      "loss": 0.9733,
+      "step": 5098
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.24445815974901e-05,
+      "loss": 0.63,
+      "step": 5099
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.244200997788407e-05,
+      "loss": 0.9294,
+      "step": 5100
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.2439438358278046e-05,
+      "loss": 0.8568,
+      "step": 5101
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.2436866738672016e-05,
+      "loss": 0.7435,
+      "step": 5102
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.243429511906599e-05,
+      "loss": 0.6978,
+      "step": 5103
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.243172349945996e-05,
+      "loss": 0.5099,
+      "step": 5104
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.242915187985393e-05,
+      "loss": 0.6008,
+      "step": 5105
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.242658026024791e-05,
+      "loss": 0.8519,
+      "step": 5106
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.242400864064188e-05,
+      "loss": 0.9843,
+      "step": 5107
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.242143702103585e-05,
+      "loss": 0.7783,
+      "step": 5108
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.2418865401429824e-05,
+      "loss": 0.8751,
+      "step": 5109
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.2416293781823794e-05,
+      "loss": 1.1743,
+      "step": 5110
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.241372216221777e-05,
+      "loss": 0.7625,
+      "step": 5111
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.241115054261174e-05,
+      "loss": 0.4814,
+      "step": 5112
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.240857892300571e-05,
+      "loss": 1.1932,
+      "step": 5113
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.240600730339968e-05,
+      "loss": 1.0416,
+      "step": 5114
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.2403435683793656e-05,
+      "loss": 1.0216,
+      "step": 5115
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.2400864064187626e-05,
+      "loss": 0.7937,
+      "step": 5116
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.23982924445816e-05,
+      "loss": 0.9037,
+      "step": 5117
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.239572082497557e-05,
+      "loss": 0.8547,
+      "step": 5118
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.239314920536954e-05,
+      "loss": 0.8431,
+      "step": 5119
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.239057758576351e-05,
+      "loss": 0.8656,
+      "step": 5120
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.238800596615749e-05,
+      "loss": 0.7014,
+      "step": 5121
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.2385434346551464e-05,
+      "loss": 1.0067,
+      "step": 5122
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.2382862726945434e-05,
+      "loss": 0.8871,
+      "step": 5123
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.23802911073394e-05,
+      "loss": 1.0342,
+      "step": 5124
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.237771948773337e-05,
+      "loss": 0.6583,
+      "step": 5125
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.237514786812735e-05,
+      "loss": 0.6237,
+      "step": 5126
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.2372576248521326e-05,
+      "loss": 0.884,
+      "step": 5127
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.2370004628915296e-05,
+      "loss": 0.5707,
+      "step": 5128
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.2367433009309265e-05,
+      "loss": 0.8352,
+      "step": 5129
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.2364861389703235e-05,
+      "loss": 1.1876,
+      "step": 5130
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.2362289770097205e-05,
+      "loss": 0.8231,
+      "step": 5131
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.235971815049118e-05,
+      "loss": 0.7943,
+      "step": 5132
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.235714653088516e-05,
+      "loss": 0.8587,
+      "step": 5133
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.235457491127913e-05,
+      "loss": 0.7426,
+      "step": 5134
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.23520032916731e-05,
+      "loss": 0.4942,
+      "step": 5135
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.234943167206707e-05,
+      "loss": 0.968,
+      "step": 5136
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.234686005246104e-05,
+      "loss": 1.1023,
+      "step": 5137
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.234428843285501e-05,
+      "loss": 1.0344,
+      "step": 5138
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.234171681324899e-05,
+      "loss": 0.716,
+      "step": 5139
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.233914519364296e-05,
+      "loss": 1.1408,
+      "step": 5140
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.233657357403693e-05,
+      "loss": 0.7494,
+      "step": 5141
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.2334001954430905e-05,
+      "loss": 0.5509,
+      "step": 5142
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.2331430334824875e-05,
+      "loss": 1.0945,
+      "step": 5143
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.2328858715218845e-05,
+      "loss": 1.3654,
+      "step": 5144
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.232628709561282e-05,
+      "loss": 0.8236,
+      "step": 5145
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.232371547600679e-05,
+      "loss": 0.8431,
+      "step": 5146
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.232114385640076e-05,
+      "loss": 0.768,
+      "step": 5147
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.231857223679474e-05,
+      "loss": 0.7593,
+      "step": 5148
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.2316000617188706e-05,
+      "loss": 1.0842,
+      "step": 5149
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.2313428997582676e-05,
+      "loss": 0.6865,
+      "step": 5150
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.231085737797665e-05,
+      "loss": 0.7805,
+      "step": 5151
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.230828575837062e-05,
+      "loss": 0.7809,
+      "step": 5152
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.23057141387646e-05,
+      "loss": 1.0524,
+      "step": 5153
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.230314251915857e-05,
+      "loss": 1.295,
+      "step": 5154
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.230057089955254e-05,
+      "loss": 0.8796,
+      "step": 5155
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.229799927994651e-05,
+      "loss": 0.9421,
+      "step": 5156
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.2295427660340484e-05,
+      "loss": 1.2088,
+      "step": 5157
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.229285604073446e-05,
+      "loss": 1.0523,
+      "step": 5158
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.229028442112843e-05,
+      "loss": 1.1031,
+      "step": 5159
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.22877128015224e-05,
+      "loss": 0.7702,
+      "step": 5160
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.228514118191637e-05,
+      "loss": 0.9114,
+      "step": 5161
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.228256956231034e-05,
+      "loss": 1.0605,
+      "step": 5162
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.2279997942704316e-05,
+      "loss": 0.9834,
+      "step": 5163
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.227742632309829e-05,
+      "loss": 0.7951,
+      "step": 5164
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.227485470349226e-05,
+      "loss": 0.7102,
+      "step": 5165
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.227228308388623e-05,
+      "loss": 1.0213,
+      "step": 5166
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.22697114642802e-05,
+      "loss": 0.7714,
+      "step": 5167
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.226713984467418e-05,
+      "loss": 0.7982,
+      "step": 5168
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.2264568225068154e-05,
+      "loss": 0.6769,
+      "step": 5169
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.2261996605462124e-05,
+      "loss": 0.8527,
+      "step": 5170
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.2259424985856094e-05,
+      "loss": 0.6233,
+      "step": 5171
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.2256853366250063e-05,
+      "loss": 0.8438,
+      "step": 5172
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.225428174664404e-05,
+      "loss": 0.7927,
+      "step": 5173
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.225171012703801e-05,
+      "loss": 0.7608,
+      "step": 5174
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.2249138507431986e-05,
+      "loss": 1.0233,
+      "step": 5175
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.2246566887825956e-05,
+      "loss": 0.8799,
+      "step": 5176
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.2243995268219925e-05,
+      "loss": 1.0616,
+      "step": 5177
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.2241423648613895e-05,
+      "loss": 0.647,
+      "step": 5178
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.223885202900787e-05,
+      "loss": 0.6096,
+      "step": 5179
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.223628040940185e-05,
+      "loss": 0.8673,
+      "step": 5180
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.223370878979582e-05,
+      "loss": 0.8849,
+      "step": 5181
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.223113717018979e-05,
+      "loss": 1.0319,
+      "step": 5182
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.222856555058376e-05,
+      "loss": 0.7131,
+      "step": 5183
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.2225993930977734e-05,
+      "loss": 0.7236,
+      "step": 5184
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.22234223113717e-05,
+      "loss": 0.8731,
+      "step": 5185
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.222085069176568e-05,
+      "loss": 0.9671,
+      "step": 5186
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.221827907215965e-05,
+      "loss": 0.4066,
+      "step": 5187
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.221570745255362e-05,
+      "loss": 0.7763,
+      "step": 5188
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.2213135832947596e-05,
+      "loss": 0.6221,
+      "step": 5189
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.2210564213341565e-05,
+      "loss": 0.7669,
+      "step": 5190
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.2207992593735535e-05,
+      "loss": 0.709,
+      "step": 5191
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.220542097412951e-05,
+      "loss": 0.6023,
+      "step": 5192
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.220284935452348e-05,
+      "loss": 0.816,
+      "step": 5193
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.220027773491745e-05,
+      "loss": 0.8301,
+      "step": 5194
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.219770611531143e-05,
+      "loss": 0.7771,
+      "step": 5195
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.21951344957054e-05,
+      "loss": 0.8348,
+      "step": 5196
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.2192562876099367e-05,
+      "loss": 1.0224,
+      "step": 5197
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.218999125649334e-05,
+      "loss": 0.8729,
+      "step": 5198
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.218741963688731e-05,
+      "loss": 0.7937,
+      "step": 5199
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.218484801728129e-05,
+      "loss": 0.528,
+      "step": 5200
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.218227639767526e-05,
+      "loss": 0.9669,
+      "step": 5201
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.217970477806923e-05,
+      "loss": 0.8785,
+      "step": 5202
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.21771331584632e-05,
+      "loss": 0.6823,
+      "step": 5203
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.2174561538857175e-05,
+      "loss": 1.0836,
+      "step": 5204
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.217198991925115e-05,
+      "loss": 0.597,
+      "step": 5205
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.216941829964512e-05,
+      "loss": 1.2264,
+      "step": 5206
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.216684668003909e-05,
+      "loss": 0.9294,
+      "step": 5207
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.216427506043306e-05,
+      "loss": 0.8424,
+      "step": 5208
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.216170344082703e-05,
+      "loss": 1.0099,
+      "step": 5209
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.2159131821221006e-05,
+      "loss": 1.2565,
+      "step": 5210
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.215656020161498e-05,
+      "loss": 1.0104,
+      "step": 5211
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.215398858200895e-05,
+      "loss": 0.9789,
+      "step": 5212
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.215141696240292e-05,
+      "loss": 0.6546,
+      "step": 5213
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.214884534279689e-05,
+      "loss": 0.8099,
+      "step": 5214
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.214627372319087e-05,
+      "loss": 0.7404,
+      "step": 5215
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.2143702103584845e-05,
+      "loss": 0.7469,
+      "step": 5216
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.2141130483978814e-05,
+      "loss": 0.6048,
+      "step": 5217
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.2138558864372784e-05,
+      "loss": 0.8653,
+      "step": 5218
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.2135987244766754e-05,
+      "loss": 1.0643,
+      "step": 5219
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.213341562516073e-05,
+      "loss": 0.6987,
+      "step": 5220
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.21308440055547e-05,
+      "loss": 0.799,
+      "step": 5221
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.2128272385948676e-05,
+      "loss": 1.1896,
+      "step": 5222
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.2125700766342646e-05,
+      "loss": 0.8754,
+      "step": 5223
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.2123129146736616e-05,
+      "loss": 0.5214,
+      "step": 5224
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.2120557527130586e-05,
+      "loss": 0.8675,
+      "step": 5225
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.211798590752456e-05,
+      "loss": 1.0158,
+      "step": 5226
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.211541428791853e-05,
+      "loss": 0.7564,
+      "step": 5227
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.211284266831251e-05,
+      "loss": 0.7554,
+      "step": 5228
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.211027104870648e-05,
+      "loss": 1.0899,
+      "step": 5229
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.210769942910045e-05,
+      "loss": 0.8914,
+      "step": 5230
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.2105127809494424e-05,
+      "loss": 0.9797,
+      "step": 5231
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.2102556189888394e-05,
+      "loss": 0.8437,
+      "step": 5232
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.209998457028236e-05,
+      "loss": 0.7743,
+      "step": 5233
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.209741295067634e-05,
+      "loss": 0.863,
+      "step": 5234
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.209484133107031e-05,
+      "loss": 0.9374,
+      "step": 5235
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.2092269711464286e-05,
+      "loss": 0.7217,
+      "step": 5236
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.2089698091858256e-05,
+      "loss": 1.0097,
+      "step": 5237
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.2087126472252225e-05,
+      "loss": 0.9067,
+      "step": 5238
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.2084554852646195e-05,
+      "loss": 0.7778,
+      "step": 5239
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.208198323304017e-05,
+      "loss": 0.9082,
+      "step": 5240
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.207941161343414e-05,
+      "loss": 0.7406,
+      "step": 5241
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.207683999382812e-05,
+      "loss": 0.9183,
+      "step": 5242
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.207426837422209e-05,
+      "loss": 1.2089,
+      "step": 5243
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.207169675461606e-05,
+      "loss": 0.9448,
+      "step": 5244
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.206912513501003e-05,
+      "loss": 0.7936,
+      "step": 5245
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.2066553515404e-05,
+      "loss": 0.8842,
+      "step": 5246
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.206398189579798e-05,
+      "loss": 0.5921,
+      "step": 5247
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.206141027619195e-05,
+      "loss": 0.9907,
+      "step": 5248
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.205883865658592e-05,
+      "loss": 0.8354,
+      "step": 5249
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.205626703697989e-05,
+      "loss": 0.6259,
+      "step": 5250
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.2053695417373865e-05,
+      "loss": 0.7001,
+      "step": 5251
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.205112379776784e-05,
+      "loss": 0.6005,
+      "step": 5252
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.204855217816181e-05,
+      "loss": 0.9171,
+      "step": 5253
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.204598055855578e-05,
+      "loss": 0.947,
+      "step": 5254
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.204340893894975e-05,
+      "loss": 0.9692,
+      "step": 5255
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.204083731934372e-05,
+      "loss": 0.7277,
+      "step": 5256
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.20382656997377e-05,
+      "loss": 1.1569,
+      "step": 5257
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.203569408013167e-05,
+      "loss": 0.9482,
+      "step": 5258
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.203312246052564e-05,
+      "loss": 1.1616,
+      "step": 5259
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.203055084091961e-05,
+      "loss": 0.9387,
+      "step": 5260
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.202797922131358e-05,
+      "loss": 0.9129,
+      "step": 5261
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.202540760170756e-05,
+      "loss": 0.787,
+      "step": 5262
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.202283598210153e-05,
+      "loss": 0.8782,
+      "step": 5263
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.2020264362495505e-05,
+      "loss": 0.6647,
+      "step": 5264
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.2017692742889475e-05,
+      "loss": 0.9162,
+      "step": 5265
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.2015121123283444e-05,
+      "loss": 0.7771,
+      "step": 5266
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.201254950367742e-05,
+      "loss": 0.7917,
+      "step": 5267
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.200997788407139e-05,
+      "loss": 0.4827,
+      "step": 5268
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.200740626446536e-05,
+      "loss": 0.6447,
+      "step": 5269
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.2004834644859337e-05,
+      "loss": 0.7388,
+      "step": 5270
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.2002263025253306e-05,
+      "loss": 0.8422,
+      "step": 5271
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.1999691405647276e-05,
+      "loss": 0.8754,
+      "step": 5272
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.199711978604125e-05,
+      "loss": 0.8373,
+      "step": 5273
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.199454816643522e-05,
+      "loss": 0.7684,
+      "step": 5274
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.199197654682919e-05,
+      "loss": 0.8519,
+      "step": 5275
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.198940492722317e-05,
+      "loss": 0.8803,
+      "step": 5276
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.198683330761714e-05,
+      "loss": 0.7665,
+      "step": 5277
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.1984261688011114e-05,
+      "loss": 0.8657,
+      "step": 5278
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.1981690068405084e-05,
+      "loss": 0.8156,
+      "step": 5279
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.1979118448799054e-05,
+      "loss": 1.1655,
+      "step": 5280
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.1976546829193023e-05,
+      "loss": 1.0531,
+      "step": 5281
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.1973975209587e-05,
+      "loss": 0.4101,
+      "step": 5282
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.1971403589980976e-05,
+      "loss": 0.8221,
+      "step": 5283
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.1968831970374946e-05,
+      "loss": 1.0994,
+      "step": 5284
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.1966260350768916e-05,
+      "loss": 0.6305,
+      "step": 5285
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.1963688731162885e-05,
+      "loss": 0.8223,
+      "step": 5286
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.1961117111556855e-05,
+      "loss": 1.2592,
+      "step": 5287
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.195854549195083e-05,
+      "loss": 0.9257,
+      "step": 5288
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.195597387234481e-05,
+      "loss": 1.286,
+      "step": 5289
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.195340225273878e-05,
+      "loss": 0.7452,
+      "step": 5290
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.195083063313275e-05,
+      "loss": 1.0023,
+      "step": 5291
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.194825901352672e-05,
+      "loss": 0.9687,
+      "step": 5292
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.1945687393920694e-05,
+      "loss": 0.6147,
+      "step": 5293
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.194311577431467e-05,
+      "loss": 0.7754,
+      "step": 5294
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.194054415470864e-05,
+      "loss": 0.7209,
+      "step": 5295
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.193797253510261e-05,
+      "loss": 0.7848,
+      "step": 5296
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.193540091549658e-05,
+      "loss": 1.102,
+      "step": 5297
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.1932829295890556e-05,
+      "loss": 1.0271,
+      "step": 5298
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.1930257676284525e-05,
+      "loss": 0.8797,
+      "step": 5299
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.19276860566785e-05,
+      "loss": 0.8096,
+      "step": 5300
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.192511443707247e-05,
+      "loss": 0.6666,
+      "step": 5301
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.192254281746644e-05,
+      "loss": 0.7652,
+      "step": 5302
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.191997119786041e-05,
+      "loss": 1.2269,
+      "step": 5303
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.191739957825439e-05,
+      "loss": 0.8396,
+      "step": 5304
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.191482795864836e-05,
+      "loss": 0.9454,
+      "step": 5305
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.191225633904233e-05,
+      "loss": 0.8399,
+      "step": 5306
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.19096847194363e-05,
+      "loss": 0.8603,
+      "step": 5307
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.190711309983027e-05,
+      "loss": 0.6687,
+      "step": 5308
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.190454148022425e-05,
+      "loss": 1.0174,
+      "step": 5309
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.190196986061822e-05,
+      "loss": 1.0184,
+      "step": 5310
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.189939824101219e-05,
+      "loss": 0.722,
+      "step": 5311
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.1896826621406165e-05,
+      "loss": 0.6142,
+      "step": 5312
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.1894255001800135e-05,
+      "loss": 0.7104,
+      "step": 5313
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.189168338219411e-05,
+      "loss": 0.631,
+      "step": 5314
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.188911176258808e-05,
+      "loss": 0.8325,
+      "step": 5315
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.188654014298205e-05,
+      "loss": 0.776,
+      "step": 5316
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.188396852337602e-05,
+      "loss": 0.7396,
+      "step": 5317
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.188139690377e-05,
+      "loss": 0.7513,
+      "step": 5318
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.1878825284163966e-05,
+      "loss": 0.8691,
+      "step": 5319
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.187625366455794e-05,
+      "loss": 0.9432,
+      "step": 5320
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.187368204495191e-05,
+      "loss": 0.7641,
+      "step": 5321
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.187111042534588e-05,
+      "loss": 1.0385,
+      "step": 5322
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.186853880573985e-05,
+      "loss": 0.893,
+      "step": 5323
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.186596718613383e-05,
+      "loss": 0.9156,
+      "step": 5324
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.1863395566527805e-05,
+      "loss": 0.6379,
+      "step": 5325
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.1860823946921774e-05,
+      "loss": 0.9337,
+      "step": 5326
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.1858252327315744e-05,
+      "loss": 0.8858,
+      "step": 5327
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.1855680707709714e-05,
+      "loss": 0.972,
+      "step": 5328
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.185310908810369e-05,
+      "loss": 0.9363,
+      "step": 5329
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.185053746849767e-05,
+      "loss": 0.9693,
+      "step": 5330
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.1847965848891636e-05,
+      "loss": 0.8015,
+      "step": 5331
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.1845394229285606e-05,
+      "loss": 1.1279,
+      "step": 5332
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.1842822609679576e-05,
+      "loss": 0.7279,
+      "step": 5333
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.1840250990073546e-05,
+      "loss": 0.8321,
+      "step": 5334
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.183767937046752e-05,
+      "loss": 0.7231,
+      "step": 5335
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.18351077508615e-05,
+      "loss": 1.1358,
+      "step": 5336
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.183253613125547e-05,
+      "loss": 0.8437,
+      "step": 5337
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.182996451164944e-05,
+      "loss": 0.8975,
+      "step": 5338
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.182739289204341e-05,
+      "loss": 0.9526,
+      "step": 5339
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.1824821272437384e-05,
+      "loss": 0.555,
+      "step": 5340
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.182224965283136e-05,
+      "loss": 0.9628,
+      "step": 5341
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.181967803322533e-05,
+      "loss": 0.9513,
+      "step": 5342
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.18171064136193e-05,
+      "loss": 0.795,
+      "step": 5343
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.181453479401327e-05,
+      "loss": 0.9769,
+      "step": 5344
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.1811963174407246e-05,
+      "loss": 0.7238,
+      "step": 5345
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.1809391554801216e-05,
+      "loss": 0.8301,
+      "step": 5346
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.180681993519519e-05,
+      "loss": 1.0724,
+      "step": 5347
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.180424831558916e-05,
+      "loss": 0.7421,
+      "step": 5348
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.180167669598313e-05,
+      "loss": 1.2524,
+      "step": 5349
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.17991050763771e-05,
+      "loss": 0.8721,
+      "step": 5350
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.179653345677108e-05,
+      "loss": 0.7913,
+      "step": 5351
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.179396183716505e-05,
+      "loss": 0.7025,
+      "step": 5352
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.1791390217559024e-05,
+      "loss": 1.0546,
+      "step": 5353
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.1788818597952993e-05,
+      "loss": 0.8716,
+      "step": 5354
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.178624697834696e-05,
+      "loss": 0.6327,
+      "step": 5355
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.178367535874094e-05,
+      "loss": 1.0162,
+      "step": 5356
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.178110373913491e-05,
+      "loss": 0.7546,
+      "step": 5357
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.177853211952888e-05,
+      "loss": 0.74,
+      "step": 5358
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.1775960499922855e-05,
+      "loss": 1.0077,
+      "step": 5359
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.1773388880316825e-05,
+      "loss": 0.6645,
+      "step": 5360
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.17708172607108e-05,
+      "loss": 0.9752,
+      "step": 5361
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.176824564110477e-05,
+      "loss": 0.807,
+      "step": 5362
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.176567402149874e-05,
+      "loss": 0.7932,
+      "step": 5363
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.176310240189271e-05,
+      "loss": 0.5561,
+      "step": 5364
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.176053078228669e-05,
+      "loss": 0.4693,
+      "step": 5365
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.175795916268066e-05,
+      "loss": 1.1602,
+      "step": 5366
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.175538754307463e-05,
+      "loss": 0.9217,
+      "step": 5367
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.17528159234686e-05,
+      "loss": 0.8401,
+      "step": 5368
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.175024430386257e-05,
+      "loss": 0.6467,
+      "step": 5369
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.174767268425654e-05,
+      "loss": 0.7687,
+      "step": 5370
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.174510106465052e-05,
+      "loss": 1.4309,
+      "step": 5371
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.1742529445044495e-05,
+      "loss": 0.8275,
+      "step": 5372
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.1739957825438465e-05,
+      "loss": 0.8033,
+      "step": 5373
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.1737386205832435e-05,
+      "loss": 0.8526,
+      "step": 5374
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.1734814586226404e-05,
+      "loss": 0.6837,
+      "step": 5375
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.173224296662038e-05,
+      "loss": 0.7342,
+      "step": 5376
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.172967134701436e-05,
+      "loss": 1.0441,
+      "step": 5377
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.172709972740833e-05,
+      "loss": 0.9925,
+      "step": 5378
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.1724528107802297e-05,
+      "loss": 0.8016,
+      "step": 5379
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.1721956488196266e-05,
+      "loss": 0.4939,
+      "step": 5380
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.1719384868590236e-05,
+      "loss": 0.765,
+      "step": 5381
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.171681324898421e-05,
+      "loss": 0.8097,
+      "step": 5382
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.171424162937819e-05,
+      "loss": 0.8675,
+      "step": 5383
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.171167000977216e-05,
+      "loss": 0.8416,
+      "step": 5384
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.170909839016613e-05,
+      "loss": 0.6914,
+      "step": 5385
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.17065267705601e-05,
+      "loss": 0.9384,
+      "step": 5386
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.1703955150954074e-05,
+      "loss": 0.6861,
+      "step": 5387
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.1701383531348044e-05,
+      "loss": 0.5811,
+      "step": 5388
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.169881191174202e-05,
+      "loss": 0.9161,
+      "step": 5389
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.169624029213599e-05,
+      "loss": 0.9115,
+      "step": 5390
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.169366867252996e-05,
+      "loss": 0.7471,
+      "step": 5391
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.1691097052923936e-05,
+      "loss": 0.7082,
+      "step": 5392
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.1688525433317906e-05,
+      "loss": 1.2912,
+      "step": 5393
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.1685953813711876e-05,
+      "loss": 0.8657,
+      "step": 5394
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.168338219410585e-05,
+      "loss": 0.9558,
+      "step": 5395
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.168081057449982e-05,
+      "loss": 1.0073,
+      "step": 5396
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.167823895489379e-05,
+      "loss": 0.9849,
+      "step": 5397
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.167566733528777e-05,
+      "loss": 0.7745,
+      "step": 5398
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.167309571568174e-05,
+      "loss": 0.6988,
+      "step": 5399
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.167052409607571e-05,
+      "loss": 0.98,
+      "step": 5400
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.1667952476469684e-05,
+      "loss": 1.0767,
+      "step": 5401
+    },
+    {
+      "epoch": 1.0,
+      "eval_accuracy": 0.7086942183187822,
+      "eval_loss": 0.8446734547615051,
+      "eval_runtime": 42.0438,
+      "eval_samples_per_second": 1370.311,
+      "eval_steps_per_second": 42.836,
+      "step": 5401
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.1665380856863654e-05,
+      "loss": 1.0363,
+      "step": 5402
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.166280923725763e-05,
+      "loss": 1.1286,
+      "step": 5403
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.16602376176516e-05,
+      "loss": 0.4987,
+      "step": 5404
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.165766599804557e-05,
+      "loss": 0.5814,
+      "step": 5405
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.165509437843954e-05,
+      "loss": 0.7951,
+      "step": 5406
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.1652522758833515e-05,
+      "loss": 0.7535,
+      "step": 5407
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.164995113922749e-05,
+      "loss": 0.783,
+      "step": 5408
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.164737951962146e-05,
+      "loss": 0.7723,
+      "step": 5409
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.164480790001543e-05,
+      "loss": 0.5825,
+      "step": 5410
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.16422362804094e-05,
+      "loss": 0.8792,
+      "step": 5411
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.163966466080337e-05,
+      "loss": 1.1684,
+      "step": 5412
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.163709304119735e-05,
+      "loss": 0.7389,
+      "step": 5413
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.1634521421591324e-05,
+      "loss": 0.8386,
+      "step": 5414
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.163194980198529e-05,
+      "loss": 0.5089,
+      "step": 5415
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.162937818237926e-05,
+      "loss": 1.045,
+      "step": 5416
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.162680656277323e-05,
+      "loss": 0.685,
+      "step": 5417
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.162423494316721e-05,
+      "loss": 0.7332,
+      "step": 5418
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.1621663323561186e-05,
+      "loss": 0.7272,
+      "step": 5419
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.1619091703955155e-05,
+      "loss": 0.7162,
+      "step": 5420
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.1616520084349125e-05,
+      "loss": 0.5683,
+      "step": 5421
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.1613948464743095e-05,
+      "loss": 0.5655,
+      "step": 5422
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.161137684513707e-05,
+      "loss": 0.8097,
+      "step": 5423
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.160880522553104e-05,
+      "loss": 1.1089,
+      "step": 5424
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.160623360592502e-05,
+      "loss": 0.4658,
+      "step": 5425
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.160366198631899e-05,
+      "loss": 0.8775,
+      "step": 5426
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.160109036671296e-05,
+      "loss": 0.6451,
+      "step": 5427
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.1598518747106926e-05,
+      "loss": 0.8333,
+      "step": 5428
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.15959471275009e-05,
+      "loss": 0.926,
+      "step": 5429
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.159337550789487e-05,
+      "loss": 1.1586,
+      "step": 5430
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.159080388828885e-05,
+      "loss": 0.5491,
+      "step": 5431
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.158823226868282e-05,
+      "loss": 0.6898,
+      "step": 5432
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.158566064907679e-05,
+      "loss": 0.7617,
+      "step": 5433
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.1583089029470765e-05,
+      "loss": 0.546,
+      "step": 5434
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.1580517409864734e-05,
+      "loss": 0.6697,
+      "step": 5435
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.1577945790258704e-05,
+      "loss": 0.8,
+      "step": 5436
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.157537417065268e-05,
+      "loss": 0.7777,
+      "step": 5437
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.157280255104665e-05,
+      "loss": 1.1658,
+      "step": 5438
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.157023093144063e-05,
+      "loss": 0.4825,
+      "step": 5439
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.1567659311834596e-05,
+      "loss": 0.9032,
+      "step": 5440
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.1565087692228566e-05,
+      "loss": 0.7825,
+      "step": 5441
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.1562516072622536e-05,
+      "loss": 0.7221,
+      "step": 5442
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.155994445301651e-05,
+      "loss": 0.8738,
+      "step": 5443
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.155737283341048e-05,
+      "loss": 0.8532,
+      "step": 5444
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.155480121380446e-05,
+      "loss": 0.7447,
+      "step": 5445
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.155222959419843e-05,
+      "loss": 0.7996,
+      "step": 5446
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.15496579745924e-05,
+      "loss": 0.6416,
+      "step": 5447
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.154708635498637e-05,
+      "loss": 0.8627,
+      "step": 5448
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.1544514735380344e-05,
+      "loss": 0.7363,
+      "step": 5449
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.154194311577432e-05,
+      "loss": 0.8921,
+      "step": 5450
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.153937149616829e-05,
+      "loss": 0.6561,
+      "step": 5451
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.153679987656226e-05,
+      "loss": 0.5396,
+      "step": 5452
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.153422825695623e-05,
+      "loss": 0.867,
+      "step": 5453
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.1531656637350206e-05,
+      "loss": 0.8911,
+      "step": 5454
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.152908501774418e-05,
+      "loss": 0.8077,
+      "step": 5455
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.152651339813815e-05,
+      "loss": 0.8119,
+      "step": 5456
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.152394177853212e-05,
+      "loss": 0.8039,
+      "step": 5457
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.152137015892609e-05,
+      "loss": 0.8801,
+      "step": 5458
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.151879853932006e-05,
+      "loss": 0.5812,
+      "step": 5459
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.151622691971404e-05,
+      "loss": 0.8839,
+      "step": 5460
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.1513655300108014e-05,
+      "loss": 0.8088,
+      "step": 5461
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.1511083680501984e-05,
+      "loss": 0.7527,
+      "step": 5462
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.150851206089595e-05,
+      "loss": 0.8498,
+      "step": 5463
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.150594044128992e-05,
+      "loss": 0.6405,
+      "step": 5464
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.15033688216839e-05,
+      "loss": 0.5311,
+      "step": 5465
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.150079720207787e-05,
+      "loss": 0.7977,
+      "step": 5466
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.1498225582471846e-05,
+      "loss": 0.7435,
+      "step": 5467
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.1495653962865815e-05,
+      "loss": 0.9302,
+      "step": 5468
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.1493082343259785e-05,
+      "loss": 1.1099,
+      "step": 5469
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.149051072365376e-05,
+      "loss": 0.9888,
+      "step": 5470
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.148793910404773e-05,
+      "loss": 0.7164,
+      "step": 5471
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.14853674844417e-05,
+      "loss": 0.8525,
+      "step": 5472
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.148279586483568e-05,
+      "loss": 0.7742,
+      "step": 5473
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.148022424522965e-05,
+      "loss": 0.8016,
+      "step": 5474
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.147765262562362e-05,
+      "loss": 0.8215,
+      "step": 5475
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.147508100601759e-05,
+      "loss": 0.8449,
+      "step": 5476
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.147250938641156e-05,
+      "loss": 0.7766,
+      "step": 5477
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.146993776680554e-05,
+      "loss": 0.8608,
+      "step": 5478
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.146736614719951e-05,
+      "loss": 0.9559,
+      "step": 5479
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.146479452759348e-05,
+      "loss": 0.6755,
+      "step": 5480
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.1462222907987455e-05,
+      "loss": 0.99,
+      "step": 5481
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.1459651288381425e-05,
+      "loss": 0.6849,
+      "step": 5482
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.1457079668775395e-05,
+      "loss": 0.8731,
+      "step": 5483
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.145450804916937e-05,
+      "loss": 0.5566,
+      "step": 5484
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.145193642956334e-05,
+      "loss": 1.2157,
+      "step": 5485
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.144936480995732e-05,
+      "loss": 0.637,
+      "step": 5486
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.144679319035129e-05,
+      "loss": 0.9331,
+      "step": 5487
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.1444221570745257e-05,
+      "loss": 0.9402,
+      "step": 5488
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.1441649951139226e-05,
+      "loss": 0.5804,
+      "step": 5489
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.14390783315332e-05,
+      "loss": 0.8495,
+      "step": 5490
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.143650671192717e-05,
+      "loss": 0.4789,
+      "step": 5491
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.143393509232115e-05,
+      "loss": 0.5706,
+      "step": 5492
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.143136347271512e-05,
+      "loss": 0.6739,
+      "step": 5493
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.142879185310909e-05,
+      "loss": 0.9962,
+      "step": 5494
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.142622023350306e-05,
+      "loss": 0.8892,
+      "step": 5495
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.1423648613897034e-05,
+      "loss": 0.8268,
+      "step": 5496
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.142107699429101e-05,
+      "loss": 0.7556,
+      "step": 5497
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.141850537468498e-05,
+      "loss": 0.9656,
+      "step": 5498
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.141593375507895e-05,
+      "loss": 0.6776,
+      "step": 5499
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.141336213547292e-05,
+      "loss": 0.8812,
+      "step": 5500
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.1410790515866896e-05,
+      "loss": 0.6942,
+      "step": 5501
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.140821889626087e-05,
+      "loss": 0.5382,
+      "step": 5502
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.140564727665484e-05,
+      "loss": 0.6741,
+      "step": 5503
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.140307565704881e-05,
+      "loss": 0.7914,
+      "step": 5504
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.140050403744278e-05,
+      "loss": 0.9826,
+      "step": 5505
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.139793241783675e-05,
+      "loss": 0.8358,
+      "step": 5506
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.139536079823073e-05,
+      "loss": 0.8083,
+      "step": 5507
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.1392789178624704e-05,
+      "loss": 0.9722,
+      "step": 5508
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.1390217559018674e-05,
+      "loss": 0.7309,
+      "step": 5509
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.1387645939412644e-05,
+      "loss": 0.6571,
+      "step": 5510
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.1385074319806613e-05,
+      "loss": 0.6394,
+      "step": 5511
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.138250270020059e-05,
+      "loss": 0.8335,
+      "step": 5512
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.137993108059456e-05,
+      "loss": 0.6463,
+      "step": 5513
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.1377359460988536e-05,
+      "loss": 0.8829,
+      "step": 5514
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.1374787841382506e-05,
+      "loss": 1.0009,
+      "step": 5515
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.1372216221776475e-05,
+      "loss": 0.747,
+      "step": 5516
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.136964460217045e-05,
+      "loss": 0.793,
+      "step": 5517
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.136707298256442e-05,
+      "loss": 0.707,
+      "step": 5518
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.136450136295839e-05,
+      "loss": 0.783,
+      "step": 5519
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.136192974335237e-05,
+      "loss": 0.8431,
+      "step": 5520
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.135935812374634e-05,
+      "loss": 0.6255,
+      "step": 5521
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.135678650414031e-05,
+      "loss": 0.7334,
+      "step": 5522
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.1354214884534284e-05,
+      "loss": 0.7556,
+      "step": 5523
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.135164326492825e-05,
+      "loss": 0.7317,
+      "step": 5524
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.134907164532222e-05,
+      "loss": 0.8537,
+      "step": 5525
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.13465000257162e-05,
+      "loss": 0.7702,
+      "step": 5526
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.134392840611017e-05,
+      "loss": 0.9135,
+      "step": 5527
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.1341356786504146e-05,
+      "loss": 0.652,
+      "step": 5528
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.1338785166898115e-05,
+      "loss": 0.724,
+      "step": 5529
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.1336213547292085e-05,
+      "loss": 0.8182,
+      "step": 5530
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.1333641927686055e-05,
+      "loss": 0.6231,
+      "step": 5531
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.133107030808003e-05,
+      "loss": 0.6469,
+      "step": 5532
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.132849868847401e-05,
+      "loss": 0.9703,
+      "step": 5533
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.132592706886798e-05,
+      "loss": 0.6673,
+      "step": 5534
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.132335544926195e-05,
+      "loss": 0.9683,
+      "step": 5535
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.1320783829655917e-05,
+      "loss": 0.6324,
+      "step": 5536
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.1318212210049886e-05,
+      "loss": 0.8589,
+      "step": 5537
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.131564059044386e-05,
+      "loss": 0.7149,
+      "step": 5538
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.131306897083784e-05,
+      "loss": 0.8904,
+      "step": 5539
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.131049735123181e-05,
+      "loss": 0.5731,
+      "step": 5540
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.130792573162578e-05,
+      "loss": 0.9565,
+      "step": 5541
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.130535411201975e-05,
+      "loss": 0.7622,
+      "step": 5542
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.1302782492413725e-05,
+      "loss": 1.2813,
+      "step": 5543
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.13002108728077e-05,
+      "loss": 0.74,
+      "step": 5544
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.129763925320167e-05,
+      "loss": 0.9954,
+      "step": 5545
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.129506763359564e-05,
+      "loss": 0.7422,
+      "step": 5546
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.129249601398961e-05,
+      "loss": 0.6867,
+      "step": 5547
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.128992439438359e-05,
+      "loss": 0.7988,
+      "step": 5548
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.1287352774777556e-05,
+      "loss": 0.8498,
+      "step": 5549
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.128478115517153e-05,
+      "loss": 0.732,
+      "step": 5550
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.12822095355655e-05,
+      "loss": 0.7174,
+      "step": 5551
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.127963791595947e-05,
+      "loss": 0.9532,
+      "step": 5552
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.127706629635344e-05,
+      "loss": 0.6795,
+      "step": 5553
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.127449467674742e-05,
+      "loss": 1.0195,
+      "step": 5554
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.127192305714139e-05,
+      "loss": 0.9389,
+      "step": 5555
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.1269351437535365e-05,
+      "loss": 1.0608,
+      "step": 5556
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.1266779817929334e-05,
+      "loss": 0.7128,
+      "step": 5557
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.1264208198323304e-05,
+      "loss": 0.777,
+      "step": 5558
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.126163657871728e-05,
+      "loss": 0.7145,
+      "step": 5559
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.125906495911125e-05,
+      "loss": 0.9742,
+      "step": 5560
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.125649333950522e-05,
+      "loss": 0.5442,
+      "step": 5561
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.1253921719899196e-05,
+      "loss": 0.9064,
+      "step": 5562
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.1251350100293166e-05,
+      "loss": 1.2121,
+      "step": 5563
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.124877848068714e-05,
+      "loss": 0.9234,
+      "step": 5564
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.124620686108111e-05,
+      "loss": 0.8756,
+      "step": 5565
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.124363524147508e-05,
+      "loss": 0.8803,
+      "step": 5566
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.124106362186905e-05,
+      "loss": 0.3292,
+      "step": 5567
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.123849200226303e-05,
+      "loss": 0.6031,
+      "step": 5568
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.1235920382657e-05,
+      "loss": 0.551,
+      "step": 5569
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.1233348763050974e-05,
+      "loss": 0.704,
+      "step": 5570
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.1230777143444944e-05,
+      "loss": 0.5394,
+      "step": 5571
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.122820552383891e-05,
+      "loss": 0.6244,
+      "step": 5572
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.122563390423288e-05,
+      "loss": 0.623,
+      "step": 5573
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.122306228462686e-05,
+      "loss": 0.6291,
+      "step": 5574
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.1220490665020836e-05,
+      "loss": 0.8072,
+      "step": 5575
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.1217919045414806e-05,
+      "loss": 0.848,
+      "step": 5576
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.1215347425808775e-05,
+      "loss": 0.691,
+      "step": 5577
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.1212775806202745e-05,
+      "loss": 0.7674,
+      "step": 5578
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.121020418659672e-05,
+      "loss": 0.6338,
+      "step": 5579
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.12076325669907e-05,
+      "loss": 0.6704,
+      "step": 5580
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.120506094738467e-05,
+      "loss": 0.9369,
+      "step": 5581
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.120248932777864e-05,
+      "loss": 0.7846,
+      "step": 5582
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.119991770817261e-05,
+      "loss": 0.828,
+      "step": 5583
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.119734608856658e-05,
+      "loss": 0.6848,
+      "step": 5584
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.119477446896055e-05,
+      "loss": 0.7121,
+      "step": 5585
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.119220284935453e-05,
+      "loss": 0.7741,
+      "step": 5586
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.11896312297485e-05,
+      "loss": 1.0047,
+      "step": 5587
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.118705961014247e-05,
+      "loss": 0.9971,
+      "step": 5588
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.118448799053644e-05,
+      "loss": 0.7674,
+      "step": 5589
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.1181916370930415e-05,
+      "loss": 0.7274,
+      "step": 5590
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.1179344751324385e-05,
+      "loss": 1.0334,
+      "step": 5591
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.117677313171836e-05,
+      "loss": 0.799,
+      "step": 5592
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.117420151211233e-05,
+      "loss": 1.1505,
+      "step": 5593
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.11716298925063e-05,
+      "loss": 0.659,
+      "step": 5594
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.116905827290028e-05,
+      "loss": 0.6201,
+      "step": 5595
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.116648665329425e-05,
+      "loss": 0.9277,
+      "step": 5596
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.1163915033688216e-05,
+      "loss": 0.7231,
+      "step": 5597
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.116134341408219e-05,
+      "loss": 0.7413,
+      "step": 5598
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.115877179447616e-05,
+      "loss": 0.5875,
+      "step": 5599
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.115620017487013e-05,
+      "loss": 0.6533,
+      "step": 5600
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.115362855526411e-05,
+      "loss": 0.8934,
+      "step": 5601
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.115105693565808e-05,
+      "loss": 0.9665,
+      "step": 5602
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.114848531605205e-05,
+      "loss": 0.7496,
+      "step": 5603
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.1145913696446025e-05,
+      "loss": 0.4635,
+      "step": 5604
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.1143342076839994e-05,
+      "loss": 0.6885,
+      "step": 5605
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.114077045723397e-05,
+      "loss": 0.5568,
+      "step": 5606
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.113819883762794e-05,
+      "loss": 0.6987,
+      "step": 5607
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.113562721802191e-05,
+      "loss": 0.4999,
+      "step": 5608
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.113305559841588e-05,
+      "loss": 0.8054,
+      "step": 5609
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.1130483978809856e-05,
+      "loss": 0.8037,
+      "step": 5610
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.112791235920383e-05,
+      "loss": 0.7271,
+      "step": 5611
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.11253407395978e-05,
+      "loss": 0.9443,
+      "step": 5612
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.112276911999177e-05,
+      "loss": 0.5808,
+      "step": 5613
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.112019750038574e-05,
+      "loss": 0.918,
+      "step": 5614
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.111762588077971e-05,
+      "loss": 0.7117,
+      "step": 5615
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.111505426117369e-05,
+      "loss": 0.6984,
+      "step": 5616
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.1112482641567664e-05,
+      "loss": 0.6868,
+      "step": 5617
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.1109911021961634e-05,
+      "loss": 1.0366,
+      "step": 5618
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.1107339402355604e-05,
+      "loss": 0.9131,
+      "step": 5619
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.1104767782749573e-05,
+      "loss": 0.8531,
+      "step": 5620
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.110219616314355e-05,
+      "loss": 0.8781,
+      "step": 5621
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.1099624543537526e-05,
+      "loss": 0.8257,
+      "step": 5622
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.1097052923931496e-05,
+      "loss": 0.9713,
+      "step": 5623
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.1094481304325466e-05,
+      "loss": 1.0947,
+      "step": 5624
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.1091909684719435e-05,
+      "loss": 0.6323,
+      "step": 5625
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.108933806511341e-05,
+      "loss": 0.6841,
+      "step": 5626
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.108676644550738e-05,
+      "loss": 0.7762,
+      "step": 5627
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.108419482590136e-05,
+      "loss": 0.7434,
+      "step": 5628
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.108162320629533e-05,
+      "loss": 0.6371,
+      "step": 5629
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.10790515866893e-05,
+      "loss": 1.0772,
+      "step": 5630
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.107647996708327e-05,
+      "loss": 0.8055,
+      "step": 5631
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.1073908347477244e-05,
+      "loss": 0.5632,
+      "step": 5632
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.107133672787122e-05,
+      "loss": 0.9429,
+      "step": 5633
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.106876510826519e-05,
+      "loss": 1.0447,
+      "step": 5634
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.106619348865916e-05,
+      "loss": 0.6217,
+      "step": 5635
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.106362186905313e-05,
+      "loss": 0.7651,
+      "step": 5636
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.1061050249447106e-05,
+      "loss": 0.6843,
+      "step": 5637
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.1058478629841075e-05,
+      "loss": 0.6375,
+      "step": 5638
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.105590701023505e-05,
+      "loss": 0.7542,
+      "step": 5639
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.105333539062902e-05,
+      "loss": 0.4277,
+      "step": 5640
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.105076377102299e-05,
+      "loss": 1.2993,
+      "step": 5641
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.104819215141697e-05,
+      "loss": 1.0804,
+      "step": 5642
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.104562053181094e-05,
+      "loss": 0.8829,
+      "step": 5643
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.104304891220491e-05,
+      "loss": 0.8481,
+      "step": 5644
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.104047729259888e-05,
+      "loss": 1.1197,
+      "step": 5645
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.103790567299285e-05,
+      "loss": 0.7152,
+      "step": 5646
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.103533405338682e-05,
+      "loss": 0.6397,
+      "step": 5647
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.10327624337808e-05,
+      "loss": 1.0085,
+      "step": 5648
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.103019081417477e-05,
+      "loss": 0.7278,
+      "step": 5649
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.102761919456874e-05,
+      "loss": 1.0309,
+      "step": 5650
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.1025047574962715e-05,
+      "loss": 0.6772,
+      "step": 5651
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.1022475955356685e-05,
+      "loss": 0.6072,
+      "step": 5652
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.101990433575066e-05,
+      "loss": 0.5871,
+      "step": 5653
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.101733271614463e-05,
+      "loss": 0.6679,
+      "step": 5654
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.10147610965386e-05,
+      "loss": 0.6556,
+      "step": 5655
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.101218947693257e-05,
+      "loss": 0.7827,
+      "step": 5656
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.100961785732655e-05,
+      "loss": 1.0751,
+      "step": 5657
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.100704623772052e-05,
+      "loss": 0.987,
+      "step": 5658
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.100447461811449e-05,
+      "loss": 0.6404,
+      "step": 5659
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.100190299850846e-05,
+      "loss": 1.1396,
+      "step": 5660
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.099933137890243e-05,
+      "loss": 0.6456,
+      "step": 5661
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.09967597592964e-05,
+      "loss": 0.817,
+      "step": 5662
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.099418813969038e-05,
+      "loss": 0.7047,
+      "step": 5663
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.0991616520084355e-05,
+      "loss": 0.7583,
+      "step": 5664
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.0989044900478324e-05,
+      "loss": 0.7,
+      "step": 5665
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.0986473280872294e-05,
+      "loss": 0.7539,
+      "step": 5666
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.0983901661266264e-05,
+      "loss": 0.6364,
+      "step": 5667
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.098133004166024e-05,
+      "loss": 0.7966,
+      "step": 5668
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.097875842205422e-05,
+      "loss": 0.72,
+      "step": 5669
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.0976186802448186e-05,
+      "loss": 1.3191,
+      "step": 5670
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.0973615182842156e-05,
+      "loss": 0.6624,
+      "step": 5671
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.0971043563236126e-05,
+      "loss": 0.9196,
+      "step": 5672
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.09684719436301e-05,
+      "loss": 0.8293,
+      "step": 5673
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.096590032402407e-05,
+      "loss": 0.965,
+      "step": 5674
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.096332870441805e-05,
+      "loss": 0.7948,
+      "step": 5675
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.096075708481202e-05,
+      "loss": 0.6907,
+      "step": 5676
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.095818546520599e-05,
+      "loss": 0.7639,
+      "step": 5677
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.095561384559996e-05,
+      "loss": 0.3987,
+      "step": 5678
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.0953042225993934e-05,
+      "loss": 0.7239,
+      "step": 5679
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.0950470606387904e-05,
+      "loss": 0.915,
+      "step": 5680
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.094789898678188e-05,
+      "loss": 0.5704,
+      "step": 5681
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.094532736717585e-05,
+      "loss": 0.8227,
+      "step": 5682
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.094275574756982e-05,
+      "loss": 0.9775,
+      "step": 5683
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.0940184127963796e-05,
+      "loss": 0.6075,
+      "step": 5684
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.0937612508357766e-05,
+      "loss": 1.1627,
+      "step": 5685
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.0935040888751735e-05,
+      "loss": 0.6203,
+      "step": 5686
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.093246926914571e-05,
+      "loss": 1.1891,
+      "step": 5687
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.092989764953968e-05,
+      "loss": 0.6758,
+      "step": 5688
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.092732602993366e-05,
+      "loss": 0.4284,
+      "step": 5689
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.092475441032763e-05,
+      "loss": 0.7826,
+      "step": 5690
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.09221827907216e-05,
+      "loss": 0.5421,
+      "step": 5691
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.091961117111557e-05,
+      "loss": 0.5815,
+      "step": 5692
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.0917039551509543e-05,
+      "loss": 0.5058,
+      "step": 5693
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.091446793190351e-05,
+      "loss": 0.6127,
+      "step": 5694
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.091189631229749e-05,
+      "loss": 0.9063,
+      "step": 5695
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.090932469269146e-05,
+      "loss": 0.6669,
+      "step": 5696
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.090675307308543e-05,
+      "loss": 0.6633,
+      "step": 5697
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.09041814534794e-05,
+      "loss": 0.7512,
+      "step": 5698
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.0901609833873375e-05,
+      "loss": 0.8996,
+      "step": 5699
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.089903821426735e-05,
+      "loss": 0.7919,
+      "step": 5700
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.089646659466132e-05,
+      "loss": 0.8458,
+      "step": 5701
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.089389497505529e-05,
+      "loss": 0.702,
+      "step": 5702
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.089132335544926e-05,
+      "loss": 0.6528,
+      "step": 5703
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.088875173584324e-05,
+      "loss": 0.7816,
+      "step": 5704
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.0886180116237214e-05,
+      "loss": 0.6348,
+      "step": 5705
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.088360849663118e-05,
+      "loss": 0.7052,
+      "step": 5706
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.088103687702515e-05,
+      "loss": 1.0779,
+      "step": 5707
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.087846525741912e-05,
+      "loss": 0.7021,
+      "step": 5708
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.087589363781309e-05,
+      "loss": 0.7499,
+      "step": 5709
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.087332201820707e-05,
+      "loss": 0.7808,
+      "step": 5710
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.0870750398601045e-05,
+      "loss": 0.6203,
+      "step": 5711
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.0868178778995015e-05,
+      "loss": 0.6604,
+      "step": 5712
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.0865607159388985e-05,
+      "loss": 0.8974,
+      "step": 5713
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.0863035539782954e-05,
+      "loss": 0.7393,
+      "step": 5714
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.086046392017693e-05,
+      "loss": 0.7139,
+      "step": 5715
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.08578923005709e-05,
+      "loss": 0.6993,
+      "step": 5716
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.085532068096488e-05,
+      "loss": 1.0068,
+      "step": 5717
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.0852749061358847e-05,
+      "loss": 0.7477,
+      "step": 5718
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.0850177441752816e-05,
+      "loss": 0.5328,
+      "step": 5719
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.084760582214679e-05,
+      "loss": 0.8818,
+      "step": 5720
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.084503420254076e-05,
+      "loss": 0.7918,
+      "step": 5721
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.084246258293473e-05,
+      "loss": 0.8372,
+      "step": 5722
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.083989096332871e-05,
+      "loss": 0.6685,
+      "step": 5723
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.083731934372268e-05,
+      "loss": 0.5118,
+      "step": 5724
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.083474772411665e-05,
+      "loss": 0.797,
+      "step": 5725
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.0832176104510624e-05,
+      "loss": 0.7598,
+      "step": 5726
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.0829604484904594e-05,
+      "loss": 0.7036,
+      "step": 5727
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.0827032865298564e-05,
+      "loss": 0.8029,
+      "step": 5728
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.082446124569254e-05,
+      "loss": 0.8233,
+      "step": 5729
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.082188962608651e-05,
+      "loss": 0.9743,
+      "step": 5730
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.0819318006480486e-05,
+      "loss": 0.9734,
+      "step": 5731
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.0816746386874456e-05,
+      "loss": 1.1636,
+      "step": 5732
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.0814174767268426e-05,
+      "loss": 0.9725,
+      "step": 5733
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.0811603147662395e-05,
+      "loss": 0.7955,
+      "step": 5734
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.080903152805637e-05,
+      "loss": 0.7485,
+      "step": 5735
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.080645990845035e-05,
+      "loss": 0.8759,
+      "step": 5736
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.080388828884432e-05,
+      "loss": 0.6671,
+      "step": 5737
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.080131666923829e-05,
+      "loss": 0.7131,
+      "step": 5738
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.079874504963226e-05,
+      "loss": 0.8893,
+      "step": 5739
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.079617343002623e-05,
+      "loss": 0.7947,
+      "step": 5740
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.0793601810420204e-05,
+      "loss": 0.8335,
+      "step": 5741
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.079103019081418e-05,
+      "loss": 0.6503,
+      "step": 5742
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.078845857120815e-05,
+      "loss": 0.7455,
+      "step": 5743
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.078588695160212e-05,
+      "loss": 0.5569,
+      "step": 5744
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.078331533199609e-05,
+      "loss": 0.9628,
+      "step": 5745
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.0780743712390065e-05,
+      "loss": 1.1442,
+      "step": 5746
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.077817209278404e-05,
+      "loss": 0.9163,
+      "step": 5747
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.077560047317801e-05,
+      "loss": 0.7781,
+      "step": 5748
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.077302885357198e-05,
+      "loss": 0.9764,
+      "step": 5749
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.077045723396595e-05,
+      "loss": 0.8438,
+      "step": 5750
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.076788561435993e-05,
+      "loss": 0.9784,
+      "step": 5751
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.07653139947539e-05,
+      "loss": 1.0408,
+      "step": 5752
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.0762742375147874e-05,
+      "loss": 0.7939,
+      "step": 5753
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.076017075554184e-05,
+      "loss": 0.8995,
+      "step": 5754
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.075759913593581e-05,
+      "loss": 0.7894,
+      "step": 5755
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.075502751632978e-05,
+      "loss": 0.6291,
+      "step": 5756
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.075245589672376e-05,
+      "loss": 0.7338,
+      "step": 5757
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.074988427711773e-05,
+      "loss": 1.087,
+      "step": 5758
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.0747312657511705e-05,
+      "loss": 0.8967,
+      "step": 5759
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.0744741037905675e-05,
+      "loss": 0.622,
+      "step": 5760
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.0742169418299645e-05,
+      "loss": 0.558,
+      "step": 5761
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.073959779869362e-05,
+      "loss": 0.5115,
+      "step": 5762
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.073702617908759e-05,
+      "loss": 0.4834,
+      "step": 5763
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.073445455948156e-05,
+      "loss": 1.0553,
+      "step": 5764
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.073188293987554e-05,
+      "loss": 0.6379,
+      "step": 5765
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.072931132026951e-05,
+      "loss": 0.6143,
+      "step": 5766
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.072673970066348e-05,
+      "loss": 0.8584,
+      "step": 5767
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.072416808105745e-05,
+      "loss": 0.3824,
+      "step": 5768
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.072159646145142e-05,
+      "loss": 0.5033,
+      "step": 5769
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.071902484184539e-05,
+      "loss": 0.9154,
+      "step": 5770
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.071645322223937e-05,
+      "loss": 0.8912,
+      "step": 5771
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.071388160263334e-05,
+      "loss": 0.7623,
+      "step": 5772
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.0711309983027315e-05,
+      "loss": 0.5089,
+      "step": 5773
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.0708738363421284e-05,
+      "loss": 0.6255,
+      "step": 5774
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.0706166743815254e-05,
+      "loss": 0.9246,
+      "step": 5775
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.0703595124209224e-05,
+      "loss": 0.7652,
+      "step": 5776
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.07010235046032e-05,
+      "loss": 0.5728,
+      "step": 5777
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.069845188499718e-05,
+      "loss": 0.9443,
+      "step": 5778
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.0695880265391146e-05,
+      "loss": 1.0416,
+      "step": 5779
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.0693308645785116e-05,
+      "loss": 0.7776,
+      "step": 5780
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.0690737026179086e-05,
+      "loss": 0.9027,
+      "step": 5781
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.068816540657306e-05,
+      "loss": 0.959,
+      "step": 5782
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.068559378696704e-05,
+      "loss": 0.6432,
+      "step": 5783
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.068302216736101e-05,
+      "loss": 1.0787,
+      "step": 5784
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.068045054775498e-05,
+      "loss": 0.5215,
+      "step": 5785
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.067787892814895e-05,
+      "loss": 0.6065,
+      "step": 5786
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.067530730854292e-05,
+      "loss": 0.7569,
+      "step": 5787
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.0672735688936894e-05,
+      "loss": 0.8169,
+      "step": 5788
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.067016406933087e-05,
+      "loss": 1.0143,
+      "step": 5789
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.066759244972484e-05,
+      "loss": 0.7774,
+      "step": 5790
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.066502083011881e-05,
+      "loss": 0.8923,
+      "step": 5791
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.066244921051278e-05,
+      "loss": 1.1026,
+      "step": 5792
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.0659877590906756e-05,
+      "loss": 0.8318,
+      "step": 5793
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.065730597130073e-05,
+      "loss": 0.8032,
+      "step": 5794
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.06547343516947e-05,
+      "loss": 0.809,
+      "step": 5795
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.065216273208867e-05,
+      "loss": 0.6109,
+      "step": 5796
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.064959111248264e-05,
+      "loss": 0.8087,
+      "step": 5797
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.064701949287662e-05,
+      "loss": 0.76,
+      "step": 5798
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.064444787327059e-05,
+      "loss": 0.9523,
+      "step": 5799
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.0641876253664564e-05,
+      "loss": 0.8391,
+      "step": 5800
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.0639304634058534e-05,
+      "loss": 0.5052,
+      "step": 5801
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.0636733014452503e-05,
+      "loss": 0.6549,
+      "step": 5802
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.063416139484647e-05,
+      "loss": 0.5477,
+      "step": 5803
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.063158977524045e-05,
+      "loss": 0.5592,
+      "step": 5804
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.062901815563442e-05,
+      "loss": 1.1897,
+      "step": 5805
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.0626446536028396e-05,
+      "loss": 1.1219,
+      "step": 5806
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.0623874916422365e-05,
+      "loss": 0.6703,
+      "step": 5807
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.0621303296816335e-05,
+      "loss": 0.841,
+      "step": 5808
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.061873167721031e-05,
+      "loss": 0.9214,
+      "step": 5809
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.061616005760428e-05,
+      "loss": 0.6237,
+      "step": 5810
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.061358843799825e-05,
+      "loss": 0.8485,
+      "step": 5811
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.061101681839223e-05,
+      "loss": 0.5331,
+      "step": 5812
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.06084451987862e-05,
+      "loss": 1.0341,
+      "step": 5813
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.0605873579180174e-05,
+      "loss": 0.5887,
+      "step": 5814
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.060330195957414e-05,
+      "loss": 0.8534,
+      "step": 5815
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.060073033996811e-05,
+      "loss": 0.9407,
+      "step": 5816
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.059815872036208e-05,
+      "loss": 0.7156,
+      "step": 5817
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.059558710075606e-05,
+      "loss": 0.7696,
+      "step": 5818
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.059301548115003e-05,
+      "loss": 0.6911,
+      "step": 5819
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.0590443861544005e-05,
+      "loss": 1.2466,
+      "step": 5820
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.0587872241937975e-05,
+      "loss": 0.7549,
+      "step": 5821
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.0585300622331945e-05,
+      "loss": 0.832,
+      "step": 5822
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.0582729002725914e-05,
+      "loss": 0.6622,
+      "step": 5823
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.058015738311989e-05,
+      "loss": 0.7059,
+      "step": 5824
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.057758576351387e-05,
+      "loss": 0.5719,
+      "step": 5825
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.057501414390784e-05,
+      "loss": 0.6226,
+      "step": 5826
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.0572442524301807e-05,
+      "loss": 0.7577,
+      "step": 5827
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.0569870904695776e-05,
+      "loss": 0.6652,
+      "step": 5828
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.056729928508975e-05,
+      "loss": 0.7182,
+      "step": 5829
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.056472766548373e-05,
+      "loss": 1.1524,
+      "step": 5830
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.05621560458777e-05,
+      "loss": 1.0901,
+      "step": 5831
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.055958442627167e-05,
+      "loss": 0.6656,
+      "step": 5832
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.055701280666564e-05,
+      "loss": 0.7428,
+      "step": 5833
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.055444118705961e-05,
+      "loss": 0.4828,
+      "step": 5834
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.0551869567453584e-05,
+      "loss": 0.5828,
+      "step": 5835
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.054929794784756e-05,
+      "loss": 1.0363,
+      "step": 5836
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.054672632824153e-05,
+      "loss": 0.7054,
+      "step": 5837
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.05441547086355e-05,
+      "loss": 1.0343,
+      "step": 5838
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.054158308902947e-05,
+      "loss": 0.6538,
+      "step": 5839
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.0539011469423446e-05,
+      "loss": 0.7525,
+      "step": 5840
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.0536439849817416e-05,
+      "loss": 0.7144,
+      "step": 5841
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.053386823021139e-05,
+      "loss": 0.5788,
+      "step": 5842
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.053129661060536e-05,
+      "loss": 0.7859,
+      "step": 5843
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.052872499099933e-05,
+      "loss": 0.9926,
+      "step": 5844
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.052615337139331e-05,
+      "loss": 0.9527,
+      "step": 5845
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.052358175178728e-05,
+      "loss": 1.1917,
+      "step": 5846
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.052101013218125e-05,
+      "loss": 0.7968,
+      "step": 5847
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.0518438512575224e-05,
+      "loss": 0.8296,
+      "step": 5848
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.0515866892969194e-05,
+      "loss": 0.8695,
+      "step": 5849
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.0513295273363163e-05,
+      "loss": 1.1658,
+      "step": 5850
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.051072365375714e-05,
+      "loss": 0.6064,
+      "step": 5851
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.050815203415111e-05,
+      "loss": 1.105,
+      "step": 5852
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.050558041454508e-05,
+      "loss": 0.9165,
+      "step": 5853
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.0503008794939056e-05,
+      "loss": 0.5985,
+      "step": 5854
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.0500437175333025e-05,
+      "loss": 0.7926,
+      "step": 5855
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.0497865555727e-05,
+      "loss": 0.6368,
+      "step": 5856
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.049529393612097e-05,
+      "loss": 0.8991,
+      "step": 5857
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.049272231651494e-05,
+      "loss": 0.9304,
+      "step": 5858
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.049015069690891e-05,
+      "loss": 0.8399,
+      "step": 5859
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.048757907730289e-05,
+      "loss": 0.8851,
+      "step": 5860
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.0485007457696864e-05,
+      "loss": 0.9346,
+      "step": 5861
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.0482435838090834e-05,
+      "loss": 0.8545,
+      "step": 5862
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.04798642184848e-05,
+      "loss": 0.6115,
+      "step": 5863
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.047729259887877e-05,
+      "loss": 0.9014,
+      "step": 5864
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.047472097927274e-05,
+      "loss": 0.8345,
+      "step": 5865
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.047214935966672e-05,
+      "loss": 0.955,
+      "step": 5866
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.0469577740060696e-05,
+      "loss": 1.1685,
+      "step": 5867
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.0467006120454665e-05,
+      "loss": 0.8272,
+      "step": 5868
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.0464434500848635e-05,
+      "loss": 0.9444,
+      "step": 5869
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.0461862881242605e-05,
+      "loss": 0.8074,
+      "step": 5870
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.045929126163658e-05,
+      "loss": 0.6114,
+      "step": 5871
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.045671964203056e-05,
+      "loss": 0.833,
+      "step": 5872
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.045414802242453e-05,
+      "loss": 0.5856,
+      "step": 5873
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.04515764028185e-05,
+      "loss": 0.7717,
+      "step": 5874
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.0449004783212467e-05,
+      "loss": 0.6754,
+      "step": 5875
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.044643316360644e-05,
+      "loss": 0.8382,
+      "step": 5876
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.044386154400041e-05,
+      "loss": 0.714,
+      "step": 5877
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.044128992439439e-05,
+      "loss": 0.9253,
+      "step": 5878
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.043871830478836e-05,
+      "loss": 0.8874,
+      "step": 5879
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.043614668518233e-05,
+      "loss": 0.9288,
+      "step": 5880
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.04335750655763e-05,
+      "loss": 0.7195,
+      "step": 5881
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.0431003445970275e-05,
+      "loss": 0.8353,
+      "step": 5882
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.0428431826364244e-05,
+      "loss": 0.9827,
+      "step": 5883
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.042586020675822e-05,
+      "loss": 0.8917,
+      "step": 5884
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.042328858715219e-05,
+      "loss": 0.9059,
+      "step": 5885
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.042071696754616e-05,
+      "loss": 0.7387,
+      "step": 5886
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.041814534794014e-05,
+      "loss": 0.9612,
+      "step": 5887
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.0415573728334106e-05,
+      "loss": 0.7745,
+      "step": 5888
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.0413002108728076e-05,
+      "loss": 0.8336,
+      "step": 5889
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.041043048912205e-05,
+      "loss": 1.2449,
+      "step": 5890
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.040785886951602e-05,
+      "loss": 0.8937,
+      "step": 5891
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.040528724991e-05,
+      "loss": 0.7908,
+      "step": 5892
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.040271563030397e-05,
+      "loss": 0.5393,
+      "step": 5893
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.040014401069794e-05,
+      "loss": 0.861,
+      "step": 5894
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.039757239109191e-05,
+      "loss": 0.9631,
+      "step": 5895
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.0395000771485884e-05,
+      "loss": 1.0496,
+      "step": 5896
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.0392429151879854e-05,
+      "loss": 0.6339,
+      "step": 5897
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.038985753227383e-05,
+      "loss": 1.0248,
+      "step": 5898
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.03872859126678e-05,
+      "loss": 0.846,
+      "step": 5899
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.038471429306177e-05,
+      "loss": 0.7911,
+      "step": 5900
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.038214267345574e-05,
+      "loss": 0.7401,
+      "step": 5901
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.0379571053849716e-05,
+      "loss": 0.7995,
+      "step": 5902
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.037699943424369e-05,
+      "loss": 0.6706,
+      "step": 5903
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.037442781463766e-05,
+      "loss": 0.9732,
+      "step": 5904
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.037185619503163e-05,
+      "loss": 1.0058,
+      "step": 5905
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.03692845754256e-05,
+      "loss": 0.9013,
+      "step": 5906
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.036671295581957e-05,
+      "loss": 0.7954,
+      "step": 5907
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.0364141336213554e-05,
+      "loss": 0.8276,
+      "step": 5908
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.0361569716607524e-05,
+      "loss": 0.981,
+      "step": 5909
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.0358998097001494e-05,
+      "loss": 0.6738,
+      "step": 5910
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.035642647739546e-05,
+      "loss": 0.9942,
+      "step": 5911
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.035385485778943e-05,
+      "loss": 0.549,
+      "step": 5912
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.035128323818341e-05,
+      "loss": 0.8283,
+      "step": 5913
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.0348711618577386e-05,
+      "loss": 0.8503,
+      "step": 5914
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.0346139998971356e-05,
+      "loss": 1.0127,
+      "step": 5915
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.0343568379365325e-05,
+      "loss": 0.7567,
+      "step": 5916
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.0340996759759295e-05,
+      "loss": 0.8292,
+      "step": 5917
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.033842514015327e-05,
+      "loss": 0.8003,
+      "step": 5918
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.033585352054724e-05,
+      "loss": 0.7379,
+      "step": 5919
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.033328190094122e-05,
+      "loss": 0.5138,
+      "step": 5920
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.033071028133519e-05,
+      "loss": 0.8104,
+      "step": 5921
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.032813866172916e-05,
+      "loss": 0.8421,
+      "step": 5922
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.0325567042123133e-05,
+      "loss": 0.9614,
+      "step": 5923
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.03229954225171e-05,
+      "loss": 0.6251,
+      "step": 5924
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.032042380291107e-05,
+      "loss": 0.8873,
+      "step": 5925
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.031785218330505e-05,
+      "loss": 0.9834,
+      "step": 5926
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.031528056369902e-05,
+      "loss": 0.8351,
+      "step": 5927
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.031270894409299e-05,
+      "loss": 0.4682,
+      "step": 5928
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.0310137324486965e-05,
+      "loss": 1.0294,
+      "step": 5929
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.0307565704880935e-05,
+      "loss": 0.9182,
+      "step": 5930
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.030499408527491e-05,
+      "loss": 0.7632,
+      "step": 5931
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.030242246566888e-05,
+      "loss": 0.8796,
+      "step": 5932
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.029985084606285e-05,
+      "loss": 0.7889,
+      "step": 5933
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.029727922645683e-05,
+      "loss": 1.1037,
+      "step": 5934
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.02947076068508e-05,
+      "loss": 0.472,
+      "step": 5935
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.0292135987244766e-05,
+      "loss": 0.8362,
+      "step": 5936
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.028956436763874e-05,
+      "loss": 0.4794,
+      "step": 5937
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.028699274803271e-05,
+      "loss": 0.8586,
+      "step": 5938
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.028442112842669e-05,
+      "loss": 0.9286,
+      "step": 5939
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.028184950882066e-05,
+      "loss": 1.0137,
+      "step": 5940
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.027927788921463e-05,
+      "loss": 0.7309,
+      "step": 5941
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.02767062696086e-05,
+      "loss": 0.7236,
+      "step": 5942
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.0274134650002575e-05,
+      "loss": 0.8453,
+      "step": 5943
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.0271563030396544e-05,
+      "loss": 0.9306,
+      "step": 5944
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.026899141079052e-05,
+      "loss": 0.5899,
+      "step": 5945
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.026641979118449e-05,
+      "loss": 0.6956,
+      "step": 5946
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.026384817157846e-05,
+      "loss": 0.7394,
+      "step": 5947
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.026127655197243e-05,
+      "loss": 0.9167,
+      "step": 5948
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.0258704932366406e-05,
+      "loss": 0.6704,
+      "step": 5949
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.025613331276038e-05,
+      "loss": 0.814,
+      "step": 5950
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.025356169315435e-05,
+      "loss": 0.9129,
+      "step": 5951
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.025099007354832e-05,
+      "loss": 0.8847,
+      "step": 5952
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.024841845394229e-05,
+      "loss": 0.9001,
+      "step": 5953
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.024584683433627e-05,
+      "loss": 0.788,
+      "step": 5954
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.0243275214730245e-05,
+      "loss": 0.694,
+      "step": 5955
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.0240703595124214e-05,
+      "loss": 0.7869,
+      "step": 5956
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.0238131975518184e-05,
+      "loss": 0.9929,
+      "step": 5957
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.0235560355912154e-05,
+      "loss": 0.9953,
+      "step": 5958
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.0232988736306123e-05,
+      "loss": 0.5597,
+      "step": 5959
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.02304171167001e-05,
+      "loss": 0.6001,
+      "step": 5960
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.0227845497094076e-05,
+      "loss": 0.6916,
+      "step": 5961
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.0225273877488046e-05,
+      "loss": 0.7715,
+      "step": 5962
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.0222702257882016e-05,
+      "loss": 0.7843,
+      "step": 5963
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.0220130638275985e-05,
+      "loss": 0.6735,
+      "step": 5964
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.021755901866996e-05,
+      "loss": 0.5646,
+      "step": 5965
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.021498739906393e-05,
+      "loss": 0.7219,
+      "step": 5966
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.021241577945791e-05,
+      "loss": 0.789,
+      "step": 5967
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.020984415985188e-05,
+      "loss": 0.6112,
+      "step": 5968
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.020727254024585e-05,
+      "loss": 0.9653,
+      "step": 5969
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.0204700920639824e-05,
+      "loss": 0.78,
+      "step": 5970
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.0202129301033794e-05,
+      "loss": 0.7038,
+      "step": 5971
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.019955768142776e-05,
+      "loss": 0.7512,
+      "step": 5972
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.019698606182174e-05,
+      "loss": 1.3302,
+      "step": 5973
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.019441444221571e-05,
+      "loss": 0.4532,
+      "step": 5974
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.019184282260968e-05,
+      "loss": 0.8636,
+      "step": 5975
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.0189271203003656e-05,
+      "loss": 0.8615,
+      "step": 5976
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.0186699583397625e-05,
+      "loss": 0.6376,
+      "step": 5977
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.0184127963791595e-05,
+      "loss": 0.9686,
+      "step": 5978
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.018155634418557e-05,
+      "loss": 1.0767,
+      "step": 5979
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.017898472457954e-05,
+      "loss": 0.5433,
+      "step": 5980
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.017641310497352e-05,
+      "loss": 0.9319,
+      "step": 5981
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.017384148536749e-05,
+      "loss": 0.6696,
+      "step": 5982
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.017126986576146e-05,
+      "loss": 0.6702,
+      "step": 5983
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.0168698246155427e-05,
+      "loss": 1.1131,
+      "step": 5984
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.01661266265494e-05,
+      "loss": 0.7967,
+      "step": 5985
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.016355500694338e-05,
+      "loss": 0.5708,
+      "step": 5986
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.016098338733735e-05,
+      "loss": 0.8435,
+      "step": 5987
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.015841176773132e-05,
+      "loss": 0.8369,
+      "step": 5988
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.015584014812529e-05,
+      "loss": 0.6304,
+      "step": 5989
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.015326852851926e-05,
+      "loss": 0.6554,
+      "step": 5990
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.0150696908913235e-05,
+      "loss": 1.0846,
+      "step": 5991
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.014812528930721e-05,
+      "loss": 1.1062,
+      "step": 5992
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.014555366970118e-05,
+      "loss": 0.9069,
+      "step": 5993
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.014298205009515e-05,
+      "loss": 0.6709,
+      "step": 5994
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.014041043048912e-05,
+      "loss": 0.7098,
+      "step": 5995
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.01378388108831e-05,
+      "loss": 0.6409,
+      "step": 5996
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.013526719127707e-05,
+      "loss": 0.6121,
+      "step": 5997
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.013269557167104e-05,
+      "loss": 0.7162,
+      "step": 5998
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.013012395206501e-05,
+      "loss": 0.9464,
+      "step": 5999
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.012755233245898e-05,
+      "loss": 1.0341,
+      "step": 6000
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.012498071285296e-05,
+      "loss": 0.8107,
+      "step": 6001
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.012240909324693e-05,
+      "loss": 0.857,
+      "step": 6002
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.0119837473640905e-05,
+      "loss": 0.7936,
+      "step": 6003
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.0117265854034874e-05,
+      "loss": 0.5778,
+      "step": 6004
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.0114694234428844e-05,
+      "loss": 0.6405,
+      "step": 6005
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.0112122614822814e-05,
+      "loss": 1.1133,
+      "step": 6006
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.010955099521679e-05,
+      "loss": 0.8125,
+      "step": 6007
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.010697937561076e-05,
+      "loss": 0.646,
+      "step": 6008
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.0104407756004736e-05,
+      "loss": 1.0046,
+      "step": 6009
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.0101836136398706e-05,
+      "loss": 0.7876,
+      "step": 6010
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.0099264516792676e-05,
+      "loss": 0.7624,
+      "step": 6011
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.009669289718665e-05,
+      "loss": 0.5935,
+      "step": 6012
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.009412127758062e-05,
+      "loss": 1.1071,
+      "step": 6013
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.009154965797459e-05,
+      "loss": 0.5648,
+      "step": 6014
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.008897803836857e-05,
+      "loss": 0.5805,
+      "step": 6015
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.008640641876254e-05,
+      "loss": 0.8027,
+      "step": 6016
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.0083834799156514e-05,
+      "loss": 1.0026,
+      "step": 6017
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.0081263179550484e-05,
+      "loss": 0.9415,
+      "step": 6018
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.0078691559944454e-05,
+      "loss": 0.907,
+      "step": 6019
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.007611994033842e-05,
+      "loss": 0.3992,
+      "step": 6020
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.00735483207324e-05,
+      "loss": 0.8553,
+      "step": 6021
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.007097670112637e-05,
+      "loss": 0.9651,
+      "step": 6022
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.0068405081520346e-05,
+      "loss": 0.694,
+      "step": 6023
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.0065833461914316e-05,
+      "loss": 0.698,
+      "step": 6024
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.0063261842308285e-05,
+      "loss": 0.682,
+      "step": 6025
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.0060690222702255e-05,
+      "loss": 0.9465,
+      "step": 6026
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.005811860309623e-05,
+      "loss": 0.6988,
+      "step": 6027
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.005554698349021e-05,
+      "loss": 0.584,
+      "step": 6028
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.005297536388418e-05,
+      "loss": 0.7472,
+      "step": 6029
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.005040374427815e-05,
+      "loss": 0.5746,
+      "step": 6030
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.004783212467212e-05,
+      "loss": 0.6955,
+      "step": 6031
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.0045260505066093e-05,
+      "loss": 0.916,
+      "step": 6032
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.004268888546007e-05,
+      "loss": 0.6395,
+      "step": 6033
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.004011726585404e-05,
+      "loss": 0.7085,
+      "step": 6034
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.003754564624801e-05,
+      "loss": 0.5765,
+      "step": 6035
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.003497402664198e-05,
+      "loss": 0.7716,
+      "step": 6036
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.003240240703595e-05,
+      "loss": 0.8231,
+      "step": 6037
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.0029830787429925e-05,
+      "loss": 1.1924,
+      "step": 6038
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.00272591678239e-05,
+      "loss": 0.8045,
+      "step": 6039
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.002468754821787e-05,
+      "loss": 0.7358,
+      "step": 6040
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.002211592861184e-05,
+      "loss": 0.8256,
+      "step": 6041
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.001954430900581e-05,
+      "loss": 1.0764,
+      "step": 6042
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.001697268939979e-05,
+      "loss": 1.0612,
+      "step": 6043
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.001440106979376e-05,
+      "loss": 0.633,
+      "step": 6044
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.001182945018773e-05,
+      "loss": 0.7492,
+      "step": 6045
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.00092578305817e-05,
+      "loss": 0.4853,
+      "step": 6046
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.000668621097567e-05,
+      "loss": 0.9765,
+      "step": 6047
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.000411459136965e-05,
+      "loss": 0.9943,
+      "step": 6048
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.000154297176362e-05,
+      "loss": 0.7079,
+      "step": 6049
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 3.999897135215759e-05,
+      "loss": 0.8077,
+      "step": 6050
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 3.9996399732551565e-05,
+      "loss": 0.6906,
+      "step": 6051
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 3.9993828112945535e-05,
+      "loss": 0.6269,
+      "step": 6052
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 3.9991256493339504e-05,
+      "loss": 0.8595,
+      "step": 6053
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 3.998868487373348e-05,
+      "loss": 0.7855,
+      "step": 6054
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 3.998611325412745e-05,
+      "loss": 0.5705,
+      "step": 6055
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 3.998354163452142e-05,
+      "loss": 0.8816,
+      "step": 6056
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 3.9980970014915397e-05,
+      "loss": 0.88,
+      "step": 6057
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 3.9978398395309366e-05,
+      "loss": 0.9452,
+      "step": 6058
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 3.997582677570334e-05,
+      "loss": 0.9727,
+      "step": 6059
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 3.997325515609731e-05,
+      "loss": 0.7436,
+      "step": 6060
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 3.997068353649128e-05,
+      "loss": 0.7367,
+      "step": 6061
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 3.996811191688525e-05,
+      "loss": 0.7767,
+      "step": 6062
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 3.996554029727923e-05,
+      "loss": 0.6838,
+      "step": 6063
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 3.9962968677673205e-05,
+      "loss": 0.858,
+      "step": 6064
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 3.9960397058067174e-05,
+      "loss": 0.9598,
+      "step": 6065
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 3.9957825438461144e-05,
+      "loss": 0.9649,
+      "step": 6066
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 3.9955253818855114e-05,
+      "loss": 1.155,
+      "step": 6067
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 3.9952682199249083e-05,
+      "loss": 0.8507,
+      "step": 6068
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 3.995011057964306e-05,
+      "loss": 0.8197,
+      "step": 6069
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 3.9947538960037036e-05,
+      "loss": 0.6194,
+      "step": 6070
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 3.9944967340431006e-05,
+      "loss": 0.589,
+      "step": 6071
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 3.9942395720824976e-05,
+      "loss": 1.0981,
+      "step": 6072
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 3.9939824101218945e-05,
+      "loss": 0.928,
+      "step": 6073
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 3.993725248161292e-05,
+      "loss": 0.7186,
+      "step": 6074
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 3.99346808620069e-05,
+      "loss": 0.6993,
+      "step": 6075
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 3.993210924240087e-05,
+      "loss": 0.8558,
+      "step": 6076
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 3.992953762279484e-05,
+      "loss": 0.97,
+      "step": 6077
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 3.992696600318881e-05,
+      "loss": 0.7239,
+      "step": 6078
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 3.9924394383582784e-05,
+      "loss": 1.1337,
+      "step": 6079
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 3.992182276397676e-05,
+      "loss": 0.9027,
+      "step": 6080
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 3.991925114437073e-05,
+      "loss": 0.6481,
+      "step": 6081
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 3.99166795247647e-05,
+      "loss": 0.7618,
+      "step": 6082
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 3.991410790515867e-05,
+      "loss": 0.9496,
+      "step": 6083
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 3.991153628555264e-05,
+      "loss": 0.8495,
+      "step": 6084
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 3.9908964665946616e-05,
+      "loss": 0.6754,
+      "step": 6085
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 3.990639304634059e-05,
+      "loss": 0.7229,
+      "step": 6086
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 3.990382142673456e-05,
+      "loss": 0.6513,
+      "step": 6087
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 3.990124980712853e-05,
+      "loss": 0.8084,
+      "step": 6088
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 3.98986781875225e-05,
+      "loss": 0.9789,
+      "step": 6089
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 3.989610656791648e-05,
+      "loss": 1.0855,
+      "step": 6090
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 3.989353494831045e-05,
+      "loss": 0.7031,
+      "step": 6091
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 3.9890963328704424e-05,
+      "loss": 0.7197,
+      "step": 6092
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 3.988839170909839e-05,
+      "loss": 0.8551,
+      "step": 6093
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 3.988582008949236e-05,
+      "loss": 0.6018,
+      "step": 6094
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 3.988324846988634e-05,
+      "loss": 0.7317,
+      "step": 6095
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 3.988067685028031e-05,
+      "loss": 0.5666,
+      "step": 6096
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 3.987810523067428e-05,
+      "loss": 0.8265,
+      "step": 6097
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 3.9875533611068255e-05,
+      "loss": 0.7696,
+      "step": 6098
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 3.9872961991462225e-05,
+      "loss": 0.6039,
+      "step": 6099
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 3.9870390371856195e-05,
+      "loss": 1.0819,
+      "step": 6100
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 3.986781875225017e-05,
+      "loss": 0.9485,
+      "step": 6101
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 3.986524713264414e-05,
+      "loss": 0.5055,
+      "step": 6102
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 3.986267551303811e-05,
+      "loss": 0.6439,
+      "step": 6103
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 3.986010389343209e-05,
+      "loss": 0.8817,
+      "step": 6104
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 3.985753227382606e-05,
+      "loss": 0.9788,
+      "step": 6105
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 3.985496065422003e-05,
+      "loss": 0.8646,
+      "step": 6106
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 3.9852389034614e-05,
+      "loss": 0.8118,
+      "step": 6107
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 3.984981741500797e-05,
+      "loss": 0.642,
+      "step": 6108
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 3.984724579540194e-05,
+      "loss": 0.852,
+      "step": 6109
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 3.984467417579592e-05,
+      "loss": 0.7439,
+      "step": 6110
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 3.9842102556189895e-05,
+      "loss": 0.9216,
+      "step": 6111
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 3.9839530936583865e-05,
+      "loss": 0.7781,
+      "step": 6112
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 3.9836959316977834e-05,
+      "loss": 0.8172,
+      "step": 6113
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 3.9834387697371804e-05,
+      "loss": 0.9091,
+      "step": 6114
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 3.9831816077765774e-05,
+      "loss": 0.6751,
+      "step": 6115
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 3.982924445815975e-05,
+      "loss": 0.7263,
+      "step": 6116
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 3.982667283855373e-05,
+      "loss": 0.9416,
+      "step": 6117
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 3.9824101218947696e-05,
+      "loss": 0.926,
+      "step": 6118
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 3.9821529599341666e-05,
+      "loss": 1.006,
+      "step": 6119
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 3.9818957979735636e-05,
+      "loss": 0.5925,
+      "step": 6120
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 3.981638636012961e-05,
+      "loss": 0.8382,
+      "step": 6121
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 3.981381474052359e-05,
+      "loss": 0.6843,
+      "step": 6122
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 3.981124312091756e-05,
+      "loss": 0.9806,
+      "step": 6123
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 3.980867150131153e-05,
+      "loss": 0.9476,
+      "step": 6124
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 3.98060998817055e-05,
+      "loss": 0.748,
+      "step": 6125
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 3.9803528262099474e-05,
+      "loss": 0.6729,
+      "step": 6126
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 3.9800956642493444e-05,
+      "loss": 0.9318,
+      "step": 6127
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 3.979838502288742e-05,
+      "loss": 0.8311,
+      "step": 6128
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 3.979581340328139e-05,
+      "loss": 0.7963,
+      "step": 6129
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 3.979324178367536e-05,
+      "loss": 0.845,
+      "step": 6130
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 3.979067016406933e-05,
+      "loss": 0.8324,
+      "step": 6131
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 3.9788098544463306e-05,
+      "loss": 1.0189,
+      "step": 6132
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 3.9785526924857276e-05,
+      "loss": 0.808,
+      "step": 6133
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 3.978295530525125e-05,
+      "loss": 0.5965,
+      "step": 6134
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 3.978038368564522e-05,
+      "loss": 0.621,
+      "step": 6135
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 3.977781206603919e-05,
+      "loss": 0.8474,
+      "step": 6136
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 3.977524044643317e-05,
+      "loss": 0.9724,
+      "step": 6137
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 3.977266882682714e-05,
+      "loss": 0.7615,
+      "step": 6138
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 3.977009720722111e-05,
+      "loss": 0.5248,
+      "step": 6139
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 3.9767525587615084e-05,
+      "loss": 0.7934,
+      "step": 6140
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 3.9764953968009053e-05,
+      "loss": 0.8488,
+      "step": 6141
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 3.976238234840303e-05,
+      "loss": 0.4396,
+      "step": 6142
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 3.9759810728797e-05,
+      "loss": 0.8036,
+      "step": 6143
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 3.975723910919097e-05,
+      "loss": 0.905,
+      "step": 6144
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 3.975466748958494e-05,
+      "loss": 0.5743,
+      "step": 6145
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 3.9752095869978915e-05,
+      "loss": 1.0029,
+      "step": 6146
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 3.9749524250372885e-05,
+      "loss": 1.1485,
+      "step": 6147
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 3.974695263076686e-05,
+      "loss": 0.9104,
+      "step": 6148
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 3.974438101116083e-05,
+      "loss": 0.7137,
+      "step": 6149
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 3.97418093915548e-05,
+      "loss": 0.937,
+      "step": 6150
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 3.973923777194877e-05,
+      "loss": 0.8959,
+      "step": 6151
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 3.973666615234275e-05,
+      "loss": 0.7011,
+      "step": 6152
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 3.9734094532736724e-05,
+      "loss": 0.4804,
+      "step": 6153
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 3.973152291313069e-05,
+      "loss": 0.4938,
+      "step": 6154
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 3.972895129352466e-05,
+      "loss": 0.8042,
+      "step": 6155
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 3.972637967391863e-05,
+      "loss": 0.6011,
+      "step": 6156
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 3.972380805431261e-05,
+      "loss": 0.6463,
+      "step": 6157
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 3.9721236434706585e-05,
+      "loss": 0.6868,
+      "step": 6158
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 3.9718664815100555e-05,
+      "loss": 0.7361,
+      "step": 6159
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 3.9716093195494525e-05,
+      "loss": 0.939,
+      "step": 6160
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 3.9713521575888495e-05,
+      "loss": 0.577,
+      "step": 6161
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 3.9710949956282464e-05,
+      "loss": 0.888,
+      "step": 6162
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 3.970837833667644e-05,
+      "loss": 0.4837,
+      "step": 6163
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 3.970580671707042e-05,
+      "loss": 0.5891,
+      "step": 6164
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 3.970323509746439e-05,
+      "loss": 0.9104,
+      "step": 6165
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 3.9700663477858357e-05,
+      "loss": 0.8274,
+      "step": 6166
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 3.9698091858252326e-05,
+      "loss": 0.6907,
+      "step": 6167
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 3.96955202386463e-05,
+      "loss": 1.0461,
+      "step": 6168
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 3.969294861904027e-05,
+      "loss": 0.7931,
+      "step": 6169
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 3.969037699943425e-05,
+      "loss": 0.664,
+      "step": 6170
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 3.968780537982822e-05,
+      "loss": 0.9691,
+      "step": 6171
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 3.968523376022219e-05,
+      "loss": 1.009,
+      "step": 6172
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 3.9682662140616165e-05,
+      "loss": 0.6647,
+      "step": 6173
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 3.9680090521010134e-05,
+      "loss": 0.626,
+      "step": 6174
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 3.9677518901404104e-05,
+      "loss": 0.8449,
+      "step": 6175
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 3.967494728179808e-05,
+      "loss": 0.7525,
+      "step": 6176
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 3.967237566219205e-05,
+      "loss": 0.8927,
+      "step": 6177
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 3.966980404258602e-05,
+      "loss": 0.9708,
+      "step": 6178
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 3.9667232422979996e-05,
+      "loss": 0.7687,
+      "step": 6179
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 3.9664660803373966e-05,
+      "loss": 0.6638,
+      "step": 6180
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 3.9662089183767936e-05,
+      "loss": 0.7233,
+      "step": 6181
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 3.965951756416191e-05,
+      "loss": 0.572,
+      "step": 6182
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 3.965694594455588e-05,
+      "loss": 1.1026,
+      "step": 6183
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 3.965437432494986e-05,
+      "loss": 0.7987,
+      "step": 6184
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 3.965180270534383e-05,
+      "loss": 1.0194,
+      "step": 6185
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 3.96492310857378e-05,
+      "loss": 0.8882,
+      "step": 6186
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 3.964665946613177e-05,
+      "loss": 0.5584,
+      "step": 6187
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 3.9644087846525744e-05,
+      "loss": 0.7767,
+      "step": 6188
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 3.964151622691972e-05,
+      "loss": 1.1024,
+      "step": 6189
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 3.963894460731369e-05,
+      "loss": 0.6344,
+      "step": 6190
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 3.963637298770766e-05,
+      "loss": 0.8567,
+      "step": 6191
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 3.963380136810163e-05,
+      "loss": 0.6439,
+      "step": 6192
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 3.96312297484956e-05,
+      "loss": 0.8729,
+      "step": 6193
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 3.9628658128889575e-05,
+      "loss": 0.8283,
+      "step": 6194
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 3.962608650928355e-05,
+      "loss": 0.9959,
+      "step": 6195
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 3.962351488967752e-05,
+      "loss": 0.9479,
+      "step": 6196
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 3.962094327007149e-05,
+      "loss": 0.5392,
+      "step": 6197
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 3.961837165046546e-05,
+      "loss": 0.6807,
+      "step": 6198
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 3.961580003085944e-05,
+      "loss": 0.961,
+      "step": 6199
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 3.9613228411253414e-05,
+      "loss": 0.6873,
+      "step": 6200
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 3.9610656791647384e-05,
+      "loss": 0.6972,
+      "step": 6201
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 3.960808517204135e-05,
+      "loss": 0.6402,
+      "step": 6202
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 3.960551355243532e-05,
+      "loss": 0.9507,
+      "step": 6203
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 3.96029419328293e-05,
+      "loss": 0.7794,
+      "step": 6204
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 3.960037031322327e-05,
+      "loss": 0.6971,
+      "step": 6205
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 3.9597798693617246e-05,
+      "loss": 0.6929,
+      "step": 6206
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 3.9595227074011215e-05,
+      "loss": 0.502,
+      "step": 6207
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 3.9592655454405185e-05,
+      "loss": 1.1613,
+      "step": 6208
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 3.9590083834799155e-05,
+      "loss": 0.7977,
+      "step": 6209
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 3.958751221519313e-05,
+      "loss": 0.5445,
+      "step": 6210
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 3.95849405955871e-05,
+      "loss": 0.8955,
+      "step": 6211
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 3.958236897598108e-05,
+      "loss": 0.4447,
+      "step": 6212
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 3.957979735637505e-05,
+      "loss": 0.4201,
+      "step": 6213
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 3.957722573676902e-05,
+      "loss": 0.9699,
+      "step": 6214
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 3.957465411716299e-05,
+      "loss": 0.7835,
+      "step": 6215
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 3.957208249755696e-05,
+      "loss": 0.9812,
+      "step": 6216
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 3.956951087795093e-05,
+      "loss": 0.8673,
+      "step": 6217
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 3.956693925834491e-05,
+      "loss": 0.6216,
+      "step": 6218
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 3.956436763873888e-05,
+      "loss": 1.0541,
+      "step": 6219
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 3.9561796019132855e-05,
+      "loss": 0.978,
+      "step": 6220
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 3.9559224399526825e-05,
+      "loss": 0.4237,
+      "step": 6221
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 3.9556652779920794e-05,
+      "loss": 0.8609,
+      "step": 6222
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 3.9554081160314764e-05,
+      "loss": 0.6143,
+      "step": 6223
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 3.955150954070874e-05,
+      "loss": 1.0422,
+      "step": 6224
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 3.954893792110271e-05,
+      "loss": 0.8306,
+      "step": 6225
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 3.954636630149669e-05,
+      "loss": 0.7343,
+      "step": 6226
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 3.9543794681890656e-05,
+      "loss": 0.7686,
+      "step": 6227
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 3.9541223062284626e-05,
+      "loss": 0.5928,
+      "step": 6228
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 3.9538651442678596e-05,
+      "loss": 0.6992,
+      "step": 6229
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 3.953607982307257e-05,
+      "loss": 1.09,
+      "step": 6230
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 3.953350820346655e-05,
+      "loss": 0.6004,
+      "step": 6231
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 3.953093658386052e-05,
+      "loss": 0.7241,
+      "step": 6232
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 3.952836496425449e-05,
+      "loss": 0.5542,
+      "step": 6233
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 3.952579334464846e-05,
+      "loss": 0.809,
+      "step": 6234
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 3.9523221725042434e-05,
+      "loss": 1.0106,
+      "step": 6235
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 3.952065010543641e-05,
+      "loss": 0.6064,
+      "step": 6236
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 3.951807848583038e-05,
+      "loss": 0.6502,
+      "step": 6237
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 3.951550686622435e-05,
+      "loss": 0.6772,
+      "step": 6238
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 3.951293524661832e-05,
+      "loss": 0.7191,
+      "step": 6239
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 3.951036362701229e-05,
+      "loss": 0.8894,
+      "step": 6240
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 3.9507792007406266e-05,
+      "loss": 0.5585,
+      "step": 6241
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 3.950522038780024e-05,
+      "loss": 0.766,
+      "step": 6242
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 3.950264876819421e-05,
+      "loss": 0.88,
+      "step": 6243
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 3.950007714858818e-05,
+      "loss": 0.9527,
+      "step": 6244
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 3.949750552898215e-05,
+      "loss": 0.8412,
+      "step": 6245
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 3.949493390937613e-05,
+      "loss": 1.0038,
+      "step": 6246
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 3.9492362289770104e-05,
+      "loss": 0.6873,
+      "step": 6247
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 3.9489790670164074e-05,
+      "loss": 1.1227,
+      "step": 6248
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 3.9487219050558044e-05,
+      "loss": 0.7317,
+      "step": 6249
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 3.948464743095201e-05,
+      "loss": 0.7766,
+      "step": 6250
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 3.948207581134599e-05,
+      "loss": 0.6657,
+      "step": 6251
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 3.947950419173996e-05,
+      "loss": 0.7094,
+      "step": 6252
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 3.9476932572133936e-05,
+      "loss": 0.8955,
+      "step": 6253
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 3.9474360952527906e-05,
+      "loss": 0.79,
+      "step": 6254
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 3.9471789332921875e-05,
+      "loss": 0.8748,
+      "step": 6255
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 3.9469217713315845e-05,
+      "loss": 1.0596,
+      "step": 6256
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 3.946664609370982e-05,
+      "loss": 1.1143,
+      "step": 6257
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 3.946407447410379e-05,
+      "loss": 0.7959,
+      "step": 6258
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 3.946150285449777e-05,
+      "loss": 0.4034,
+      "step": 6259
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 3.945893123489174e-05,
+      "loss": 0.8675,
+      "step": 6260
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 3.945635961528571e-05,
+      "loss": 1.0736,
+      "step": 6261
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 3.9453787995679683e-05,
+      "loss": 0.939,
+      "step": 6262
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 3.945121637607365e-05,
+      "loss": 0.9178,
+      "step": 6263
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 3.944864475646762e-05,
+      "loss": 0.7458,
+      "step": 6264
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 3.94460731368616e-05,
+      "loss": 0.399,
+      "step": 6265
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 3.944350151725557e-05,
+      "loss": 0.9494,
+      "step": 6266
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 3.9440929897649545e-05,
+      "loss": 0.7411,
+      "step": 6267
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 3.9438358278043515e-05,
+      "loss": 0.6267,
+      "step": 6268
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 3.9435786658437485e-05,
+      "loss": 0.5888,
+      "step": 6269
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 3.9433215038831455e-05,
+      "loss": 0.5931,
+      "step": 6270
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 3.943064341922543e-05,
+      "loss": 0.892,
+      "step": 6271
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 3.94280717996194e-05,
+      "loss": 0.7494,
+      "step": 6272
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 3.942550018001338e-05,
+      "loss": 0.8275,
+      "step": 6273
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 3.942292856040735e-05,
+      "loss": 0.9708,
+      "step": 6274
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 3.9420356940801317e-05,
+      "loss": 0.8454,
+      "step": 6275
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 3.9417785321195286e-05,
+      "loss": 0.932,
+      "step": 6276
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 3.941521370158926e-05,
+      "loss": 0.5525,
+      "step": 6277
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 3.941264208198324e-05,
+      "loss": 0.5445,
+      "step": 6278
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 3.941007046237721e-05,
+      "loss": 1.2509,
+      "step": 6279
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 3.940749884277118e-05,
+      "loss": 1.008,
+      "step": 6280
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 3.940492722316515e-05,
+      "loss": 0.6851,
+      "step": 6281
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 3.9402355603559125e-05,
+      "loss": 0.7244,
+      "step": 6282
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 3.93997839839531e-05,
+      "loss": 0.6746,
+      "step": 6283
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 3.939721236434707e-05,
+      "loss": 1.1025,
+      "step": 6284
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 3.939464074474104e-05,
+      "loss": 0.997,
+      "step": 6285
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 3.939206912513501e-05,
+      "loss": 0.9654,
+      "step": 6286
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 3.938949750552898e-05,
+      "loss": 0.876,
+      "step": 6287
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 3.9386925885922956e-05,
+      "loss": 0.4546,
+      "step": 6288
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 3.938435426631693e-05,
+      "loss": 0.8664,
+      "step": 6289
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 3.93817826467109e-05,
+      "loss": 0.9336,
+      "step": 6290
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 3.937921102710487e-05,
+      "loss": 0.5646,
+      "step": 6291
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 3.937663940749884e-05,
+      "loss": 1.1338,
+      "step": 6292
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 3.937406778789282e-05,
+      "loss": 0.7774,
+      "step": 6293
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 3.937149616828679e-05,
+      "loss": 0.7714,
+      "step": 6294
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 3.9368924548680764e-05,
+      "loss": 0.5424,
+      "step": 6295
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 3.9366352929074734e-05,
+      "loss": 0.6943,
+      "step": 6296
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 3.9363781309468704e-05,
+      "loss": 0.7374,
+      "step": 6297
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 3.936120968986268e-05,
+      "loss": 0.9437,
+      "step": 6298
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 3.935863807025665e-05,
+      "loss": 0.8178,
+      "step": 6299
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 3.935606645065062e-05,
+      "loss": 0.713,
+      "step": 6300
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 3.9353494831044596e-05,
+      "loss": 0.825,
+      "step": 6301
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 3.9350923211438566e-05,
+      "loss": 0.7901,
+      "step": 6302
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 3.9348351591832535e-05,
+      "loss": 1.0365,
+      "step": 6303
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 3.934577997222651e-05,
+      "loss": 1.0431,
+      "step": 6304
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 3.934320835262048e-05,
+      "loss": 0.9043,
+      "step": 6305
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 3.934063673301445e-05,
+      "loss": 0.9597,
+      "step": 6306
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 3.933806511340843e-05,
+      "loss": 0.7487,
+      "step": 6307
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 3.93354934938024e-05,
+      "loss": 0.776,
+      "step": 6308
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 3.9332921874196374e-05,
+      "loss": 0.5428,
+      "step": 6309
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 3.9330350254590344e-05,
+      "loss": 0.6506,
+      "step": 6310
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 3.932777863498431e-05,
+      "loss": 0.6042,
+      "step": 6311
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 3.932520701537828e-05,
+      "loss": 0.3921,
+      "step": 6312
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 3.932263539577226e-05,
+      "loss": 0.8064,
+      "step": 6313
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 3.9320063776166236e-05,
+      "loss": 0.8045,
+      "step": 6314
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 3.9317492156560206e-05,
+      "loss": 0.5039,
+      "step": 6315
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 3.9314920536954175e-05,
+      "loss": 0.7869,
+      "step": 6316
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 3.9312348917348145e-05,
+      "loss": 0.8912,
+      "step": 6317
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 3.9309777297742115e-05,
+      "loss": 1.0228,
+      "step": 6318
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 3.930720567813609e-05,
+      "loss": 0.8225,
+      "step": 6319
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 3.930463405853007e-05,
+      "loss": 0.7677,
+      "step": 6320
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 3.930206243892404e-05,
+      "loss": 0.8122,
+      "step": 6321
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 3.929949081931801e-05,
+      "loss": 0.7892,
+      "step": 6322
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 3.9296919199711977e-05,
+      "loss": 0.778,
+      "step": 6323
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 3.929434758010595e-05,
+      "loss": 1.1692,
+      "step": 6324
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 3.929177596049993e-05,
+      "loss": 0.7448,
+      "step": 6325
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 3.92892043408939e-05,
+      "loss": 0.6387,
+      "step": 6326
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 3.928663272128787e-05,
+      "loss": 0.5663,
+      "step": 6327
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 3.928406110168184e-05,
+      "loss": 0.8007,
+      "step": 6328
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 3.9281489482075815e-05,
+      "loss": 0.9814,
+      "step": 6329
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 3.9278917862469785e-05,
+      "loss": 0.8743,
+      "step": 6330
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 3.927634624286376e-05,
+      "loss": 0.6595,
+      "step": 6331
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 3.927377462325773e-05,
+      "loss": 0.7996,
+      "step": 6332
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 3.92712030036517e-05,
+      "loss": 0.9327,
+      "step": 6333
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 3.926863138404567e-05,
+      "loss": 0.7021,
+      "step": 6334
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 3.926605976443965e-05,
+      "loss": 0.874,
+      "step": 6335
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 3.9263488144833616e-05,
+      "loss": 0.8912,
+      "step": 6336
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 3.926091652522759e-05,
+      "loss": 0.957,
+      "step": 6337
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 3.925834490562156e-05,
+      "loss": 0.8422,
+      "step": 6338
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 3.925577328601553e-05,
+      "loss": 0.9636,
+      "step": 6339
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 3.925320166640951e-05,
+      "loss": 0.7556,
+      "step": 6340
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 3.925063004680348e-05,
+      "loss": 0.5681,
+      "step": 6341
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 3.924805842719745e-05,
+      "loss": 1.0708,
+      "step": 6342
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 3.9245486807591425e-05,
+      "loss": 0.9587,
+      "step": 6343
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 3.9242915187985394e-05,
+      "loss": 0.6164,
+      "step": 6344
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 3.924034356837937e-05,
+      "loss": 0.678,
+      "step": 6345
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 3.923777194877334e-05,
+      "loss": 1.0126,
+      "step": 6346
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 3.923520032916731e-05,
+      "loss": 0.7206,
+      "step": 6347
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 3.923262870956128e-05,
+      "loss": 0.7641,
+      "step": 6348
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 3.9230057089955256e-05,
+      "loss": 0.6704,
+      "step": 6349
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 3.9227485470349226e-05,
+      "loss": 1.1126,
+      "step": 6350
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 3.92249138507432e-05,
+      "loss": 0.8025,
+      "step": 6351
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 3.922234223113717e-05,
+      "loss": 1.2854,
+      "step": 6352
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 3.921977061153114e-05,
+      "loss": 0.8018,
+      "step": 6353
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 3.921719899192511e-05,
+      "loss": 0.5154,
+      "step": 6354
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 3.921462737231909e-05,
+      "loss": 0.8172,
+      "step": 6355
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 3.9212055752713064e-05,
+      "loss": 0.7413,
+      "step": 6356
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 3.9209484133107034e-05,
+      "loss": 0.8595,
+      "step": 6357
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 3.9206912513501004e-05,
+      "loss": 0.9874,
+      "step": 6358
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 3.920434089389497e-05,
+      "loss": 0.7484,
+      "step": 6359
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 3.920176927428894e-05,
+      "loss": 0.5891,
+      "step": 6360
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 3.9199197654682926e-05,
+      "loss": 0.706,
+      "step": 6361
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 3.9196626035076896e-05,
+      "loss": 0.9314,
+      "step": 6362
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 3.9194054415470866e-05,
+      "loss": 0.4499,
+      "step": 6363
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 3.9191482795864835e-05,
+      "loss": 0.3994,
+      "step": 6364
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 3.9188911176258805e-05,
+      "loss": 1.0093,
+      "step": 6365
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 3.918633955665278e-05,
+      "loss": 0.7573,
+      "step": 6366
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 3.918376793704676e-05,
+      "loss": 0.847,
+      "step": 6367
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 3.918119631744073e-05,
+      "loss": 1.0886,
+      "step": 6368
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 3.91786246978347e-05,
+      "loss": 0.8385,
+      "step": 6369
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 3.917605307822867e-05,
+      "loss": 0.8268,
+      "step": 6370
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 3.9173481458622643e-05,
+      "loss": 0.873,
+      "step": 6371
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 3.917090983901661e-05,
+      "loss": 0.9085,
+      "step": 6372
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 3.916833821941059e-05,
+      "loss": 0.7312,
+      "step": 6373
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 3.916576659980456e-05,
+      "loss": 0.8478,
+      "step": 6374
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 3.916319498019853e-05,
+      "loss": 0.8222,
+      "step": 6375
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 3.9160623360592505e-05,
+      "loss": 0.786,
+      "step": 6376
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 3.9158051740986475e-05,
+      "loss": 1.0083,
+      "step": 6377
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 3.9155480121380445e-05,
+      "loss": 1.0691,
+      "step": 6378
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 3.915290850177442e-05,
+      "loss": 0.7144,
+      "step": 6379
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 3.915033688216839e-05,
+      "loss": 0.6814,
+      "step": 6380
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 3.914776526256236e-05,
+      "loss": 0.8747,
+      "step": 6381
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 3.914519364295634e-05,
+      "loss": 0.7617,
+      "step": 6382
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 3.914262202335031e-05,
+      "loss": 0.7422,
+      "step": 6383
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 3.914005040374428e-05,
+      "loss": 0.6683,
+      "step": 6384
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 3.913747878413825e-05,
+      "loss": 0.9247,
+      "step": 6385
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 3.913490716453222e-05,
+      "loss": 1.0146,
+      "step": 6386
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 3.91323355449262e-05,
+      "loss": 0.7941,
+      "step": 6387
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 3.912976392532017e-05,
+      "loss": 0.7602,
+      "step": 6388
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 3.912719230571414e-05,
+      "loss": 0.8928,
+      "step": 6389
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 3.9124620686108115e-05,
+      "loss": 0.8187,
+      "step": 6390
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 3.9122049066502085e-05,
+      "loss": 0.6715,
+      "step": 6391
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 3.911947744689606e-05,
+      "loss": 0.901,
+      "step": 6392
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 3.911690582729003e-05,
+      "loss": 0.839,
+      "step": 6393
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 3.9114334207684e-05,
+      "loss": 0.7309,
+      "step": 6394
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 3.911176258807797e-05,
+      "loss": 0.7677,
+      "step": 6395
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 3.9109190968471947e-05,
+      "loss": 1.191,
+      "step": 6396
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 3.9106619348865916e-05,
+      "loss": 0.708,
+      "step": 6397
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 3.910404772925989e-05,
+      "loss": 0.6758,
+      "step": 6398
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 3.910147610965386e-05,
+      "loss": 0.9139,
+      "step": 6399
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 3.909890449004783e-05,
+      "loss": 0.9541,
+      "step": 6400
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 3.90963328704418e-05,
+      "loss": 0.6435,
+      "step": 6401
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 3.909376125083578e-05,
+      "loss": 0.8382,
+      "step": 6402
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 3.9091189631229755e-05,
+      "loss": 1.1357,
+      "step": 6403
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 3.9088618011623724e-05,
+      "loss": 0.8723,
+      "step": 6404
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 3.9086046392017694e-05,
+      "loss": 0.6753,
+      "step": 6405
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 3.9083474772411664e-05,
+      "loss": 0.6698,
+      "step": 6406
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 3.908090315280564e-05,
+      "loss": 0.8346,
+      "step": 6407
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 3.907833153319962e-05,
+      "loss": 0.6868,
+      "step": 6408
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 3.9075759913593586e-05,
+      "loss": 0.6411,
+      "step": 6409
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 3.9073188293987556e-05,
+      "loss": 0.9553,
+      "step": 6410
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 3.9070616674381526e-05,
+      "loss": 0.7899,
+      "step": 6411
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 3.9068045054775495e-05,
+      "loss": 0.8169,
+      "step": 6412
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 3.906547343516947e-05,
+      "loss": 0.6559,
+      "step": 6413
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 3.906290181556345e-05,
+      "loss": 0.8047,
+      "step": 6414
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 3.906033019595742e-05,
+      "loss": 0.6011,
+      "step": 6415
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 3.905775857635139e-05,
+      "loss": 0.9828,
+      "step": 6416
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 3.905518695674536e-05,
+      "loss": 0.9126,
+      "step": 6417
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 3.9052615337139334e-05,
+      "loss": 0.7415,
+      "step": 6418
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 3.9050043717533304e-05,
+      "loss": 0.7836,
+      "step": 6419
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 3.904747209792728e-05,
+      "loss": 0.7566,
+      "step": 6420
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 3.904490047832125e-05,
+      "loss": 0.5611,
+      "step": 6421
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 3.904232885871522e-05,
+      "loss": 0.6673,
+      "step": 6422
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 3.9039757239109196e-05,
+      "loss": 0.4765,
+      "step": 6423
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 3.9037185619503166e-05,
+      "loss": 0.9095,
+      "step": 6424
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 3.9034613999897135e-05,
+      "loss": 0.8726,
+      "step": 6425
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 3.903204238029111e-05,
+      "loss": 0.6696,
+      "step": 6426
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 3.902947076068508e-05,
+      "loss": 0.6469,
+      "step": 6427
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 3.902689914107905e-05,
+      "loss": 0.7728,
+      "step": 6428
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 3.902432752147303e-05,
+      "loss": 0.5965,
+      "step": 6429
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 3.9021755901867e-05,
+      "loss": 0.98,
+      "step": 6430
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 3.901918428226097e-05,
+      "loss": 0.3509,
+      "step": 6431
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 3.901661266265494e-05,
+      "loss": 0.8256,
+      "step": 6432
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 3.901404104304891e-05,
+      "loss": 0.6418,
+      "step": 6433
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 3.901146942344289e-05,
+      "loss": 0.741,
+      "step": 6434
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 3.900889780383686e-05,
+      "loss": 0.7963,
+      "step": 6435
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 3.900632618423083e-05,
+      "loss": 0.7352,
+      "step": 6436
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 3.90037545646248e-05,
+      "loss": 0.9053,
+      "step": 6437
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 3.9001182945018775e-05,
+      "loss": 0.8117,
+      "step": 6438
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 3.899861132541275e-05,
+      "loss": 1.0607,
+      "step": 6439
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 3.899603970580672e-05,
+      "loss": 0.9308,
+      "step": 6440
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 3.899346808620069e-05,
+      "loss": 0.9733,
+      "step": 6441
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 3.899089646659466e-05,
+      "loss": 0.4787,
+      "step": 6442
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 3.898832484698863e-05,
+      "loss": 0.7978,
+      "step": 6443
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 3.898575322738261e-05,
+      "loss": 0.6124,
+      "step": 6444
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 3.898318160777658e-05,
+      "loss": 0.7088,
+      "step": 6445
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 3.898060998817055e-05,
+      "loss": 0.8942,
+      "step": 6446
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 3.897803836856452e-05,
+      "loss": 0.859,
+      "step": 6447
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 3.897546674895849e-05,
+      "loss": 0.8703,
+      "step": 6448
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 3.897289512935247e-05,
+      "loss": 0.6262,
+      "step": 6449
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 3.8970323509746445e-05,
+      "loss": 0.6937,
+      "step": 6450
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 3.8967751890140415e-05,
+      "loss": 0.7221,
+      "step": 6451
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 3.8965180270534384e-05,
+      "loss": 0.7966,
+      "step": 6452
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 3.8962608650928354e-05,
+      "loss": 0.4993,
+      "step": 6453
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 3.896003703132233e-05,
+      "loss": 0.9303,
+      "step": 6454
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 3.89574654117163e-05,
+      "loss": 0.6812,
+      "step": 6455
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 3.895489379211028e-05,
+      "loss": 0.7535,
+      "step": 6456
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 3.8952322172504246e-05,
+      "loss": 0.5614,
+      "step": 6457
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 3.8949750552898216e-05,
+      "loss": 1.0177,
+      "step": 6458
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 3.8947178933292186e-05,
+      "loss": 1.0378,
+      "step": 6459
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 3.894460731368616e-05,
+      "loss": 1.0898,
+      "step": 6460
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 3.894203569408013e-05,
+      "loss": 0.9991,
+      "step": 6461
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 3.893946407447411e-05,
+      "loss": 0.6178,
+      "step": 6462
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 3.893689245486808e-05,
+      "loss": 1.006,
+      "step": 6463
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 3.893432083526205e-05,
+      "loss": 0.6704,
+      "step": 6464
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 3.8931749215656024e-05,
+      "loss": 0.6319,
+      "step": 6465
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 3.8929177596049994e-05,
+      "loss": 0.7766,
+      "step": 6466
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 3.8926605976443964e-05,
+      "loss": 0.8266,
+      "step": 6467
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 3.892403435683794e-05,
+      "loss": 0.6755,
+      "step": 6468
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 3.892146273723191e-05,
+      "loss": 1.0032,
+      "step": 6469
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 3.8918891117625886e-05,
+      "loss": 0.9251,
+      "step": 6470
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 3.8916319498019856e-05,
+      "loss": 0.5689,
+      "step": 6471
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 3.8913747878413826e-05,
+      "loss": 0.9323,
+      "step": 6472
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 3.8911176258807795e-05,
+      "loss": 0.7822,
+      "step": 6473
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 3.890860463920177e-05,
+      "loss": 0.7792,
+      "step": 6474
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 3.890603301959574e-05,
+      "loss": 0.7525,
+      "step": 6475
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 3.890346139998972e-05,
+      "loss": 0.9546,
+      "step": 6476
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 3.890088978038369e-05,
+      "loss": 0.506,
+      "step": 6477
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 3.889831816077766e-05,
+      "loss": 0.637,
+      "step": 6478
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 3.889574654117163e-05,
+      "loss": 0.6401,
+      "step": 6479
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 3.8893174921565603e-05,
+      "loss": 0.7636,
+      "step": 6480
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 3.889060330195958e-05,
+      "loss": 0.665,
+      "step": 6481
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 3.888803168235355e-05,
+      "loss": 1.0572,
+      "step": 6482
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 3.888546006274752e-05,
+      "loss": 0.8415,
+      "step": 6483
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 3.888288844314149e-05,
+      "loss": 0.9243,
+      "step": 6484
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 3.888031682353546e-05,
+      "loss": 0.7888,
+      "step": 6485
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 3.887774520392944e-05,
+      "loss": 0.7175,
+      "step": 6486
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 3.887517358432341e-05,
+      "loss": 0.6624,
+      "step": 6487
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 3.887260196471738e-05,
+      "loss": 0.8833,
+      "step": 6488
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 3.887003034511135e-05,
+      "loss": 0.7607,
+      "step": 6489
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 3.886745872550532e-05,
+      "loss": 0.6918,
+      "step": 6490
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 3.88648871058993e-05,
+      "loss": 0.6716,
+      "step": 6491
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 3.8862315486293274e-05,
+      "loss": 0.6726,
+      "step": 6492
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 3.885974386668724e-05,
+      "loss": 0.5695,
+      "step": 6493
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 3.885717224708121e-05,
+      "loss": 0.7199,
+      "step": 6494
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 3.885460062747518e-05,
+      "loss": 0.8502,
+      "step": 6495
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 3.885202900786916e-05,
+      "loss": 0.8182,
+      "step": 6496
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 3.884945738826313e-05,
+      "loss": 0.6921,
+      "step": 6497
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 3.8846885768657105e-05,
+      "loss": 0.6483,
+      "step": 6498
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 3.8844314149051075e-05,
+      "loss": 0.6506,
+      "step": 6499
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 3.8841742529445045e-05,
+      "loss": 0.707,
+      "step": 6500
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 3.883917090983902e-05,
+      "loss": 0.7061,
+      "step": 6501
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 3.883659929023299e-05,
+      "loss": 0.7186,
+      "step": 6502
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 3.883402767062696e-05,
+      "loss": 0.9052,
+      "step": 6503
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 3.883145605102094e-05,
+      "loss": 0.8081,
+      "step": 6504
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 3.8828884431414907e-05,
+      "loss": 0.5622,
+      "step": 6505
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 3.8826312811808876e-05,
+      "loss": 0.806,
+      "step": 6506
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 3.882374119220285e-05,
+      "loss": 0.7067,
+      "step": 6507
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 3.882116957259682e-05,
+      "loss": 0.8708,
+      "step": 6508
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 3.881859795299079e-05,
+      "loss": 1.1094,
+      "step": 6509
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 3.881602633338477e-05,
+      "loss": 1.0358,
+      "step": 6510
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 3.881345471377874e-05,
+      "loss": 0.9051,
+      "step": 6511
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 3.8810883094172715e-05,
+      "loss": 0.7047,
+      "step": 6512
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 3.8808311474566684e-05,
+      "loss": 1.141,
+      "step": 6513
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 3.8805739854960654e-05,
+      "loss": 0.6532,
+      "step": 6514
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 3.8803168235354624e-05,
+      "loss": 0.4291,
+      "step": 6515
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 3.88005966157486e-05,
+      "loss": 0.8621,
+      "step": 6516
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 3.879802499614258e-05,
+      "loss": 0.7483,
+      "step": 6517
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 3.8795453376536546e-05,
+      "loss": 0.6354,
+      "step": 6518
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 3.8792881756930516e-05,
+      "loss": 0.8928,
+      "step": 6519
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 3.8790310137324486e-05,
+      "loss": 0.8058,
+      "step": 6520
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 3.8787738517718455e-05,
+      "loss": 0.9016,
+      "step": 6521
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 3.878516689811243e-05,
+      "loss": 0.9213,
+      "step": 6522
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 3.878259527850641e-05,
+      "loss": 0.5857,
+      "step": 6523
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 3.878002365890038e-05,
+      "loss": 1.0359,
+      "step": 6524
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 3.877745203929435e-05,
+      "loss": 0.8414,
+      "step": 6525
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 3.877488041968832e-05,
+      "loss": 0.9402,
+      "step": 6526
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 3.8772308800082294e-05,
+      "loss": 0.4519,
+      "step": 6527
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 3.876973718047627e-05,
+      "loss": 0.6878,
+      "step": 6528
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 3.876716556087024e-05,
+      "loss": 0.639,
+      "step": 6529
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 3.876459394126421e-05,
+      "loss": 0.5153,
+      "step": 6530
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 3.876202232165818e-05,
+      "loss": 0.6911,
+      "step": 6531
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 3.8759450702052156e-05,
+      "loss": 0.7474,
+      "step": 6532
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 3.875687908244613e-05,
+      "loss": 0.6242,
+      "step": 6533
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 3.87543074628401e-05,
+      "loss": 0.7321,
+      "step": 6534
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 3.875173584323407e-05,
+      "loss": 0.9101,
+      "step": 6535
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 3.874916422362804e-05,
+      "loss": 0.5791,
+      "step": 6536
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 3.874659260402201e-05,
+      "loss": 0.6181,
+      "step": 6537
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 3.874402098441599e-05,
+      "loss": 0.4273,
+      "step": 6538
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 3.8741449364809964e-05,
+      "loss": 0.7659,
+      "step": 6539
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 3.8738877745203934e-05,
+      "loss": 0.9523,
+      "step": 6540
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 3.87363061255979e-05,
+      "loss": 1.0146,
+      "step": 6541
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 3.873373450599187e-05,
+      "loss": 0.9549,
+      "step": 6542
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 3.873116288638585e-05,
+      "loss": 0.4983,
+      "step": 6543
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 3.872859126677982e-05,
+      "loss": 0.9753,
+      "step": 6544
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 3.8726019647173796e-05,
+      "loss": 0.7859,
+      "step": 6545
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 3.8723448027567765e-05,
+      "loss": 0.91,
+      "step": 6546
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 3.8720876407961735e-05,
+      "loss": 0.9072,
+      "step": 6547
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 3.871830478835571e-05,
+      "loss": 0.6334,
+      "step": 6548
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 3.871573316874968e-05,
+      "loss": 0.7147,
+      "step": 6549
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 3.871316154914365e-05,
+      "loss": 0.8128,
+      "step": 6550
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 3.871058992953763e-05,
+      "loss": 0.717,
+      "step": 6551
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 3.87080183099316e-05,
+      "loss": 0.7864,
+      "step": 6552
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 3.870544669032557e-05,
+      "loss": 0.5945,
+      "step": 6553
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 3.870287507071954e-05,
+      "loss": 1.0838,
+      "step": 6554
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 3.870030345111351e-05,
+      "loss": 0.6683,
+      "step": 6555
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 3.869773183150748e-05,
+      "loss": 0.9259,
+      "step": 6556
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 3.869516021190146e-05,
+      "loss": 0.9507,
+      "step": 6557
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 3.869258859229543e-05,
+      "loss": 0.5061,
+      "step": 6558
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 3.8690016972689405e-05,
+      "loss": 1.2379,
+      "step": 6559
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 3.8687445353083375e-05,
+      "loss": 0.8998,
+      "step": 6560
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 3.8684873733477344e-05,
+      "loss": 0.7382,
+      "step": 6561
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 3.8682302113871314e-05,
+      "loss": 0.7264,
+      "step": 6562
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 3.867973049426529e-05,
+      "loss": 0.8028,
+      "step": 6563
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 3.867715887465927e-05,
+      "loss": 0.5917,
+      "step": 6564
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 3.867458725505324e-05,
+      "loss": 0.7536,
+      "step": 6565
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 3.8672015635447206e-05,
+      "loss": 0.656,
+      "step": 6566
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 3.8669444015841176e-05,
+      "loss": 0.789,
+      "step": 6567
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 3.8666872396235146e-05,
+      "loss": 1.0657,
+      "step": 6568
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 3.866430077662912e-05,
+      "loss": 0.7193,
+      "step": 6569
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 3.86617291570231e-05,
+      "loss": 0.9337,
+      "step": 6570
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 3.865915753741707e-05,
+      "loss": 0.5426,
+      "step": 6571
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 3.865658591781104e-05,
+      "loss": 1.0079,
+      "step": 6572
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 3.865401429820501e-05,
+      "loss": 0.9089,
+      "step": 6573
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 3.8651442678598984e-05,
+      "loss": 0.6443,
+      "step": 6574
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 3.864887105899296e-05,
+      "loss": 0.8049,
+      "step": 6575
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 3.864629943938693e-05,
+      "loss": 0.7919,
+      "step": 6576
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 3.86437278197809e-05,
+      "loss": 0.8912,
+      "step": 6577
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 3.864115620017487e-05,
+      "loss": 0.6993,
+      "step": 6578
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 3.8638584580568846e-05,
+      "loss": 0.6379,
+      "step": 6579
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 3.8636012960962816e-05,
+      "loss": 0.8128,
+      "step": 6580
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 3.863344134135679e-05,
+      "loss": 0.834,
+      "step": 6581
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 3.863086972175076e-05,
+      "loss": 0.9096,
+      "step": 6582
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 3.862829810214473e-05,
+      "loss": 0.9002,
+      "step": 6583
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 3.86257264825387e-05,
+      "loss": 1.0219,
+      "step": 6584
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 3.862315486293268e-05,
+      "loss": 0.6567,
+      "step": 6585
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 3.862058324332665e-05,
+      "loss": 0.8344,
+      "step": 6586
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 3.8618011623720624e-05,
+      "loss": 0.6547,
+      "step": 6587
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 3.8615440004114594e-05,
+      "loss": 0.5052,
+      "step": 6588
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 3.8612868384508563e-05,
+      "loss": 0.7981,
+      "step": 6589
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 3.861029676490254e-05,
+      "loss": 0.5907,
+      "step": 6590
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 3.860772514529651e-05,
+      "loss": 0.9092,
+      "step": 6591
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 3.860515352569048e-05,
+      "loss": 0.6043,
+      "step": 6592
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 3.8602581906084456e-05,
+      "loss": 0.6953,
+      "step": 6593
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 3.8600010286478425e-05,
+      "loss": 0.8251,
+      "step": 6594
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 3.85974386668724e-05,
+      "loss": 0.9224,
+      "step": 6595
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 3.859486704726637e-05,
+      "loss": 0.7257,
+      "step": 6596
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 3.859229542766034e-05,
+      "loss": 0.6534,
+      "step": 6597
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 3.858972380805431e-05,
+      "loss": 0.6639,
+      "step": 6598
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 3.858715218844829e-05,
+      "loss": 0.9491,
+      "step": 6599
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 3.858458056884226e-05,
+      "loss": 0.7977,
+      "step": 6600
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 3.8582008949236234e-05,
+      "loss": 0.8673,
+      "step": 6601
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 3.85794373296302e-05,
+      "loss": 0.9232,
+      "step": 6602
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 3.857686571002417e-05,
+      "loss": 1.031,
+      "step": 6603
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 3.857429409041814e-05,
+      "loss": 0.9001,
+      "step": 6604
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 3.857172247081212e-05,
+      "loss": 0.7037,
+      "step": 6605
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 3.8569150851206095e-05,
+      "loss": 0.6269,
+      "step": 6606
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 3.8566579231600065e-05,
+      "loss": 0.7206,
+      "step": 6607
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 3.8564007611994035e-05,
+      "loss": 1.0341,
+      "step": 6608
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 3.8561435992388005e-05,
+      "loss": 0.4932,
+      "step": 6609
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 3.8558864372781974e-05,
+      "loss": 0.6856,
+      "step": 6610
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 3.855629275317596e-05,
+      "loss": 0.909,
+      "step": 6611
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 3.855372113356993e-05,
+      "loss": 0.5677,
+      "step": 6612
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 3.85511495139639e-05,
+      "loss": 0.7856,
+      "step": 6613
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 3.8548577894357867e-05,
+      "loss": 0.5864,
+      "step": 6614
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 3.8546006274751836e-05,
+      "loss": 0.4665,
+      "step": 6615
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 3.854343465514581e-05,
+      "loss": 0.7578,
+      "step": 6616
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 3.854086303553979e-05,
+      "loss": 0.642,
+      "step": 6617
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 3.853829141593376e-05,
+      "loss": 1.0233,
+      "step": 6618
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 3.853571979632773e-05,
+      "loss": 0.9262,
+      "step": 6619
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 3.85331481767217e-05,
+      "loss": 0.6584,
+      "step": 6620
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 3.8530576557115675e-05,
+      "loss": 0.4514,
+      "step": 6621
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 3.8528004937509644e-05,
+      "loss": 0.7159,
+      "step": 6622
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 3.852543331790362e-05,
+      "loss": 0.6831,
+      "step": 6623
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 3.852286169829759e-05,
+      "loss": 0.8524,
+      "step": 6624
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 3.852029007869156e-05,
+      "loss": 0.6069,
+      "step": 6625
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 3.851771845908554e-05,
+      "loss": 0.5795,
+      "step": 6626
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 3.8515146839479506e-05,
+      "loss": 0.9333,
+      "step": 6627
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 3.8512575219873476e-05,
+      "loss": 0.8851,
+      "step": 6628
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 3.851000360026745e-05,
+      "loss": 0.4891,
+      "step": 6629
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 3.850743198066142e-05,
+      "loss": 0.6911,
+      "step": 6630
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 3.850486036105539e-05,
+      "loss": 0.9386,
+      "step": 6631
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 3.850228874144937e-05,
+      "loss": 1.0641,
+      "step": 6632
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 3.849971712184334e-05,
+      "loss": 0.625,
+      "step": 6633
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 3.849714550223731e-05,
+      "loss": 0.412,
+      "step": 6634
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 3.8494573882631284e-05,
+      "loss": 0.9518,
+      "step": 6635
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 3.8492002263025254e-05,
+      "loss": 1.0571,
+      "step": 6636
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 3.848943064341923e-05,
+      "loss": 0.6358,
+      "step": 6637
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 3.84868590238132e-05,
+      "loss": 0.611,
+      "step": 6638
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 3.848428740420717e-05,
+      "loss": 1.0493,
+      "step": 6639
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 3.848171578460114e-05,
+      "loss": 0.4972,
+      "step": 6640
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 3.8479144164995116e-05,
+      "loss": 0.8904,
+      "step": 6641
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 3.847657254538909e-05,
+      "loss": 0.5304,
+      "step": 6642
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 3.847400092578306e-05,
+      "loss": 0.9456,
+      "step": 6643
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 3.847142930617703e-05,
+      "loss": 1.0246,
+      "step": 6644
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 3.8468857686571e-05,
+      "loss": 0.784,
+      "step": 6645
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 3.846628606696497e-05,
+      "loss": 0.7275,
+      "step": 6646
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 3.846371444735895e-05,
+      "loss": 0.7609,
+      "step": 6647
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 3.8461142827752924e-05,
+      "loss": 0.9369,
+      "step": 6648
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 3.8458571208146894e-05,
+      "loss": 0.5118,
+      "step": 6649
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 3.845599958854086e-05,
+      "loss": 0.7388,
+      "step": 6650
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 3.845342796893483e-05,
+      "loss": 0.6765,
+      "step": 6651
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 3.845085634932881e-05,
+      "loss": 1.1567,
+      "step": 6652
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 3.8448284729722786e-05,
+      "loss": 1.0271,
+      "step": 6653
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 3.8445713110116756e-05,
+      "loss": 0.964,
+      "step": 6654
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 3.8443141490510725e-05,
+      "loss": 0.7237,
+      "step": 6655
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 3.8440569870904695e-05,
+      "loss": 0.8782,
+      "step": 6656
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 3.843799825129867e-05,
+      "loss": 0.6119,
+      "step": 6657
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 3.843542663169264e-05,
+      "loss": 1.1899,
+      "step": 6658
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 3.843285501208662e-05,
+      "loss": 0.9509,
+      "step": 6659
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 3.843028339248059e-05,
+      "loss": 0.8484,
+      "step": 6660
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 3.842771177287456e-05,
+      "loss": 0.7931,
+      "step": 6661
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 3.8425140153268527e-05,
+      "loss": 0.6363,
+      "step": 6662
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 3.84225685336625e-05,
+      "loss": 0.8642,
+      "step": 6663
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 3.841999691405647e-05,
+      "loss": 0.7974,
+      "step": 6664
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 3.841742529445045e-05,
+      "loss": 0.7226,
+      "step": 6665
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 3.841485367484442e-05,
+      "loss": 0.8575,
+      "step": 6666
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 3.841228205523839e-05,
+      "loss": 0.6876,
+      "step": 6667
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 3.8409710435632365e-05,
+      "loss": 0.7598,
+      "step": 6668
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 3.8407138816026335e-05,
+      "loss": 1.1325,
+      "step": 6669
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 3.8404567196420304e-05,
+      "loss": 0.8998,
+      "step": 6670
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 3.840199557681428e-05,
+      "loss": 0.9475,
+      "step": 6671
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 3.839942395720825e-05,
+      "loss": 0.8554,
+      "step": 6672
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 3.839685233760223e-05,
+      "loss": 0.6278,
+      "step": 6673
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 3.83942807179962e-05,
+      "loss": 0.6856,
+      "step": 6674
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 3.8391709098390166e-05,
+      "loss": 1.1967,
+      "step": 6675
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 3.8389137478784136e-05,
+      "loss": 0.5274,
+      "step": 6676
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 3.838656585917811e-05,
+      "loss": 0.7921,
+      "step": 6677
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 3.838399423957208e-05,
+      "loss": 0.7513,
+      "step": 6678
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 3.838142261996606e-05,
+      "loss": 0.9392,
+      "step": 6679
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 3.837885100036003e-05,
+      "loss": 0.7891,
+      "step": 6680
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 3.8376279380754e-05,
+      "loss": 0.7989,
+      "step": 6681
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 3.837370776114797e-05,
+      "loss": 0.7135,
+      "step": 6682
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 3.8371136141541944e-05,
+      "loss": 1.0202,
+      "step": 6683
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 3.836856452193592e-05,
+      "loss": 0.8613,
+      "step": 6684
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 3.836599290232989e-05,
+      "loss": 0.7445,
+      "step": 6685
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 3.836342128272386e-05,
+      "loss": 1.096,
+      "step": 6686
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 3.836084966311783e-05,
+      "loss": 0.5727,
+      "step": 6687
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 3.8358278043511806e-05,
+      "loss": 1.0717,
+      "step": 6688
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 3.835570642390578e-05,
+      "loss": 0.7121,
+      "step": 6689
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 3.835313480429975e-05,
+      "loss": 0.9089,
+      "step": 6690
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 3.835056318469372e-05,
+      "loss": 0.7227,
+      "step": 6691
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 3.834799156508769e-05,
+      "loss": 1.0557,
+      "step": 6692
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 3.834541994548166e-05,
+      "loss": 0.6698,
+      "step": 6693
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 3.834284832587564e-05,
+      "loss": 1.0497,
+      "step": 6694
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 3.8340276706269614e-05,
+      "loss": 0.8099,
+      "step": 6695
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 3.8337705086663584e-05,
+      "loss": 0.747,
+      "step": 6696
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 3.8335133467057554e-05,
+      "loss": 0.5971,
+      "step": 6697
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 3.833256184745152e-05,
+      "loss": 0.7889,
+      "step": 6698
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 3.83299902278455e-05,
+      "loss": 0.6562,
+      "step": 6699
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 3.8327418608239476e-05,
+      "loss": 1.0878,
+      "step": 6700
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 3.8324846988633446e-05,
+      "loss": 0.6739,
+      "step": 6701
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 3.8322275369027416e-05,
+      "loss": 0.7956,
+      "step": 6702
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 3.8319703749421385e-05,
+      "loss": 1.0397,
+      "step": 6703
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 3.831713212981536e-05,
+      "loss": 0.9546,
+      "step": 6704
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 3.831456051020933e-05,
+      "loss": 0.9638,
+      "step": 6705
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 3.831198889060331e-05,
+      "loss": 0.6608,
+      "step": 6706
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 3.830941727099728e-05,
+      "loss": 1.2679,
+      "step": 6707
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 3.830684565139125e-05,
+      "loss": 0.8574,
+      "step": 6708
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 3.830427403178522e-05,
+      "loss": 0.5495,
+      "step": 6709
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 3.8301702412179193e-05,
+      "loss": 0.8366,
+      "step": 6710
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 3.829913079257316e-05,
+      "loss": 0.7231,
+      "step": 6711
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 3.829655917296714e-05,
+      "loss": 0.6222,
+      "step": 6712
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 3.829398755336111e-05,
+      "loss": 0.5986,
+      "step": 6713
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 3.829141593375508e-05,
+      "loss": 0.6325,
+      "step": 6714
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 3.8288844314149055e-05,
+      "loss": 0.5941,
+      "step": 6715
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 3.8286272694543025e-05,
+      "loss": 1.0594,
+      "step": 6716
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 3.8283701074936995e-05,
+      "loss": 0.9425,
+      "step": 6717
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 3.828112945533097e-05,
+      "loss": 0.7209,
+      "step": 6718
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 3.827855783572494e-05,
+      "loss": 0.6978,
+      "step": 6719
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 3.827598621611892e-05,
+      "loss": 0.5846,
+      "step": 6720
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 3.827341459651289e-05,
+      "loss": 0.4727,
+      "step": 6721
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 3.827084297690686e-05,
+      "loss": 0.5998,
+      "step": 6722
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 3.8268271357300826e-05,
+      "loss": 0.6547,
+      "step": 6723
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 3.82656997376948e-05,
+      "loss": 0.7916,
+      "step": 6724
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 3.826312811808877e-05,
+      "loss": 0.6108,
+      "step": 6725
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 3.826055649848275e-05,
+      "loss": 0.8623,
+      "step": 6726
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 3.825798487887672e-05,
+      "loss": 1.0121,
+      "step": 6727
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 3.825541325927069e-05,
+      "loss": 0.8715,
+      "step": 6728
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 3.825284163966466e-05,
+      "loss": 1.3789,
+      "step": 6729
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 3.8250270020058635e-05,
+      "loss": 0.5362,
+      "step": 6730
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 3.824769840045261e-05,
+      "loss": 0.9642,
+      "step": 6731
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 3.824512678084658e-05,
+      "loss": 0.7553,
+      "step": 6732
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 3.824255516124055e-05,
+      "loss": 1.2379,
+      "step": 6733
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 3.823998354163452e-05,
+      "loss": 0.7248,
+      "step": 6734
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 3.8237411922028497e-05,
+      "loss": 0.7205,
+      "step": 6735
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 3.823484030242247e-05,
+      "loss": 0.7105,
+      "step": 6736
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 3.823226868281644e-05,
+      "loss": 0.484,
+      "step": 6737
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 3.822969706321041e-05,
+      "loss": 0.7804,
+      "step": 6738
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 3.822712544360438e-05,
+      "loss": 0.7171,
+      "step": 6739
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 3.822455382399835e-05,
+      "loss": 0.5852,
+      "step": 6740
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 3.822198220439233e-05,
+      "loss": 0.4687,
+      "step": 6741
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 3.8219410584786305e-05,
+      "loss": 1.0388,
+      "step": 6742
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 3.8216838965180274e-05,
+      "loss": 0.8232,
+      "step": 6743
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 3.8214267345574244e-05,
+      "loss": 0.7724,
+      "step": 6744
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 3.8211695725968214e-05,
+      "loss": 0.7119,
+      "step": 6745
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 3.820912410636219e-05,
+      "loss": 0.7181,
+      "step": 6746
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 3.820655248675616e-05,
+      "loss": 0.8812,
+      "step": 6747
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 3.8203980867150136e-05,
+      "loss": 0.9092,
+      "step": 6748
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 3.8201409247544106e-05,
+      "loss": 0.87,
+      "step": 6749
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 3.8198837627938076e-05,
+      "loss": 0.8873,
+      "step": 6750
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 3.819626600833205e-05,
+      "loss": 0.6802,
+      "step": 6751
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 3.819369438872602e-05,
+      "loss": 0.6675,
+      "step": 6752
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 3.819112276911999e-05,
+      "loss": 0.8271,
+      "step": 6753
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 3.818855114951397e-05,
+      "loss": 0.6568,
+      "step": 6754
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 3.818597952990794e-05,
+      "loss": 0.627,
+      "step": 6755
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 3.818340791030191e-05,
+      "loss": 0.451,
+      "step": 6756
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 3.8180836290695884e-05,
+      "loss": 0.7924,
+      "step": 6757
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 3.8178264671089854e-05,
+      "loss": 1.0205,
+      "step": 6758
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 3.817569305148382e-05,
+      "loss": 0.8826,
+      "step": 6759
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 3.81731214318778e-05,
+      "loss": 0.5564,
+      "step": 6760
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 3.817054981227177e-05,
+      "loss": 1.0415,
+      "step": 6761
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 3.8167978192665746e-05,
+      "loss": 0.8132,
+      "step": 6762
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 3.8165406573059716e-05,
+      "loss": 1.239,
+      "step": 6763
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 3.8162834953453685e-05,
+      "loss": 0.6997,
+      "step": 6764
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 3.8160263333847655e-05,
+      "loss": 0.5756,
+      "step": 6765
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 3.815769171424163e-05,
+      "loss": 0.7645,
+      "step": 6766
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 3.815512009463561e-05,
+      "loss": 0.7684,
+      "step": 6767
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 3.815254847502958e-05,
+      "loss": 0.595,
+      "step": 6768
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 3.814997685542355e-05,
+      "loss": 1.0294,
+      "step": 6769
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 3.814740523581752e-05,
+      "loss": 0.6699,
+      "step": 6770
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 3.8144833616211487e-05,
+      "loss": 0.7837,
+      "step": 6771
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 3.814226199660546e-05,
+      "loss": 1.1496,
+      "step": 6772
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 3.813969037699944e-05,
+      "loss": 0.6494,
+      "step": 6773
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 3.813711875739341e-05,
+      "loss": 1.0622,
+      "step": 6774
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 3.813454713778738e-05,
+      "loss": 0.6935,
+      "step": 6775
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 3.813197551818135e-05,
+      "loss": 1.2645,
+      "step": 6776
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 3.8129403898575325e-05,
+      "loss": 0.7378,
+      "step": 6777
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 3.81268322789693e-05,
+      "loss": 0.5978,
+      "step": 6778
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 3.812426065936327e-05,
+      "loss": 0.8756,
+      "step": 6779
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 3.812168903975724e-05,
+      "loss": 0.7854,
+      "step": 6780
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 3.811911742015121e-05,
+      "loss": 0.8271,
+      "step": 6781
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 3.811654580054519e-05,
+      "loss": 0.6387,
+      "step": 6782
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 3.811397418093916e-05,
+      "loss": 0.7843,
+      "step": 6783
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 3.811140256133313e-05,
+      "loss": 1.0446,
+      "step": 6784
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 3.81088309417271e-05,
+      "loss": 0.98,
+      "step": 6785
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 3.810625932212107e-05,
+      "loss": 0.6854,
+      "step": 6786
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 3.810368770251504e-05,
+      "loss": 0.5837,
+      "step": 6787
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 3.810111608290902e-05,
+      "loss": 0.8391,
+      "step": 6788
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 3.809854446330299e-05,
+      "loss": 0.6204,
+      "step": 6789
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 3.8095972843696965e-05,
+      "loss": 0.7632,
+      "step": 6790
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 3.8093401224090934e-05,
+      "loss": 0.6978,
+      "step": 6791
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 3.8090829604484904e-05,
+      "loss": 0.7755,
+      "step": 6792
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 3.808825798487888e-05,
+      "loss": 0.8896,
+      "step": 6793
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 3.808568636527285e-05,
+      "loss": 0.7793,
+      "step": 6794
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 3.808311474566682e-05,
+      "loss": 0.5697,
+      "step": 6795
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 3.8080543126060796e-05,
+      "loss": 0.853,
+      "step": 6796
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 3.8077971506454766e-05,
+      "loss": 0.8862,
+      "step": 6797
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 3.807539988684874e-05,
+      "loss": 0.8806,
+      "step": 6798
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 3.807282826724271e-05,
+      "loss": 0.9116,
+      "step": 6799
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 3.807025664763668e-05,
+      "loss": 0.8539,
+      "step": 6800
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 3.806768502803065e-05,
+      "loss": 0.8341,
+      "step": 6801
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 3.806511340842463e-05,
+      "loss": 1.0104,
+      "step": 6802
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 3.80625417888186e-05,
+      "loss": 0.9166,
+      "step": 6803
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 3.8059970169212574e-05,
+      "loss": 0.9408,
+      "step": 6804
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 3.8057398549606544e-05,
+      "loss": 0.5358,
+      "step": 6805
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 3.8054826930000514e-05,
+      "loss": 1.057,
+      "step": 6806
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 3.805225531039448e-05,
+      "loss": 0.8183,
+      "step": 6807
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 3.804968369078846e-05,
+      "loss": 0.66,
+      "step": 6808
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 3.8047112071182436e-05,
+      "loss": 0.9444,
+      "step": 6809
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 3.8044540451576406e-05,
+      "loss": 0.8174,
+      "step": 6810
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 3.8041968831970376e-05,
+      "loss": 0.5044,
+      "step": 6811
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 3.8039397212364345e-05,
+      "loss": 0.6712,
+      "step": 6812
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 3.8036825592758315e-05,
+      "loss": 0.9175,
+      "step": 6813
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 3.80342539731523e-05,
+      "loss": 0.6987,
+      "step": 6814
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 3.803168235354627e-05,
+      "loss": 0.9684,
+      "step": 6815
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 3.802911073394024e-05,
+      "loss": 0.6619,
+      "step": 6816
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 3.802653911433421e-05,
+      "loss": 0.6338,
+      "step": 6817
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 3.802396749472818e-05,
+      "loss": 0.706,
+      "step": 6818
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 3.8021395875122153e-05,
+      "loss": 0.5928,
+      "step": 6819
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 3.801882425551613e-05,
+      "loss": 0.5256,
+      "step": 6820
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 3.80162526359101e-05,
+      "loss": 0.8366,
+      "step": 6821
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 3.801368101630407e-05,
+      "loss": 0.6104,
+      "step": 6822
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 3.801110939669804e-05,
+      "loss": 0.9588,
+      "step": 6823
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 3.8008537777092015e-05,
+      "loss": 0.9304,
+      "step": 6824
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 3.8005966157485985e-05,
+      "loss": 1.0772,
+      "step": 6825
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 3.800339453787996e-05,
+      "loss": 0.8617,
+      "step": 6826
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 3.800082291827393e-05,
+      "loss": 0.9592,
+      "step": 6827
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 3.79982512986679e-05,
+      "loss": 0.8502,
+      "step": 6828
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 3.799567967906188e-05,
+      "loss": 0.9487,
+      "step": 6829
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 3.799310805945585e-05,
+      "loss": 1.296,
+      "step": 6830
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 3.799053643984982e-05,
+      "loss": 0.7075,
+      "step": 6831
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 3.798796482024379e-05,
+      "loss": 1.0849,
+      "step": 6832
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 3.798539320063776e-05,
+      "loss": 0.9229,
+      "step": 6833
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 3.798282158103173e-05,
+      "loss": 0.6463,
+      "step": 6834
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 3.798024996142571e-05,
+      "loss": 0.7561,
+      "step": 6835
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 3.797767834181968e-05,
+      "loss": 0.6421,
+      "step": 6836
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 3.7975106722213655e-05,
+      "loss": 0.746,
+      "step": 6837
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 3.7972535102607625e-05,
+      "loss": 0.8846,
+      "step": 6838
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 3.7969963483001595e-05,
+      "loss": 0.92,
+      "step": 6839
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 3.796739186339557e-05,
+      "loss": 0.7422,
+      "step": 6840
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 3.796482024378954e-05,
+      "loss": 0.5179,
+      "step": 6841
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 3.796224862418351e-05,
+      "loss": 0.8445,
+      "step": 6842
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 3.795967700457749e-05,
+      "loss": 0.7586,
+      "step": 6843
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 3.7957105384971457e-05,
+      "loss": 0.6654,
+      "step": 6844
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 3.795453376536543e-05,
+      "loss": 0.6986,
+      "step": 6845
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 3.79519621457594e-05,
+      "loss": 0.7655,
+      "step": 6846
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 3.794939052615337e-05,
+      "loss": 0.6435,
+      "step": 6847
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 3.794681890654734e-05,
+      "loss": 0.8567,
+      "step": 6848
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 3.794424728694132e-05,
+      "loss": 0.8857,
+      "step": 6849
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 3.794167566733529e-05,
+      "loss": 0.7038,
+      "step": 6850
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 3.7939104047729265e-05,
+      "loss": 0.6763,
+      "step": 6851
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 3.7936532428123234e-05,
+      "loss": 0.6889,
+      "step": 6852
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 3.7933960808517204e-05,
+      "loss": 0.8188,
+      "step": 6853
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 3.7931389188911174e-05,
+      "loss": 0.7532,
+      "step": 6854
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 3.792881756930515e-05,
+      "loss": 0.7679,
+      "step": 6855
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 3.792624594969913e-05,
+      "loss": 0.9163,
+      "step": 6856
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 3.7923674330093096e-05,
+      "loss": 0.8315,
+      "step": 6857
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 3.7921102710487066e-05,
+      "loss": 0.9265,
+      "step": 6858
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 3.7918531090881036e-05,
+      "loss": 0.8202,
+      "step": 6859
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 3.791595947127501e-05,
+      "loss": 0.6866,
+      "step": 6860
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 3.791338785166899e-05,
+      "loss": 0.8318,
+      "step": 6861
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 3.791081623206296e-05,
+      "loss": 1.1303,
+      "step": 6862
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 3.790824461245693e-05,
+      "loss": 0.6078,
+      "step": 6863
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 3.79056729928509e-05,
+      "loss": 0.9194,
+      "step": 6864
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 3.790310137324487e-05,
+      "loss": 0.8066,
+      "step": 6865
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 3.7900529753638844e-05,
+      "loss": 1.0259,
+      "step": 6866
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 3.789795813403282e-05,
+      "loss": 0.7025,
+      "step": 6867
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 3.789538651442679e-05,
+      "loss": 0.4348,
+      "step": 6868
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 3.789281489482076e-05,
+      "loss": 0.6623,
+      "step": 6869
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 3.789024327521473e-05,
+      "loss": 0.6653,
+      "step": 6870
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 3.7887671655608706e-05,
+      "loss": 1.0534,
+      "step": 6871
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 3.7885100036002676e-05,
+      "loss": 0.7665,
+      "step": 6872
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 3.788252841639665e-05,
+      "loss": 0.6545,
+      "step": 6873
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 3.787995679679062e-05,
+      "loss": 0.6554,
+      "step": 6874
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 3.787738517718459e-05,
+      "loss": 0.6603,
+      "step": 6875
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 3.787481355757857e-05,
+      "loss": 0.6001,
+      "step": 6876
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 3.787224193797254e-05,
+      "loss": 0.9366,
+      "step": 6877
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 3.786967031836651e-05,
+      "loss": 0.777,
+      "step": 6878
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 3.7867098698760484e-05,
+      "loss": 1.0479,
+      "step": 6879
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 3.786452707915445e-05,
+      "loss": 0.5697,
+      "step": 6880
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 3.786195545954842e-05,
+      "loss": 0.8376,
+      "step": 6881
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 3.78593838399424e-05,
+      "loss": 0.6987,
+      "step": 6882
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 3.785681222033637e-05,
+      "loss": 0.8099,
+      "step": 6883
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 3.785424060073034e-05,
+      "loss": 0.8793,
+      "step": 6884
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 3.7851668981124315e-05,
+      "loss": 0.744,
+      "step": 6885
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 3.7849097361518285e-05,
+      "loss": 0.7578,
+      "step": 6886
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 3.784652574191226e-05,
+      "loss": 0.9261,
+      "step": 6887
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 3.784395412230623e-05,
+      "loss": 0.7966,
+      "step": 6888
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 3.78413825027002e-05,
+      "loss": 0.7875,
+      "step": 6889
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 3.783881088309417e-05,
+      "loss": 0.5361,
+      "step": 6890
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 3.783623926348815e-05,
+      "loss": 0.5884,
+      "step": 6891
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 3.7833667643882123e-05,
+      "loss": 0.8073,
+      "step": 6892
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 3.783109602427609e-05,
+      "loss": 0.8531,
+      "step": 6893
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 3.782852440467006e-05,
+      "loss": 1.1927,
+      "step": 6894
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 3.782595278506403e-05,
+      "loss": 0.7267,
+      "step": 6895
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 3.7823381165458e-05,
+      "loss": 0.6457,
+      "step": 6896
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 3.782080954585198e-05,
+      "loss": 0.6672,
+      "step": 6897
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 3.7818237926245955e-05,
+      "loss": 0.6375,
+      "step": 6898
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 3.7815666306639925e-05,
+      "loss": 0.8491,
+      "step": 6899
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 3.7813094687033894e-05,
+      "loss": 0.8194,
+      "step": 6900
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 3.7810523067427864e-05,
+      "loss": 0.9465,
+      "step": 6901
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 3.780795144782184e-05,
+      "loss": 0.6862,
+      "step": 6902
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 3.780537982821582e-05,
+      "loss": 0.7147,
+      "step": 6903
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 3.780280820860979e-05,
+      "loss": 0.7087,
+      "step": 6904
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 3.7800236589003756e-05,
+      "loss": 0.8163,
+      "step": 6905
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 3.7797664969397726e-05,
+      "loss": 1.0708,
+      "step": 6906
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 3.77950933497917e-05,
+      "loss": 0.8563,
+      "step": 6907
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 3.779252173018567e-05,
+      "loss": 0.9553,
+      "step": 6908
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 3.778995011057965e-05,
+      "loss": 0.5749,
+      "step": 6909
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 3.778737849097362e-05,
+      "loss": 0.7837,
+      "step": 6910
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 3.778480687136759e-05,
+      "loss": 0.7643,
+      "step": 6911
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 3.778223525176156e-05,
+      "loss": 0.9227,
+      "step": 6912
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 3.7779663632155534e-05,
+      "loss": 0.8698,
+      "step": 6913
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 3.7777092012549504e-05,
+      "loss": 0.6416,
+      "step": 6914
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 3.777452039294348e-05,
+      "loss": 0.754,
+      "step": 6915
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 3.777194877333745e-05,
+      "loss": 0.686,
+      "step": 6916
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 3.776937715373142e-05,
+      "loss": 0.5898,
+      "step": 6917
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 3.7766805534125396e-05,
+      "loss": 0.7288,
+      "step": 6918
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 3.7764233914519366e-05,
+      "loss": 0.9203,
+      "step": 6919
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 3.7761662294913336e-05,
+      "loss": 0.705,
+      "step": 6920
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 3.775909067530731e-05,
+      "loss": 0.7065,
+      "step": 6921
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 3.775651905570128e-05,
+      "loss": 0.7696,
+      "step": 6922
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 3.775394743609526e-05,
+      "loss": 0.6937,
+      "step": 6923
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 3.775137581648923e-05,
+      "loss": 1.0339,
+      "step": 6924
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 3.77488041968832e-05,
+      "loss": 0.8882,
+      "step": 6925
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 3.774623257727717e-05,
+      "loss": 0.5685,
+      "step": 6926
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 3.7743660957671144e-05,
+      "loss": 0.7219,
+      "step": 6927
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 3.7741089338065113e-05,
+      "loss": 0.6742,
+      "step": 6928
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 3.773851771845909e-05,
+      "loss": 0.8769,
+      "step": 6929
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 3.773594609885306e-05,
+      "loss": 1.2538,
+      "step": 6930
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 3.773337447924703e-05,
+      "loss": 0.979,
+      "step": 6931
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 3.7730802859641e-05,
+      "loss": 1.2575,
+      "step": 6932
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 3.7728231240034975e-05,
+      "loss": 0.6575,
+      "step": 6933
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 3.772565962042895e-05,
+      "loss": 0.7092,
+      "step": 6934
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 3.772308800082292e-05,
+      "loss": 0.6261,
+      "step": 6935
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 3.772051638121689e-05,
+      "loss": 1.1583,
+      "step": 6936
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 3.771794476161086e-05,
+      "loss": 0.6863,
+      "step": 6937
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 3.771537314200483e-05,
+      "loss": 0.6334,
+      "step": 6938
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 3.7712801522398814e-05,
+      "loss": 0.8397,
+      "step": 6939
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 3.7710229902792784e-05,
+      "loss": 0.7721,
+      "step": 6940
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 3.770765828318675e-05,
+      "loss": 0.8505,
+      "step": 6941
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 3.770508666358072e-05,
+      "loss": 0.7448,
+      "step": 6942
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 3.770251504397469e-05,
+      "loss": 0.7178,
+      "step": 6943
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 3.769994342436867e-05,
+      "loss": 0.7384,
+      "step": 6944
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 3.7697371804762645e-05,
+      "loss": 1.0717,
+      "step": 6945
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 3.7694800185156615e-05,
+      "loss": 0.8098,
+      "step": 6946
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 3.7692228565550585e-05,
+      "loss": 0.9233,
+      "step": 6947
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 3.7689656945944555e-05,
+      "loss": 0.8642,
+      "step": 6948
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 3.768708532633853e-05,
+      "loss": 0.8366,
+      "step": 6949
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 3.76845137067325e-05,
+      "loss": 0.915,
+      "step": 6950
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 3.768194208712648e-05,
+      "loss": 0.8505,
+      "step": 6951
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 3.767937046752045e-05,
+      "loss": 0.923,
+      "step": 6952
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 3.7676798847914417e-05,
+      "loss": 1.0044,
+      "step": 6953
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 3.767422722830839e-05,
+      "loss": 1.0495,
+      "step": 6954
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 3.767165560870236e-05,
+      "loss": 0.7058,
+      "step": 6955
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 3.766908398909633e-05,
+      "loss": 0.9045,
+      "step": 6956
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 3.766651236949031e-05,
+      "loss": 0.9322,
+      "step": 6957
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 3.766394074988428e-05,
+      "loss": 0.5708,
+      "step": 6958
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 3.766136913027825e-05,
+      "loss": 0.8225,
+      "step": 6959
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 3.7658797510672225e-05,
+      "loss": 0.6146,
+      "step": 6960
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 3.7656225891066194e-05,
+      "loss": 0.6893,
+      "step": 6961
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 3.7653654271460164e-05,
+      "loss": 0.7458,
+      "step": 6962
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 3.765108265185414e-05,
+      "loss": 0.8167,
+      "step": 6963
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 3.764851103224811e-05,
+      "loss": 0.515,
+      "step": 6964
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 3.764593941264209e-05,
+      "loss": 0.6892,
+      "step": 6965
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 3.7643367793036056e-05,
+      "loss": 0.8311,
+      "step": 6966
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 3.7640796173430026e-05,
+      "loss": 0.8284,
+      "step": 6967
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 3.7638224553823996e-05,
+      "loss": 0.5544,
+      "step": 6968
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 3.763565293421797e-05,
+      "loss": 0.577,
+      "step": 6969
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 3.763308131461195e-05,
+      "loss": 0.6219,
+      "step": 6970
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 3.763050969500592e-05,
+      "loss": 1.1968,
+      "step": 6971
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 3.762793807539989e-05,
+      "loss": 0.7358,
+      "step": 6972
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 3.762536645579386e-05,
+      "loss": 0.4342,
+      "step": 6973
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 3.762279483618783e-05,
+      "loss": 0.8194,
+      "step": 6974
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 3.7620223216581804e-05,
+      "loss": 0.649,
+      "step": 6975
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 3.761765159697578e-05,
+      "loss": 0.793,
+      "step": 6976
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 3.761507997736975e-05,
+      "loss": 0.9562,
+      "step": 6977
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 3.761250835776372e-05,
+      "loss": 1.0414,
+      "step": 6978
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 3.760993673815769e-05,
+      "loss": 1.0965,
+      "step": 6979
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 3.7607365118551666e-05,
+      "loss": 0.9386,
+      "step": 6980
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 3.760479349894564e-05,
+      "loss": 0.9147,
+      "step": 6981
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 3.760222187933961e-05,
+      "loss": 1.0524,
+      "step": 6982
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 3.759965025973358e-05,
+      "loss": 0.6497,
+      "step": 6983
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 3.759707864012755e-05,
+      "loss": 0.7139,
+      "step": 6984
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 3.759450702052153e-05,
+      "loss": 0.5728,
+      "step": 6985
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 3.7591935400915504e-05,
+      "loss": 0.8799,
+      "step": 6986
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 3.7589363781309474e-05,
+      "loss": 0.7288,
+      "step": 6987
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 3.7586792161703444e-05,
+      "loss": 0.9605,
+      "step": 6988
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 3.758422054209741e-05,
+      "loss": 0.7409,
+      "step": 6989
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 3.758164892249138e-05,
+      "loss": 0.691,
+      "step": 6990
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 3.757907730288536e-05,
+      "loss": 0.7579,
+      "step": 6991
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 3.7576505683279336e-05,
+      "loss": 0.8663,
+      "step": 6992
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 3.7573934063673306e-05,
+      "loss": 0.6757,
+      "step": 6993
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 3.7571362444067275e-05,
+      "loss": 1.0699,
+      "step": 6994
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 3.7568790824461245e-05,
+      "loss": 0.7284,
+      "step": 6995
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 3.756621920485522e-05,
+      "loss": 0.8733,
+      "step": 6996
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 3.756364758524919e-05,
+      "loss": 0.7823,
+      "step": 6997
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 3.756107596564317e-05,
+      "loss": 0.686,
+      "step": 6998
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 3.755850434603714e-05,
+      "loss": 0.5931,
+      "step": 6999
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 3.755593272643111e-05,
+      "loss": 0.7504,
+      "step": 7000
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 3.7553361106825083e-05,
+      "loss": 0.6386,
+      "step": 7001
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 3.755078948721905e-05,
+      "loss": 0.7126,
+      "step": 7002
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 3.754821786761302e-05,
+      "loss": 0.7837,
+      "step": 7003
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 3.7545646248007e-05,
+      "loss": 0.8446,
+      "step": 7004
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 3.754307462840097e-05,
+      "loss": 0.9181,
+      "step": 7005
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 3.754050300879494e-05,
+      "loss": 0.9349,
+      "step": 7006
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 3.7537931389188915e-05,
+      "loss": 0.801,
+      "step": 7007
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 3.7535359769582885e-05,
+      "loss": 0.6605,
+      "step": 7008
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 3.7532788149976854e-05,
+      "loss": 0.7364,
+      "step": 7009
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 3.753021653037083e-05,
+      "loss": 0.7824,
+      "step": 7010
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 3.75276449107648e-05,
+      "loss": 0.9088,
+      "step": 7011
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 3.752507329115878e-05,
+      "loss": 0.9077,
+      "step": 7012
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 3.752250167155275e-05,
+      "loss": 0.6806,
+      "step": 7013
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 3.7519930051946716e-05,
+      "loss": 0.7787,
+      "step": 7014
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 3.7517358432340686e-05,
+      "loss": 0.9397,
+      "step": 7015
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 3.751478681273466e-05,
+      "loss": 1.1388,
+      "step": 7016
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 3.751221519312864e-05,
+      "loss": 0.6036,
+      "step": 7017
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 3.750964357352261e-05,
+      "loss": 0.7492,
+      "step": 7018
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 3.750707195391658e-05,
+      "loss": 0.7816,
+      "step": 7019
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 3.750450033431055e-05,
+      "loss": 0.7367,
+      "step": 7020
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 3.750192871470452e-05,
+      "loss": 0.6308,
+      "step": 7021
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 3.7499357095098494e-05,
+      "loss": 0.4329,
+      "step": 7022
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 3.749678547549247e-05,
+      "loss": 0.7146,
+      "step": 7023
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 3.749421385588644e-05,
+      "loss": 0.8419,
+      "step": 7024
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 3.749164223628041e-05,
+      "loss": 0.968,
+      "step": 7025
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 3.748907061667438e-05,
+      "loss": 0.6498,
+      "step": 7026
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 3.7486498997068356e-05,
+      "loss": 0.6948,
+      "step": 7027
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 3.748392737746233e-05,
+      "loss": 0.551,
+      "step": 7028
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 3.74813557578563e-05,
+      "loss": 0.6979,
+      "step": 7029
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 3.747878413825027e-05,
+      "loss": 0.7294,
+      "step": 7030
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 3.747621251864424e-05,
+      "loss": 0.6955,
+      "step": 7031
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 3.747364089903822e-05,
+      "loss": 0.8699,
+      "step": 7032
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 3.747106927943219e-05,
+      "loss": 0.7801,
+      "step": 7033
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 3.7468497659826164e-05,
+      "loss": 0.9363,
+      "step": 7034
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 3.7465926040220134e-05,
+      "loss": 0.8774,
+      "step": 7035
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 3.7463354420614104e-05,
+      "loss": 0.8942,
+      "step": 7036
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 3.746078280100807e-05,
+      "loss": 0.9074,
+      "step": 7037
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 3.745821118140205e-05,
+      "loss": 0.6217,
+      "step": 7038
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 3.745563956179602e-05,
+      "loss": 0.667,
+      "step": 7039
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 3.7453067942189996e-05,
+      "loss": 0.9889,
+      "step": 7040
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 3.7450496322583966e-05,
+      "loss": 0.8216,
+      "step": 7041
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 3.7447924702977935e-05,
+      "loss": 0.6831,
+      "step": 7042
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 3.744535308337191e-05,
+      "loss": 0.547,
+      "step": 7043
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 3.744278146376588e-05,
+      "loss": 0.7279,
+      "step": 7044
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 3.744020984415985e-05,
+      "loss": 0.8174,
+      "step": 7045
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 3.743763822455383e-05,
+      "loss": 0.9223,
+      "step": 7046
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 3.74350666049478e-05,
+      "loss": 0.9034,
+      "step": 7047
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 3.7432494985341774e-05,
+      "loss": 0.9407,
+      "step": 7048
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 3.7429923365735743e-05,
+      "loss": 0.8939,
+      "step": 7049
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 3.742735174612971e-05,
+      "loss": 0.9741,
+      "step": 7050
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 3.742478012652368e-05,
+      "loss": 0.7884,
+      "step": 7051
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 3.742220850691766e-05,
+      "loss": 0.737,
+      "step": 7052
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 3.741963688731163e-05,
+      "loss": 0.7351,
+      "step": 7053
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 3.7417065267705605e-05,
+      "loss": 0.5856,
+      "step": 7054
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 3.7414493648099575e-05,
+      "loss": 0.7455,
+      "step": 7055
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 3.7411922028493545e-05,
+      "loss": 0.5108,
+      "step": 7056
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 3.7409350408887515e-05,
+      "loss": 0.719,
+      "step": 7057
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 3.740677878928149e-05,
+      "loss": 0.6741,
+      "step": 7058
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 3.740420716967547e-05,
+      "loss": 0.7564,
+      "step": 7059
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 3.740163555006944e-05,
+      "loss": 0.626,
+      "step": 7060
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 3.739906393046341e-05,
+      "loss": 0.8788,
+      "step": 7061
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 3.7396492310857376e-05,
+      "loss": 0.821,
+      "step": 7062
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 3.7393920691251346e-05,
+      "loss": 0.7856,
+      "step": 7063
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 3.739134907164533e-05,
+      "loss": 0.8145,
+      "step": 7064
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 3.73887774520393e-05,
+      "loss": 0.7158,
+      "step": 7065
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 3.738620583243327e-05,
+      "loss": 0.668,
+      "step": 7066
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 3.738363421282724e-05,
+      "loss": 0.8953,
+      "step": 7067
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 3.738106259322121e-05,
+      "loss": 0.9566,
+      "step": 7068
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 3.7378490973615185e-05,
+      "loss": 0.6005,
+      "step": 7069
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 3.737591935400916e-05,
+      "loss": 0.9862,
+      "step": 7070
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 3.737334773440313e-05,
+      "loss": 0.857,
+      "step": 7071
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 3.73707761147971e-05,
+      "loss": 0.9953,
+      "step": 7072
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 3.736820449519107e-05,
+      "loss": 1.1999,
+      "step": 7073
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 3.7365632875585047e-05,
+      "loss": 0.9172,
+      "step": 7074
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 3.7363061255979016e-05,
+      "loss": 0.9389,
+      "step": 7075
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 3.736048963637299e-05,
+      "loss": 1.0498,
+      "step": 7076
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 3.735791801676696e-05,
+      "loss": 0.9817,
+      "step": 7077
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 3.735534639716093e-05,
+      "loss": 0.6245,
+      "step": 7078
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 3.735277477755491e-05,
+      "loss": 0.9108,
+      "step": 7079
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 3.735020315794888e-05,
+      "loss": 0.6889,
+      "step": 7080
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 3.734763153834285e-05,
+      "loss": 0.7527,
+      "step": 7081
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 3.7345059918736824e-05,
+      "loss": 0.6427,
+      "step": 7082
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 3.7342488299130794e-05,
+      "loss": 0.8551,
+      "step": 7083
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 3.7339916679524764e-05,
+      "loss": 0.8153,
+      "step": 7084
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 3.733734505991874e-05,
+      "loss": 0.8423,
+      "step": 7085
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 3.733477344031271e-05,
+      "loss": 0.6418,
+      "step": 7086
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 3.733220182070668e-05,
+      "loss": 0.7955,
+      "step": 7087
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 3.7329630201100656e-05,
+      "loss": 0.94,
+      "step": 7088
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 3.7327058581494626e-05,
+      "loss": 1.032,
+      "step": 7089
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 3.73244869618886e-05,
+      "loss": 1.0352,
+      "step": 7090
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 3.732191534228257e-05,
+      "loss": 0.8431,
+      "step": 7091
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 3.731934372267654e-05,
+      "loss": 0.704,
+      "step": 7092
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 3.731677210307051e-05,
+      "loss": 0.5849,
+      "step": 7093
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 3.731420048346449e-05,
+      "loss": 1.0814,
+      "step": 7094
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 3.7311628863858464e-05,
+      "loss": 0.8917,
+      "step": 7095
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 3.7309057244252434e-05,
+      "loss": 0.7685,
+      "step": 7096
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 3.7306485624646404e-05,
+      "loss": 0.8851,
+      "step": 7097
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 3.730391400504037e-05,
+      "loss": 0.9385,
+      "step": 7098
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 3.730134238543434e-05,
+      "loss": 1.1047,
+      "step": 7099
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 3.729877076582832e-05,
+      "loss": 0.8234,
+      "step": 7100
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 3.7296199146222296e-05,
+      "loss": 1.1799,
+      "step": 7101
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 3.7293627526616266e-05,
+      "loss": 0.7086,
+      "step": 7102
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 3.7291055907010235e-05,
+      "loss": 0.8817,
+      "step": 7103
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 3.7288484287404205e-05,
+      "loss": 0.7759,
+      "step": 7104
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 3.728591266779818e-05,
+      "loss": 0.9083,
+      "step": 7105
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 3.728334104819216e-05,
+      "loss": 0.9096,
+      "step": 7106
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 3.728076942858613e-05,
+      "loss": 0.7941,
+      "step": 7107
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 3.72781978089801e-05,
+      "loss": 0.6872,
+      "step": 7108
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 3.727562618937407e-05,
+      "loss": 0.5792,
+      "step": 7109
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 3.727305456976804e-05,
+      "loss": 0.8385,
+      "step": 7110
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 3.727048295016201e-05,
+      "loss": 0.7511,
+      "step": 7111
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 3.726791133055599e-05,
+      "loss": 0.7793,
+      "step": 7112
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 3.726533971094996e-05,
+      "loss": 0.9609,
+      "step": 7113
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 3.726276809134393e-05,
+      "loss": 0.839,
+      "step": 7114
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 3.72601964717379e-05,
+      "loss": 0.4809,
+      "step": 7115
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 3.7257624852131875e-05,
+      "loss": 0.7556,
+      "step": 7116
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 3.7255053232525845e-05,
+      "loss": 0.8093,
+      "step": 7117
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 3.725248161291982e-05,
+      "loss": 0.6322,
+      "step": 7118
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 3.724990999331379e-05,
+      "loss": 0.8888,
+      "step": 7119
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 3.724733837370776e-05,
+      "loss": 0.9511,
+      "step": 7120
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 3.724476675410174e-05,
+      "loss": 1.1379,
+      "step": 7121
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 3.724219513449571e-05,
+      "loss": 0.9185,
+      "step": 7122
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 3.7239623514889676e-05,
+      "loss": 0.7406,
+      "step": 7123
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 3.723705189528365e-05,
+      "loss": 0.7515,
+      "step": 7124
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 3.723448027567762e-05,
+      "loss": 0.8708,
+      "step": 7125
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 3.72319086560716e-05,
+      "loss": 0.9726,
+      "step": 7126
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 3.722933703646557e-05,
+      "loss": 0.8179,
+      "step": 7127
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 3.722676541685954e-05,
+      "loss": 0.9869,
+      "step": 7128
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 3.722419379725351e-05,
+      "loss": 0.8814,
+      "step": 7129
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 3.7221622177647485e-05,
+      "loss": 1.0104,
+      "step": 7130
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 3.7219050558041454e-05,
+      "loss": 0.5834,
+      "step": 7131
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 3.721647893843543e-05,
+      "loss": 0.6679,
+      "step": 7132
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 3.72139073188294e-05,
+      "loss": 0.6006,
+      "step": 7133
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 3.721133569922337e-05,
+      "loss": 0.7987,
+      "step": 7134
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 3.7208764079617346e-05,
+      "loss": 0.5681,
+      "step": 7135
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 3.7206192460011316e-05,
+      "loss": 0.9963,
+      "step": 7136
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 3.720362084040529e-05,
+      "loss": 0.7064,
+      "step": 7137
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 3.720104922079926e-05,
+      "loss": 0.6865,
+      "step": 7138
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 3.719847760119323e-05,
+      "loss": 0.8141,
+      "step": 7139
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 3.71959059815872e-05,
+      "loss": 0.5732,
+      "step": 7140
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 3.719333436198118e-05,
+      "loss": 0.8374,
+      "step": 7141
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 3.7190762742375155e-05,
+      "loss": 0.7201,
+      "step": 7142
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 3.7188191122769124e-05,
+      "loss": 0.9465,
+      "step": 7143
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 3.7185619503163094e-05,
+      "loss": 0.6867,
+      "step": 7144
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 3.7183047883557064e-05,
+      "loss": 0.6271,
+      "step": 7145
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 3.718047626395103e-05,
+      "loss": 0.5561,
+      "step": 7146
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 3.717790464434501e-05,
+      "loss": 1.1639,
+      "step": 7147
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 3.7175333024738986e-05,
+      "loss": 0.8392,
+      "step": 7148
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 3.7172761405132956e-05,
+      "loss": 0.7711,
+      "step": 7149
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 3.7170189785526926e-05,
+      "loss": 0.7603,
+      "step": 7150
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 3.7167618165920895e-05,
+      "loss": 0.7488,
+      "step": 7151
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 3.716504654631487e-05,
+      "loss": 0.9288,
+      "step": 7152
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 3.716247492670885e-05,
+      "loss": 0.5814,
+      "step": 7153
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 3.715990330710282e-05,
+      "loss": 0.7244,
+      "step": 7154
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 3.715733168749679e-05,
+      "loss": 0.9683,
+      "step": 7155
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 3.715476006789076e-05,
+      "loss": 0.5719,
+      "step": 7156
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 3.7152188448284734e-05,
+      "loss": 0.8646,
+      "step": 7157
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 3.7149616828678703e-05,
+      "loss": 0.6113,
+      "step": 7158
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 3.714704520907268e-05,
+      "loss": 0.8921,
+      "step": 7159
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 3.714447358946665e-05,
+      "loss": 0.5937,
+      "step": 7160
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 3.714190196986062e-05,
+      "loss": 0.8923,
+      "step": 7161
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 3.713933035025459e-05,
+      "loss": 1.0609,
+      "step": 7162
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 3.7136758730648565e-05,
+      "loss": 0.8203,
+      "step": 7163
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 3.7134187111042535e-05,
+      "loss": 0.8487,
+      "step": 7164
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 3.713161549143651e-05,
+      "loss": 0.7869,
+      "step": 7165
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 3.712904387183048e-05,
+      "loss": 0.6082,
+      "step": 7166
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 3.712647225222445e-05,
+      "loss": 0.71,
+      "step": 7167
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 3.712390063261843e-05,
+      "loss": 0.8165,
+      "step": 7168
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 3.71213290130124e-05,
+      "loss": 0.7655,
+      "step": 7169
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 3.711875739340637e-05,
+      "loss": 0.7048,
+      "step": 7170
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 3.711618577380034e-05,
+      "loss": 0.8109,
+      "step": 7171
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 3.711361415419431e-05,
+      "loss": 1.229,
+      "step": 7172
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 3.711104253458829e-05,
+      "loss": 0.6946,
+      "step": 7173
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 3.710847091498226e-05,
+      "loss": 0.7091,
+      "step": 7174
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 3.710589929537623e-05,
+      "loss": 0.771,
+      "step": 7175
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 3.71033276757702e-05,
+      "loss": 0.7487,
+      "step": 7176
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 3.7100756056164175e-05,
+      "loss": 0.6755,
+      "step": 7177
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 3.7098184436558145e-05,
+      "loss": 0.7061,
+      "step": 7178
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 3.709561281695212e-05,
+      "loss": 0.7722,
+      "step": 7179
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 3.709304119734609e-05,
+      "loss": 0.8903,
+      "step": 7180
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 3.709046957774006e-05,
+      "loss": 0.8687,
+      "step": 7181
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 3.708789795813403e-05,
+      "loss": 1.1843,
+      "step": 7182
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 3.7085326338528007e-05,
+      "loss": 1.1135,
+      "step": 7183
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 3.708275471892198e-05,
+      "loss": 1.1423,
+      "step": 7184
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 3.708018309931595e-05,
+      "loss": 0.7006,
+      "step": 7185
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 3.707761147970992e-05,
+      "loss": 0.824,
+      "step": 7186
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 3.707503986010389e-05,
+      "loss": 0.831,
+      "step": 7187
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 3.707246824049786e-05,
+      "loss": 0.8331,
+      "step": 7188
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 3.7069896620891845e-05,
+      "loss": 0.7692,
+      "step": 7189
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 3.7067325001285815e-05,
+      "loss": 0.7815,
+      "step": 7190
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 3.7064753381679784e-05,
+      "loss": 0.9762,
+      "step": 7191
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 3.7062181762073754e-05,
+      "loss": 0.8712,
+      "step": 7192
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 3.7059610142467724e-05,
+      "loss": 0.5567,
+      "step": 7193
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 3.70570385228617e-05,
+      "loss": 0.9898,
+      "step": 7194
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 3.705446690325568e-05,
+      "loss": 0.9387,
+      "step": 7195
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 3.7051895283649646e-05,
+      "loss": 1.1763,
+      "step": 7196
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 3.7049323664043616e-05,
+      "loss": 0.7283,
+      "step": 7197
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 3.7046752044437586e-05,
+      "loss": 0.7128,
+      "step": 7198
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 3.704418042483156e-05,
+      "loss": 0.5836,
+      "step": 7199
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 3.704160880522553e-05,
+      "loss": 0.7547,
+      "step": 7200
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 3.703903718561951e-05,
+      "loss": 0.6368,
+      "step": 7201
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 3.703646556601348e-05,
+      "loss": 0.6999,
+      "step": 7202
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 3.703389394640745e-05,
+      "loss": 0.9733,
+      "step": 7203
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 3.7031322326801424e-05,
+      "loss": 0.8746,
+      "step": 7204
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 3.7028750707195394e-05,
+      "loss": 0.6236,
+      "step": 7205
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 3.7026179087589364e-05,
+      "loss": 1.0523,
+      "step": 7206
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 3.702360746798334e-05,
+      "loss": 0.9116,
+      "step": 7207
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 3.702103584837731e-05,
+      "loss": 0.6535,
+      "step": 7208
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 3.701846422877128e-05,
+      "loss": 0.8935,
+      "step": 7209
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 3.7015892609165256e-05,
+      "loss": 0.801,
+      "step": 7210
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 3.7013320989559226e-05,
+      "loss": 0.9884,
+      "step": 7211
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 3.7010749369953195e-05,
+      "loss": 0.526,
+      "step": 7212
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 3.700817775034717e-05,
+      "loss": 0.6615,
+      "step": 7213
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 3.700560613074114e-05,
+      "loss": 1.0076,
+      "step": 7214
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 3.700303451113512e-05,
+      "loss": 0.536,
+      "step": 7215
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 3.700046289152909e-05,
+      "loss": 0.9047,
+      "step": 7216
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 3.699789127192306e-05,
+      "loss": 0.7669,
+      "step": 7217
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 3.699531965231703e-05,
+      "loss": 0.7456,
+      "step": 7218
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 3.6992748032711e-05,
+      "loss": 0.902,
+      "step": 7219
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 3.699017641310498e-05,
+      "loss": 0.8045,
+      "step": 7220
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 3.698760479349895e-05,
+      "loss": 0.9678,
+      "step": 7221
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 3.698503317389292e-05,
+      "loss": 0.9114,
+      "step": 7222
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 3.698246155428689e-05,
+      "loss": 0.7642,
+      "step": 7223
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 3.697988993468086e-05,
+      "loss": 0.7557,
+      "step": 7224
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 3.6977318315074835e-05,
+      "loss": 0.8358,
+      "step": 7225
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 3.697474669546881e-05,
+      "loss": 0.5935,
+      "step": 7226
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 3.697217507586278e-05,
+      "loss": 0.4586,
+      "step": 7227
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 3.696960345625675e-05,
+      "loss": 0.4337,
+      "step": 7228
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 3.696703183665072e-05,
+      "loss": 0.7101,
+      "step": 7229
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 3.69644602170447e-05,
+      "loss": 0.7033,
+      "step": 7230
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 3.6961888597438673e-05,
+      "loss": 0.5412,
+      "step": 7231
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 3.695931697783264e-05,
+      "loss": 0.915,
+      "step": 7232
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 3.695674535822661e-05,
+      "loss": 1.0032,
+      "step": 7233
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 3.695417373862058e-05,
+      "loss": 0.6048,
+      "step": 7234
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 3.695160211901456e-05,
+      "loss": 0.482,
+      "step": 7235
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 3.694903049940853e-05,
+      "loss": 0.7121,
+      "step": 7236
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 3.6946458879802505e-05,
+      "loss": 0.7823,
+      "step": 7237
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 3.6943887260196475e-05,
+      "loss": 0.7038,
+      "step": 7238
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 3.6941315640590444e-05,
+      "loss": 0.6519,
+      "step": 7239
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 3.6938744020984414e-05,
+      "loss": 0.7757,
+      "step": 7240
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 3.693617240137839e-05,
+      "loss": 0.6256,
+      "step": 7241
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 3.693360078177236e-05,
+      "loss": 0.5399,
+      "step": 7242
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 3.693102916216634e-05,
+      "loss": 1.1216,
+      "step": 7243
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 3.6928457542560306e-05,
+      "loss": 1.2053,
+      "step": 7244
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 3.6925885922954276e-05,
+      "loss": 0.7556,
+      "step": 7245
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 3.692331430334825e-05,
+      "loss": 0.6352,
+      "step": 7246
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 3.692074268374222e-05,
+      "loss": 0.7807,
+      "step": 7247
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 3.691817106413619e-05,
+      "loss": 1.0555,
+      "step": 7248
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 3.691559944453017e-05,
+      "loss": 1.0304,
+      "step": 7249
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 3.691302782492414e-05,
+      "loss": 0.5731,
+      "step": 7250
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 3.6910456205318115e-05,
+      "loss": 0.412,
+      "step": 7251
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 3.6907884585712084e-05,
+      "loss": 0.5222,
+      "step": 7252
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 3.6905312966106054e-05,
+      "loss": 0.7487,
+      "step": 7253
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 3.6902741346500024e-05,
+      "loss": 0.6874,
+      "step": 7254
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 3.6900169726894e-05,
+      "loss": 0.5768,
+      "step": 7255
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 3.689759810728797e-05,
+      "loss": 0.6484,
+      "step": 7256
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 3.6895026487681946e-05,
+      "loss": 0.8448,
+      "step": 7257
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 3.6892454868075916e-05,
+      "loss": 0.8353,
+      "step": 7258
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 3.6889883248469886e-05,
+      "loss": 0.9871,
+      "step": 7259
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 3.6887311628863855e-05,
+      "loss": 0.7684,
+      "step": 7260
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 3.688474000925783e-05,
+      "loss": 0.7114,
+      "step": 7261
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 3.688216838965181e-05,
+      "loss": 0.7782,
+      "step": 7262
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 3.687959677004578e-05,
+      "loss": 0.8157,
+      "step": 7263
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 3.687702515043975e-05,
+      "loss": 0.7185,
+      "step": 7264
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 3.687445353083372e-05,
+      "loss": 0.9788,
+      "step": 7265
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 3.687188191122769e-05,
+      "loss": 0.6207,
+      "step": 7266
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 3.686931029162167e-05,
+      "loss": 1.0274,
+      "step": 7267
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 3.686673867201564e-05,
+      "loss": 0.6443,
+      "step": 7268
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 3.686416705240961e-05,
+      "loss": 0.67,
+      "step": 7269
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 3.686159543280358e-05,
+      "loss": 1.101,
+      "step": 7270
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 3.685902381319755e-05,
+      "loss": 1.0339,
+      "step": 7271
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 3.6856452193591525e-05,
+      "loss": 0.8125,
+      "step": 7272
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 3.68538805739855e-05,
+      "loss": 0.5258,
+      "step": 7273
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 3.685130895437947e-05,
+      "loss": 0.7699,
+      "step": 7274
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 3.684873733477344e-05,
+      "loss": 0.8451,
+      "step": 7275
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 3.684616571516741e-05,
+      "loss": 0.7476,
+      "step": 7276
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 3.684359409556139e-05,
+      "loss": 1.0336,
+      "step": 7277
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 3.684102247595536e-05,
+      "loss": 0.8912,
+      "step": 7278
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 3.6838450856349334e-05,
+      "loss": 0.95,
+      "step": 7279
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 3.68358792367433e-05,
+      "loss": 0.7374,
+      "step": 7280
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 3.683330761713727e-05,
+      "loss": 0.7358,
+      "step": 7281
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 3.683073599753125e-05,
+      "loss": 0.7598,
+      "step": 7282
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 3.682816437792522e-05,
+      "loss": 0.7483,
+      "step": 7283
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 3.682559275831919e-05,
+      "loss": 1.08,
+      "step": 7284
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 3.6823021138713165e-05,
+      "loss": 0.6436,
+      "step": 7285
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 3.6820449519107135e-05,
+      "loss": 0.8208,
+      "step": 7286
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 3.6817877899501105e-05,
+      "loss": 0.7053,
+      "step": 7287
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 3.681530627989508e-05,
+      "loss": 0.8823,
+      "step": 7288
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 3.681273466028905e-05,
+      "loss": 0.9204,
+      "step": 7289
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 3.681016304068303e-05,
+      "loss": 0.67,
+      "step": 7290
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 3.6807591421077e-05,
+      "loss": 0.7572,
+      "step": 7291
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 3.6805019801470967e-05,
+      "loss": 0.5711,
+      "step": 7292
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 3.680244818186494e-05,
+      "loss": 0.6453,
+      "step": 7293
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 3.679987656225891e-05,
+      "loss": 0.7389,
+      "step": 7294
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 3.679730494265288e-05,
+      "loss": 0.7349,
+      "step": 7295
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 3.679473332304686e-05,
+      "loss": 0.4815,
+      "step": 7296
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 3.679216170344083e-05,
+      "loss": 0.7063,
+      "step": 7297
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 3.6789590083834805e-05,
+      "loss": 0.7216,
+      "step": 7298
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 3.6787018464228775e-05,
+      "loss": 0.7885,
+      "step": 7299
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 3.6784446844622744e-05,
+      "loss": 0.5635,
+      "step": 7300
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 3.6781875225016714e-05,
+      "loss": 0.7426,
+      "step": 7301
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 3.677930360541069e-05,
+      "loss": 0.7574,
+      "step": 7302
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 3.677673198580466e-05,
+      "loss": 0.6239,
+      "step": 7303
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 3.677416036619864e-05,
+      "loss": 0.5966,
+      "step": 7304
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 3.6771588746592606e-05,
+      "loss": 0.8062,
+      "step": 7305
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 3.6769017126986576e-05,
+      "loss": 0.6377,
+      "step": 7306
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 3.6766445507380546e-05,
+      "loss": 0.5653,
+      "step": 7307
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 3.676387388777452e-05,
+      "loss": 0.8885,
+      "step": 7308
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 3.67613022681685e-05,
+      "loss": 0.66,
+      "step": 7309
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 3.675873064856247e-05,
+      "loss": 1.0375,
+      "step": 7310
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 3.675615902895644e-05,
+      "loss": 0.8889,
+      "step": 7311
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 3.675358740935041e-05,
+      "loss": 0.8914,
+      "step": 7312
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 3.675101578974438e-05,
+      "loss": 0.6905,
+      "step": 7313
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 3.674844417013836e-05,
+      "loss": 0.2998,
+      "step": 7314
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 3.674587255053233e-05,
+      "loss": 0.5372,
+      "step": 7315
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 3.67433009309263e-05,
+      "loss": 1.1343,
+      "step": 7316
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 3.674072931132027e-05,
+      "loss": 0.7958,
+      "step": 7317
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 3.673815769171424e-05,
+      "loss": 1.0645,
+      "step": 7318
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 3.6735586072108216e-05,
+      "loss": 0.9087,
+      "step": 7319
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 3.673301445250219e-05,
+      "loss": 0.4437,
+      "step": 7320
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 3.673044283289616e-05,
+      "loss": 0.768,
+      "step": 7321
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 3.672787121329013e-05,
+      "loss": 0.7326,
+      "step": 7322
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 3.67252995936841e-05,
+      "loss": 0.8654,
+      "step": 7323
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 3.672272797407808e-05,
+      "loss": 0.5263,
+      "step": 7324
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 3.672015635447205e-05,
+      "loss": 0.7986,
+      "step": 7325
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 3.6717584734866024e-05,
+      "loss": 0.8183,
+      "step": 7326
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 3.6715013115259994e-05,
+      "loss": 1.1742,
+      "step": 7327
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 3.671244149565396e-05,
+      "loss": 0.5264,
+      "step": 7328
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 3.670986987604794e-05,
+      "loss": 0.5317,
+      "step": 7329
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 3.670729825644191e-05,
+      "loss": 0.655,
+      "step": 7330
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 3.670472663683588e-05,
+      "loss": 0.7819,
+      "step": 7331
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 3.6702155017229856e-05,
+      "loss": 1.0388,
+      "step": 7332
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 3.6699583397623825e-05,
+      "loss": 0.9562,
+      "step": 7333
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 3.6697011778017795e-05,
+      "loss": 0.9046,
+      "step": 7334
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 3.669444015841177e-05,
+      "loss": 0.9811,
+      "step": 7335
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 3.669186853880574e-05,
+      "loss": 0.917,
+      "step": 7336
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 3.668929691919971e-05,
+      "loss": 0.7023,
+      "step": 7337
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 3.668672529959369e-05,
+      "loss": 0.7161,
+      "step": 7338
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 3.668415367998766e-05,
+      "loss": 0.8294,
+      "step": 7339
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 3.6681582060381633e-05,
+      "loss": 1.0941,
+      "step": 7340
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 3.66790104407756e-05,
+      "loss": 0.8365,
+      "step": 7341
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 3.667643882116957e-05,
+      "loss": 0.7733,
+      "step": 7342
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 3.667386720156354e-05,
+      "loss": 0.9576,
+      "step": 7343
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 3.667129558195752e-05,
+      "loss": 0.7319,
+      "step": 7344
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 3.6668723962351495e-05,
+      "loss": 1.0722,
+      "step": 7345
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 3.6666152342745465e-05,
+      "loss": 0.7447,
+      "step": 7346
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 3.6663580723139435e-05,
+      "loss": 0.9653,
+      "step": 7347
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 3.6661009103533404e-05,
+      "loss": 0.8269,
+      "step": 7348
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 3.6658437483927374e-05,
+      "loss": 0.8571,
+      "step": 7349
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 3.665586586432135e-05,
+      "loss": 1.0644,
+      "step": 7350
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 3.665329424471533e-05,
+      "loss": 0.9296,
+      "step": 7351
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 3.66507226251093e-05,
+      "loss": 0.9712,
+      "step": 7352
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 3.6648151005503266e-05,
+      "loss": 0.9285,
+      "step": 7353
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 3.6645579385897236e-05,
+      "loss": 0.8483,
+      "step": 7354
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 3.664300776629121e-05,
+      "loss": 0.4742,
+      "step": 7355
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 3.664043614668519e-05,
+      "loss": 0.9249,
+      "step": 7356
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 3.663786452707916e-05,
+      "loss": 0.9519,
+      "step": 7357
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 3.663529290747313e-05,
+      "loss": 0.8131,
+      "step": 7358
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 3.66327212878671e-05,
+      "loss": 0.6964,
+      "step": 7359
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 3.6630149668261075e-05,
+      "loss": 0.9637,
+      "step": 7360
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 3.6627578048655044e-05,
+      "loss": 0.6996,
+      "step": 7361
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 3.662500642904902e-05,
+      "loss": 0.8353,
+      "step": 7362
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 3.662243480944299e-05,
+      "loss": 0.9728,
+      "step": 7363
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 3.661986318983696e-05,
+      "loss": 0.4672,
+      "step": 7364
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 3.661729157023093e-05,
+      "loss": 0.8184,
+      "step": 7365
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 3.6614719950624906e-05,
+      "loss": 0.8014,
+      "step": 7366
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 3.6612148331018876e-05,
+      "loss": 0.8058,
+      "step": 7367
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 3.660957671141285e-05,
+      "loss": 0.6168,
+      "step": 7368
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 3.660700509180682e-05,
+      "loss": 0.5379,
+      "step": 7369
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 3.660443347220079e-05,
+      "loss": 0.8951,
+      "step": 7370
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 3.660186185259477e-05,
+      "loss": 1.1185,
+      "step": 7371
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 3.659929023298874e-05,
+      "loss": 0.4884,
+      "step": 7372
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 3.659671861338271e-05,
+      "loss": 0.8338,
+      "step": 7373
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 3.6594146993776684e-05,
+      "loss": 0.5171,
+      "step": 7374
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 3.6591575374170654e-05,
+      "loss": 0.9093,
+      "step": 7375
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 3.658900375456463e-05,
+      "loss": 0.5899,
+      "step": 7376
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 3.65864321349586e-05,
+      "loss": 0.7778,
+      "step": 7377
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 3.658386051535257e-05,
+      "loss": 1.104,
+      "step": 7378
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 3.658128889574654e-05,
+      "loss": 1.2834,
+      "step": 7379
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 3.6578717276140516e-05,
+      "loss": 0.7752,
+      "step": 7380
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 3.6576145656534485e-05,
+      "loss": 0.6399,
+      "step": 7381
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 3.657357403692846e-05,
+      "loss": 0.6808,
+      "step": 7382
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 3.657100241732243e-05,
+      "loss": 0.7223,
+      "step": 7383
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 3.65684307977164e-05,
+      "loss": 0.9563,
+      "step": 7384
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 3.656585917811037e-05,
+      "loss": 1.2021,
+      "step": 7385
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 3.656328755850435e-05,
+      "loss": 0.9281,
+      "step": 7386
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 3.6560715938898324e-05,
+      "loss": 0.3631,
+      "step": 7387
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 3.6558144319292293e-05,
+      "loss": 0.7853,
+      "step": 7388
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 3.655557269968626e-05,
+      "loss": 0.566,
+      "step": 7389
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 3.655300108008023e-05,
+      "loss": 1.0193,
+      "step": 7390
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 3.65504294604742e-05,
+      "loss": 0.6951,
+      "step": 7391
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 3.6547857840868186e-05,
+      "loss": 0.8027,
+      "step": 7392
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 3.6545286221262155e-05,
+      "loss": 0.6739,
+      "step": 7393
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 3.6542714601656125e-05,
+      "loss": 0.7304,
+      "step": 7394
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 3.6540142982050095e-05,
+      "loss": 0.9143,
+      "step": 7395
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 3.6537571362444065e-05,
+      "loss": 0.7125,
+      "step": 7396
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 3.653499974283804e-05,
+      "loss": 0.7277,
+      "step": 7397
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 3.653242812323202e-05,
+      "loss": 0.7105,
+      "step": 7398
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 3.652985650362599e-05,
+      "loss": 0.5876,
+      "step": 7399
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 3.652728488401996e-05,
+      "loss": 0.6153,
+      "step": 7400
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 3.6524713264413927e-05,
+      "loss": 0.9735,
+      "step": 7401
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 3.65221416448079e-05,
+      "loss": 0.6719,
+      "step": 7402
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 3.651957002520187e-05,
+      "loss": 0.8221,
+      "step": 7403
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 3.651699840559585e-05,
+      "loss": 0.3257,
+      "step": 7404
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 3.651442678598982e-05,
+      "loss": 0.9252,
+      "step": 7405
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 3.651185516638379e-05,
+      "loss": 0.7424,
+      "step": 7406
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 3.6509283546777765e-05,
+      "loss": 0.7963,
+      "step": 7407
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 3.6506711927171735e-05,
+      "loss": 1.1851,
+      "step": 7408
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 3.6504140307565704e-05,
+      "loss": 0.6569,
+      "step": 7409
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 3.650156868795968e-05,
+      "loss": 0.6525,
+      "step": 7410
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 3.649899706835365e-05,
+      "loss": 0.7211,
+      "step": 7411
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 3.649642544874762e-05,
+      "loss": 0.9288,
+      "step": 7412
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 3.64938538291416e-05,
+      "loss": 1.1139,
+      "step": 7413
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 3.6491282209535566e-05,
+      "loss": 0.7243,
+      "step": 7414
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 3.6488710589929536e-05,
+      "loss": 1.0841,
+      "step": 7415
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 3.648613897032351e-05,
+      "loss": 0.7558,
+      "step": 7416
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 3.648356735071748e-05,
+      "loss": 0.5672,
+      "step": 7417
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 3.648099573111146e-05,
+      "loss": 1.2613,
+      "step": 7418
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 3.647842411150543e-05,
+      "loss": 0.6425,
+      "step": 7419
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 3.64758524918994e-05,
+      "loss": 0.8287,
+      "step": 7420
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 3.647328087229337e-05,
+      "loss": 0.5934,
+      "step": 7421
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 3.6470709252687344e-05,
+      "loss": 0.995,
+      "step": 7422
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 3.646813763308132e-05,
+      "loss": 0.7342,
+      "step": 7423
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 3.646556601347529e-05,
+      "loss": 0.7997,
+      "step": 7424
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 3.646299439386926e-05,
+      "loss": 0.6168,
+      "step": 7425
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 3.646042277426323e-05,
+      "loss": 1.1338,
+      "step": 7426
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 3.64578511546572e-05,
+      "loss": 0.8329,
+      "step": 7427
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 3.6455279535051176e-05,
+      "loss": 0.7751,
+      "step": 7428
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 3.645270791544515e-05,
+      "loss": 1.0151,
+      "step": 7429
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 3.645013629583912e-05,
+      "loss": 0.9428,
+      "step": 7430
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 3.644756467623309e-05,
+      "loss": 0.7968,
+      "step": 7431
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 3.644499305662706e-05,
+      "loss": 0.7945,
+      "step": 7432
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 3.644242143702104e-05,
+      "loss": 0.9402,
+      "step": 7433
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 3.6439849817415014e-05,
+      "loss": 0.7975,
+      "step": 7434
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 3.6437278197808984e-05,
+      "loss": 0.7122,
+      "step": 7435
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 3.6434706578202954e-05,
+      "loss": 1.0963,
+      "step": 7436
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 3.643213495859692e-05,
+      "loss": 0.9017,
+      "step": 7437
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 3.64295633389909e-05,
+      "loss": 0.7799,
+      "step": 7438
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 3.6426991719384876e-05,
+      "loss": 0.923,
+      "step": 7439
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 3.6424420099778846e-05,
+      "loss": 0.6006,
+      "step": 7440
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 3.6421848480172816e-05,
+      "loss": 1.0517,
+      "step": 7441
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 3.6419276860566785e-05,
+      "loss": 0.8012,
+      "step": 7442
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 3.6416705240960755e-05,
+      "loss": 1.0069,
+      "step": 7443
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 3.641413362135473e-05,
+      "loss": 0.8657,
+      "step": 7444
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 3.641156200174871e-05,
+      "loss": 0.8512,
+      "step": 7445
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 3.640899038214268e-05,
+      "loss": 0.5253,
+      "step": 7446
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 3.640641876253665e-05,
+      "loss": 0.8548,
+      "step": 7447
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 3.640384714293062e-05,
+      "loss": 0.7714,
+      "step": 7448
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 3.640127552332459e-05,
+      "loss": 0.7429,
+      "step": 7449
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 3.639870390371856e-05,
+      "loss": 0.7342,
+      "step": 7450
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 3.639613228411254e-05,
+      "loss": 0.6914,
+      "step": 7451
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 3.639356066450651e-05,
+      "loss": 0.7658,
+      "step": 7452
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 3.639098904490048e-05,
+      "loss": 0.8841,
+      "step": 7453
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 3.6388417425294455e-05,
+      "loss": 0.7791,
+      "step": 7454
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 3.6385845805688425e-05,
+      "loss": 0.8307,
+      "step": 7455
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 3.6383274186082395e-05,
+      "loss": 0.6526,
+      "step": 7456
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 3.638070256647637e-05,
+      "loss": 0.6644,
+      "step": 7457
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 3.637813094687034e-05,
+      "loss": 0.8863,
+      "step": 7458
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 3.637555932726431e-05,
+      "loss": 0.6641,
+      "step": 7459
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 3.637298770765829e-05,
+      "loss": 0.6765,
+      "step": 7460
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 3.637041608805226e-05,
+      "loss": 1.216,
+      "step": 7461
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 3.6367844468446226e-05,
+      "loss": 0.8915,
+      "step": 7462
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 3.63652728488402e-05,
+      "loss": 0.6651,
+      "step": 7463
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 3.636270122923417e-05,
+      "loss": 0.7864,
+      "step": 7464
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 3.636012960962815e-05,
+      "loss": 0.6732,
+      "step": 7465
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 3.635755799002212e-05,
+      "loss": 1.0174,
+      "step": 7466
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 3.635498637041609e-05,
+      "loss": 0.7036,
+      "step": 7467
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 3.635241475081006e-05,
+      "loss": 0.6527,
+      "step": 7468
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 3.6349843131204035e-05,
+      "loss": 0.8523,
+      "step": 7469
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 3.634727151159801e-05,
+      "loss": 0.9503,
+      "step": 7470
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 3.634469989199198e-05,
+      "loss": 1.0535,
+      "step": 7471
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 3.634212827238595e-05,
+      "loss": 0.767,
+      "step": 7472
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 3.633955665277992e-05,
+      "loss": 0.7499,
+      "step": 7473
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 3.633698503317389e-05,
+      "loss": 0.6469,
+      "step": 7474
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 3.6334413413567866e-05,
+      "loss": 0.9795,
+      "step": 7475
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 3.633184179396184e-05,
+      "loss": 0.723,
+      "step": 7476
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 3.632927017435581e-05,
+      "loss": 0.6903,
+      "step": 7477
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 3.632669855474978e-05,
+      "loss": 0.7872,
+      "step": 7478
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 3.632412693514375e-05,
+      "loss": 0.9815,
+      "step": 7479
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 3.632155531553773e-05,
+      "loss": 0.6337,
+      "step": 7480
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 3.6318983695931705e-05,
+      "loss": 0.7902,
+      "step": 7481
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 3.6316412076325674e-05,
+      "loss": 0.8463,
+      "step": 7482
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 3.6313840456719644e-05,
+      "loss": 0.8047,
+      "step": 7483
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 3.6311268837113614e-05,
+      "loss": 0.7862,
+      "step": 7484
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 3.630869721750759e-05,
+      "loss": 0.7902,
+      "step": 7485
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 3.630612559790156e-05,
+      "loss": 1.0343,
+      "step": 7486
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 3.6303553978295536e-05,
+      "loss": 0.8262,
+      "step": 7487
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 3.6300982358689506e-05,
+      "loss": 0.7796,
+      "step": 7488
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 3.6298410739083476e-05,
+      "loss": 0.9562,
+      "step": 7489
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 3.6295839119477445e-05,
+      "loss": 0.85,
+      "step": 7490
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 3.629326749987142e-05,
+      "loss": 1.0365,
+      "step": 7491
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 3.629069588026539e-05,
+      "loss": 0.8011,
+      "step": 7492
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 3.628812426065937e-05,
+      "loss": 1.0055,
+      "step": 7493
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 3.628555264105334e-05,
+      "loss": 0.8735,
+      "step": 7494
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 3.628298102144731e-05,
+      "loss": 0.7539,
+      "step": 7495
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 3.6280409401841284e-05,
+      "loss": 1.2182,
+      "step": 7496
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 3.6277837782235253e-05,
+      "loss": 0.9595,
+      "step": 7497
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 3.627526616262922e-05,
+      "loss": 0.7017,
+      "step": 7498
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 3.62726945430232e-05,
+      "loss": 0.9427,
+      "step": 7499
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 3.627012292341717e-05,
+      "loss": 0.7825,
+      "step": 7500
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 3.6267551303811146e-05,
+      "loss": 0.9206,
+      "step": 7501
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 3.6264979684205115e-05,
+      "loss": 0.7072,
+      "step": 7502
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 3.6262408064599085e-05,
+      "loss": 0.9281,
+      "step": 7503
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 3.6259836444993055e-05,
+      "loss": 0.835,
+      "step": 7504
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 3.625726482538703e-05,
+      "loss": 0.7877,
+      "step": 7505
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 3.6254693205781e-05,
+      "loss": 0.8242,
+      "step": 7506
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 3.625212158617498e-05,
+      "loss": 0.6549,
+      "step": 7507
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 3.624954996656895e-05,
+      "loss": 0.5154,
+      "step": 7508
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 3.624697834696292e-05,
+      "loss": 0.7581,
+      "step": 7509
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 3.6244406727356886e-05,
+      "loss": 1.0298,
+      "step": 7510
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 3.624183510775086e-05,
+      "loss": 0.5764,
+      "step": 7511
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 3.623926348814484e-05,
+      "loss": 0.5837,
+      "step": 7512
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 3.623669186853881e-05,
+      "loss": 0.7677,
+      "step": 7513
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 3.623412024893278e-05,
+      "loss": 0.7014,
+      "step": 7514
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 3.623154862932675e-05,
+      "loss": 0.8704,
+      "step": 7515
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 3.622897700972072e-05,
+      "loss": 0.6943,
+      "step": 7516
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 3.62264053901147e-05,
+      "loss": 0.6552,
+      "step": 7517
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 3.622383377050867e-05,
+      "loss": 0.8123,
+      "step": 7518
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 3.622126215090264e-05,
+      "loss": 0.7246,
+      "step": 7519
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 3.621869053129661e-05,
+      "loss": 0.8826,
+      "step": 7520
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 3.621611891169058e-05,
+      "loss": 0.6005,
+      "step": 7521
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 3.6213547292084557e-05,
+      "loss": 0.762,
+      "step": 7522
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 3.621097567247853e-05,
+      "loss": 0.9758,
+      "step": 7523
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 3.62084040528725e-05,
+      "loss": 0.7389,
+      "step": 7524
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 3.620583243326647e-05,
+      "loss": 0.9395,
+      "step": 7525
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 3.620326081366044e-05,
+      "loss": 0.8385,
+      "step": 7526
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 3.620068919405442e-05,
+      "loss": 0.818,
+      "step": 7527
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 3.619811757444839e-05,
+      "loss": 0.8165,
+      "step": 7528
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 3.6195545954842365e-05,
+      "loss": 0.8627,
+      "step": 7529
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 3.6192974335236334e-05,
+      "loss": 0.9799,
+      "step": 7530
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 3.6190402715630304e-05,
+      "loss": 0.8414,
+      "step": 7531
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 3.618783109602428e-05,
+      "loss": 0.7065,
+      "step": 7532
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 3.618525947641825e-05,
+      "loss": 0.7905,
+      "step": 7533
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 3.618268785681222e-05,
+      "loss": 0.6107,
+      "step": 7534
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 3.6180116237206196e-05,
+      "loss": 0.6531,
+      "step": 7535
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 3.6177544617600166e-05,
+      "loss": 1.0513,
+      "step": 7536
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 3.6174972997994136e-05,
+      "loss": 0.6636,
+      "step": 7537
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 3.617240137838811e-05,
+      "loss": 0.7307,
+      "step": 7538
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 3.616982975878208e-05,
+      "loss": 0.8402,
+      "step": 7539
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 3.616725813917605e-05,
+      "loss": 1.0662,
+      "step": 7540
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 3.616468651957003e-05,
+      "loss": 0.9413,
+      "step": 7541
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 3.6162114899964e-05,
+      "loss": 1.1107,
+      "step": 7542
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 3.6159543280357974e-05,
+      "loss": 0.6425,
+      "step": 7543
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 3.6156971660751944e-05,
+      "loss": 0.9085,
+      "step": 7544
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 3.6154400041145914e-05,
+      "loss": 0.6749,
+      "step": 7545
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 3.615182842153988e-05,
+      "loss": 0.6208,
+      "step": 7546
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 3.614925680193386e-05,
+      "loss": 0.6177,
+      "step": 7547
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 3.6146685182327836e-05,
+      "loss": 0.9345,
+      "step": 7548
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 3.6144113562721806e-05,
+      "loss": 0.6248,
+      "step": 7549
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 3.6141541943115776e-05,
+      "loss": 0.958,
+      "step": 7550
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 3.6138970323509745e-05,
+      "loss": 0.5736,
+      "step": 7551
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 3.6136398703903715e-05,
+      "loss": 0.6939,
+      "step": 7552
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 3.613382708429769e-05,
+      "loss": 0.7356,
+      "step": 7553
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 3.613125546469167e-05,
+      "loss": 0.7093,
+      "step": 7554
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 3.612868384508564e-05,
+      "loss": 0.7307,
+      "step": 7555
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 3.612611222547961e-05,
+      "loss": 0.8522,
+      "step": 7556
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 3.612354060587358e-05,
+      "loss": 0.7101,
+      "step": 7557
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 3.612096898626755e-05,
+      "loss": 0.5028,
+      "step": 7558
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 3.611839736666153e-05,
+      "loss": 0.8813,
+      "step": 7559
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 3.61158257470555e-05,
+      "loss": 0.9438,
+      "step": 7560
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 3.611325412744947e-05,
+      "loss": 1.2202,
+      "step": 7561
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 3.611068250784344e-05,
+      "loss": 0.972,
+      "step": 7562
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 3.6108110888237415e-05,
+      "loss": 0.6497,
+      "step": 7563
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 3.6105539268631385e-05,
+      "loss": 0.8951,
+      "step": 7564
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 3.610296764902536e-05,
+      "loss": 1.1274,
+      "step": 7565
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 3.610039602941933e-05,
+      "loss": 1.0008,
+      "step": 7566
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 3.60978244098133e-05,
+      "loss": 0.8545,
+      "step": 7567
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 3.609525279020727e-05,
+      "loss": 0.9272,
+      "step": 7568
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 3.609268117060125e-05,
+      "loss": 0.7623,
+      "step": 7569
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 3.609010955099522e-05,
+      "loss": 0.6596,
+      "step": 7570
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 3.608753793138919e-05,
+      "loss": 0.9125,
+      "step": 7571
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 3.608496631178316e-05,
+      "loss": 0.7912,
+      "step": 7572
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 3.608239469217713e-05,
+      "loss": 0.7926,
+      "step": 7573
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 3.607982307257111e-05,
+      "loss": 0.6274,
+      "step": 7574
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 3.607725145296508e-05,
+      "loss": 0.5121,
+      "step": 7575
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 3.607467983335905e-05,
+      "loss": 0.6918,
+      "step": 7576
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 3.6072108213753025e-05,
+      "loss": 0.6553,
+      "step": 7577
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 3.6069536594146994e-05,
+      "loss": 0.6933,
+      "step": 7578
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 3.606696497454097e-05,
+      "loss": 0.8678,
+      "step": 7579
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 3.606439335493494e-05,
+      "loss": 0.6563,
+      "step": 7580
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 3.606182173532891e-05,
+      "loss": 0.8013,
+      "step": 7581
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 3.605925011572288e-05,
+      "loss": 0.5193,
+      "step": 7582
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 3.6056678496116856e-05,
+      "loss": 0.693,
+      "step": 7583
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 3.6054106876510826e-05,
+      "loss": 0.8345,
+      "step": 7584
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 3.60515352569048e-05,
+      "loss": 0.8875,
+      "step": 7585
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 3.604896363729877e-05,
+      "loss": 0.8656,
+      "step": 7586
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 3.604639201769274e-05,
+      "loss": 0.4894,
+      "step": 7587
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 3.604382039808672e-05,
+      "loss": 0.8071,
+      "step": 7588
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 3.604124877848069e-05,
+      "loss": 0.8035,
+      "step": 7589
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 3.6038677158874665e-05,
+      "loss": 0.8461,
+      "step": 7590
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 3.6036105539268634e-05,
+      "loss": 0.4824,
+      "step": 7591
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 3.6033533919662604e-05,
+      "loss": 0.7076,
+      "step": 7592
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 3.6030962300056574e-05,
+      "loss": 0.5167,
+      "step": 7593
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 3.602839068045055e-05,
+      "loss": 0.8503,
+      "step": 7594
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 3.6025819060844527e-05,
+      "loss": 1.0926,
+      "step": 7595
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 3.6023247441238496e-05,
+      "loss": 1.1561,
+      "step": 7596
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 3.6020675821632466e-05,
+      "loss": 0.5441,
+      "step": 7597
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 3.6018104202026436e-05,
+      "loss": 0.6245,
+      "step": 7598
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 3.6015532582420405e-05,
+      "loss": 0.7206,
+      "step": 7599
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 3.601296096281438e-05,
+      "loss": 0.9715,
+      "step": 7600
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 3.601038934320836e-05,
+      "loss": 0.4809,
+      "step": 7601
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 3.600781772360233e-05,
+      "loss": 0.6307,
+      "step": 7602
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 3.60052461039963e-05,
+      "loss": 0.5871,
+      "step": 7603
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 3.600267448439027e-05,
+      "loss": 1.0594,
+      "step": 7604
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 3.6000102864784244e-05,
+      "loss": 0.6088,
+      "step": 7605
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 3.599753124517822e-05,
+      "loss": 0.8175,
+      "step": 7606
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 3.599495962557219e-05,
+      "loss": 0.7361,
+      "step": 7607
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 3.599238800596616e-05,
+      "loss": 0.6467,
+      "step": 7608
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 3.598981638636013e-05,
+      "loss": 0.5984,
+      "step": 7609
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 3.5987244766754106e-05,
+      "loss": 0.8366,
+      "step": 7610
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 3.5984673147148075e-05,
+      "loss": 0.7481,
+      "step": 7611
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 3.598210152754205e-05,
+      "loss": 0.6108,
+      "step": 7612
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 3.597952990793602e-05,
+      "loss": 0.573,
+      "step": 7613
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 3.597695828832999e-05,
+      "loss": 1.3257,
+      "step": 7614
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 3.597438666872396e-05,
+      "loss": 0.5584,
+      "step": 7615
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 3.597181504911794e-05,
+      "loss": 1.3693,
+      "step": 7616
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 3.596924342951191e-05,
+      "loss": 0.5809,
+      "step": 7617
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 3.5966671809905884e-05,
+      "loss": 0.6635,
+      "step": 7618
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 3.596410019029985e-05,
+      "loss": 0.7714,
+      "step": 7619
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 3.596152857069382e-05,
+      "loss": 0.7189,
+      "step": 7620
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 3.59589569510878e-05,
+      "loss": 0.5622,
+      "step": 7621
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 3.595638533148177e-05,
+      "loss": 0.6179,
+      "step": 7622
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 3.595381371187574e-05,
+      "loss": 0.9023,
+      "step": 7623
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 3.5951242092269715e-05,
+      "loss": 0.7339,
+      "step": 7624
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 3.5948670472663685e-05,
+      "loss": 0.9639,
+      "step": 7625
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 3.594609885305766e-05,
+      "loss": 0.8643,
+      "step": 7626
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 3.594352723345163e-05,
+      "loss": 0.9077,
+      "step": 7627
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 3.59409556138456e-05,
+      "loss": 0.993,
+      "step": 7628
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 3.593838399423957e-05,
+      "loss": 0.7474,
+      "step": 7629
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 3.593581237463355e-05,
+      "loss": 0.8594,
+      "step": 7630
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 3.5933240755027517e-05,
+      "loss": 0.9318,
+      "step": 7631
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 3.593066913542149e-05,
+      "loss": 0.7511,
+      "step": 7632
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 3.592809751581546e-05,
+      "loss": 0.9314,
+      "step": 7633
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 3.592552589620943e-05,
+      "loss": 0.9008,
+      "step": 7634
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 3.59229542766034e-05,
+      "loss": 0.6596,
+      "step": 7635
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 3.592038265699738e-05,
+      "loss": 0.7671,
+      "step": 7636
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 3.5917811037391355e-05,
+      "loss": 0.6006,
+      "step": 7637
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 3.5915239417785325e-05,
+      "loss": 0.9574,
+      "step": 7638
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 3.5912667798179294e-05,
+      "loss": 0.7964,
+      "step": 7639
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 3.5910096178573264e-05,
+      "loss": 0.7016,
+      "step": 7640
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 3.5907524558967234e-05,
+      "loss": 0.8847,
+      "step": 7641
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 3.590495293936122e-05,
+      "loss": 0.8078,
+      "step": 7642
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 3.590238131975519e-05,
+      "loss": 0.6781,
+      "step": 7643
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 3.5899809700149156e-05,
+      "loss": 0.5961,
+      "step": 7644
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 3.5897238080543126e-05,
+      "loss": 0.8378,
+      "step": 7645
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 3.5894666460937096e-05,
+      "loss": 0.9871,
+      "step": 7646
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 3.589209484133107e-05,
+      "loss": 0.8671,
+      "step": 7647
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 3.588952322172505e-05,
+      "loss": 0.7494,
+      "step": 7648
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 3.588695160211902e-05,
+      "loss": 0.7179,
+      "step": 7649
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 3.588437998251299e-05,
+      "loss": 0.8315,
+      "step": 7650
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 3.588180836290696e-05,
+      "loss": 0.8314,
+      "step": 7651
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 3.5879236743300934e-05,
+      "loss": 0.8871,
+      "step": 7652
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 3.5876665123694904e-05,
+      "loss": 0.5503,
+      "step": 7653
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 3.587409350408888e-05,
+      "loss": 1.0578,
+      "step": 7654
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 3.587152188448285e-05,
+      "loss": 0.86,
+      "step": 7655
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 3.586895026487682e-05,
+      "loss": 0.8614,
+      "step": 7656
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 3.5866378645270796e-05,
+      "loss": 0.7976,
+      "step": 7657
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 3.5863807025664766e-05,
+      "loss": 0.9251,
+      "step": 7658
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 3.5861235406058736e-05,
+      "loss": 0.6326,
+      "step": 7659
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 3.585866378645271e-05,
+      "loss": 0.7238,
+      "step": 7660
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 3.585609216684668e-05,
+      "loss": 0.5435,
+      "step": 7661
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 3.585352054724065e-05,
+      "loss": 0.9655,
+      "step": 7662
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 3.585094892763463e-05,
+      "loss": 0.7921,
+      "step": 7663
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 3.58483773080286e-05,
+      "loss": 0.6587,
+      "step": 7664
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 3.584580568842257e-05,
+      "loss": 0.67,
+      "step": 7665
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 3.5843234068816544e-05,
+      "loss": 0.7685,
+      "step": 7666
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 3.584066244921051e-05,
+      "loss": 0.9394,
+      "step": 7667
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 3.583809082960449e-05,
+      "loss": 0.7326,
+      "step": 7668
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 3.583551920999846e-05,
+      "loss": 0.954,
+      "step": 7669
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 3.583294759039243e-05,
+      "loss": 1.0886,
+      "step": 7670
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 3.58303759707864e-05,
+      "loss": 0.7338,
+      "step": 7671
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 3.5827804351180375e-05,
+      "loss": 0.7654,
+      "step": 7672
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 3.582523273157435e-05,
+      "loss": 0.9186,
+      "step": 7673
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 3.582266111196832e-05,
+      "loss": 0.8891,
+      "step": 7674
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 3.582008949236229e-05,
+      "loss": 1.1144,
+      "step": 7675
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 3.581751787275626e-05,
+      "loss": 0.645,
+      "step": 7676
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 3.581494625315023e-05,
+      "loss": 0.7389,
+      "step": 7677
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 3.581237463354421e-05,
+      "loss": 0.6266,
+      "step": 7678
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 3.5809803013938183e-05,
+      "loss": 0.9428,
+      "step": 7679
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 3.580723139433215e-05,
+      "loss": 0.9348,
+      "step": 7680
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 3.580465977472612e-05,
+      "loss": 0.8648,
+      "step": 7681
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 3.580208815512009e-05,
+      "loss": 1.0986,
+      "step": 7682
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 3.579951653551407e-05,
+      "loss": 0.8195,
+      "step": 7683
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 3.5796944915908045e-05,
+      "loss": 0.8578,
+      "step": 7684
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 3.5794373296302015e-05,
+      "loss": 0.525,
+      "step": 7685
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 3.5791801676695985e-05,
+      "loss": 0.4753,
+      "step": 7686
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 3.5789230057089954e-05,
+      "loss": 0.7324,
+      "step": 7687
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 3.578665843748393e-05,
+      "loss": 0.6986,
+      "step": 7688
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 3.57840868178779e-05,
+      "loss": 0.6182,
+      "step": 7689
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 3.578151519827188e-05,
+      "loss": 0.5988,
+      "step": 7690
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 3.577894357866585e-05,
+      "loss": 0.6084,
+      "step": 7691
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 3.5776371959059816e-05,
+      "loss": 0.6944,
+      "step": 7692
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 3.5773800339453786e-05,
+      "loss": 0.6216,
+      "step": 7693
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 3.577122871984776e-05,
+      "loss": 0.9267,
+      "step": 7694
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 3.576865710024173e-05,
+      "loss": 0.6539,
+      "step": 7695
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 3.576608548063571e-05,
+      "loss": 0.8613,
+      "step": 7696
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 3.576351386102968e-05,
+      "loss": 0.7865,
+      "step": 7697
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.576094224142365e-05,
+      "loss": 0.7583,
+      "step": 7698
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.5758370621817625e-05,
+      "loss": 0.7488,
+      "step": 7699
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.5755799002211594e-05,
+      "loss": 0.7419,
+      "step": 7700
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.5753227382605564e-05,
+      "loss": 0.8058,
+      "step": 7701
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.575065576299954e-05,
+      "loss": 0.5504,
+      "step": 7702
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.574808414339351e-05,
+      "loss": 0.9912,
+      "step": 7703
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.5745512523787487e-05,
+      "loss": 0.8334,
+      "step": 7704
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.5742940904181456e-05,
+      "loss": 0.7002,
+      "step": 7705
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.5740369284575426e-05,
+      "loss": 0.9411,
+      "step": 7706
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.5737797664969396e-05,
+      "loss": 0.73,
+      "step": 7707
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.573522604536337e-05,
+      "loss": 0.7752,
+      "step": 7708
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.573265442575734e-05,
+      "loss": 0.5872,
+      "step": 7709
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.573008280615132e-05,
+      "loss": 0.8075,
+      "step": 7710
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.572751118654529e-05,
+      "loss": 0.8004,
+      "step": 7711
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.572493956693926e-05,
+      "loss": 0.6532,
+      "step": 7712
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.572236794733323e-05,
+      "loss": 0.7194,
+      "step": 7713
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.5719796327727204e-05,
+      "loss": 1.1125,
+      "step": 7714
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.571722470812118e-05,
+      "loss": 0.5205,
+      "step": 7715
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.571465308851515e-05,
+      "loss": 0.8085,
+      "step": 7716
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.571208146890912e-05,
+      "loss": 0.8694,
+      "step": 7717
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.570950984930309e-05,
+      "loss": 0.7896,
+      "step": 7718
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.570693822969706e-05,
+      "loss": 1.0473,
+      "step": 7719
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.570436661009104e-05,
+      "loss": 0.5953,
+      "step": 7720
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.570179499048501e-05,
+      "loss": 0.7287,
+      "step": 7721
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.569922337087898e-05,
+      "loss": 0.804,
+      "step": 7722
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.569665175127295e-05,
+      "loss": 0.7948,
+      "step": 7723
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.569408013166692e-05,
+      "loss": 1.124,
+      "step": 7724
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.56915085120609e-05,
+      "loss": 0.5306,
+      "step": 7725
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.5688936892454874e-05,
+      "loss": 0.7308,
+      "step": 7726
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.5686365272848844e-05,
+      "loss": 0.8777,
+      "step": 7727
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.568379365324281e-05,
+      "loss": 0.9313,
+      "step": 7728
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.568122203363678e-05,
+      "loss": 0.8759,
+      "step": 7729
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.567865041403076e-05,
+      "loss": 0.9308,
+      "step": 7730
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.567607879442473e-05,
+      "loss": 0.8642,
+      "step": 7731
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.5673507174818705e-05,
+      "loss": 0.8976,
+      "step": 7732
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.5670935555212675e-05,
+      "loss": 0.9051,
+      "step": 7733
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.5668363935606645e-05,
+      "loss": 0.9536,
+      "step": 7734
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.566579231600062e-05,
+      "loss": 0.6554,
+      "step": 7735
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.566322069639459e-05,
+      "loss": 0.8098,
+      "step": 7736
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.566064907678856e-05,
+      "loss": 0.9359,
+      "step": 7737
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.565807745718254e-05,
+      "loss": 0.8807,
+      "step": 7738
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.565550583757651e-05,
+      "loss": 1.1335,
+      "step": 7739
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.5652934217970477e-05,
+      "loss": 0.7733,
+      "step": 7740
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.565036259836445e-05,
+      "loss": 0.6941,
+      "step": 7741
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.564779097875842e-05,
+      "loss": 0.8619,
+      "step": 7742
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.56452193591524e-05,
+      "loss": 0.7297,
+      "step": 7743
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.564264773954637e-05,
+      "loss": 0.5888,
+      "step": 7744
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.564007611994034e-05,
+      "loss": 0.7589,
+      "step": 7745
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.5637504500334315e-05,
+      "loss": 0.825,
+      "step": 7746
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.5634932880728285e-05,
+      "loss": 1.1081,
+      "step": 7747
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.5632361261122254e-05,
+      "loss": 1.0893,
+      "step": 7748
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.562978964151623e-05,
+      "loss": 0.8088,
+      "step": 7749
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.56272180219102e-05,
+      "loss": 0.7089,
+      "step": 7750
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.562464640230418e-05,
+      "loss": 0.4785,
+      "step": 7751
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.562207478269815e-05,
+      "loss": 0.5135,
+      "step": 7752
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.5619503163092116e-05,
+      "loss": 0.6805,
+      "step": 7753
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.5616931543486086e-05,
+      "loss": 0.816,
+      "step": 7754
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.561435992388006e-05,
+      "loss": 0.6382,
+      "step": 7755
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.561178830427403e-05,
+      "loss": 0.5373,
+      "step": 7756
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.560921668466801e-05,
+      "loss": 0.6538,
+      "step": 7757
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.560664506506198e-05,
+      "loss": 0.8619,
+      "step": 7758
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.560407344545595e-05,
+      "loss": 0.4427,
+      "step": 7759
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.560150182584992e-05,
+      "loss": 0.8699,
+      "step": 7760
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.5598930206243894e-05,
+      "loss": 0.7125,
+      "step": 7761
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.559635858663787e-05,
+      "loss": 0.8562,
+      "step": 7762
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.559378696703184e-05,
+      "loss": 0.9475,
+      "step": 7763
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.559121534742581e-05,
+      "loss": 0.4479,
+      "step": 7764
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.558864372781978e-05,
+      "loss": 0.6547,
+      "step": 7765
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.558607210821375e-05,
+      "loss": 0.6843,
+      "step": 7766
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.558350048860773e-05,
+      "loss": 0.9214,
+      "step": 7767
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.55809288690017e-05,
+      "loss": 0.7892,
+      "step": 7768
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.557835724939567e-05,
+      "loss": 0.8324,
+      "step": 7769
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.557578562978964e-05,
+      "loss": 0.5113,
+      "step": 7770
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.557321401018361e-05,
+      "loss": 0.8661,
+      "step": 7771
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.557064239057759e-05,
+      "loss": 0.9656,
+      "step": 7772
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.5568070770971564e-05,
+      "loss": 0.5078,
+      "step": 7773
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.5565499151365534e-05,
+      "loss": 0.8422,
+      "step": 7774
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.5562927531759504e-05,
+      "loss": 0.6032,
+      "step": 7775
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.556035591215347e-05,
+      "loss": 0.8638,
+      "step": 7776
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.555778429254745e-05,
+      "loss": 0.6604,
+      "step": 7777
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.555521267294142e-05,
+      "loss": 0.6882,
+      "step": 7778
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.5552641053335396e-05,
+      "loss": 0.7996,
+      "step": 7779
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.5550069433729366e-05,
+      "loss": 0.5944,
+      "step": 7780
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.5547497814123335e-05,
+      "loss": 0.7516,
+      "step": 7781
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.554492619451731e-05,
+      "loss": 0.7365,
+      "step": 7782
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.554235457491128e-05,
+      "loss": 0.8772,
+      "step": 7783
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.553978295530525e-05,
+      "loss": 0.7579,
+      "step": 7784
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.553721133569923e-05,
+      "loss": 1.1292,
+      "step": 7785
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.55346397160932e-05,
+      "loss": 1.0707,
+      "step": 7786
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.553206809648717e-05,
+      "loss": 0.6198,
+      "step": 7787
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.5529496476881143e-05,
+      "loss": 1.0077,
+      "step": 7788
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.552692485727511e-05,
+      "loss": 0.8325,
+      "step": 7789
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.552435323766908e-05,
+      "loss": 0.4889,
+      "step": 7790
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.552178161806306e-05,
+      "loss": 0.4904,
+      "step": 7791
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.551920999845703e-05,
+      "loss": 0.9831,
+      "step": 7792
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.5516638378851005e-05,
+      "loss": 0.7372,
+      "step": 7793
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.5514066759244975e-05,
+      "loss": 0.9316,
+      "step": 7794
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.5511495139638945e-05,
+      "loss": 0.5834,
+      "step": 7795
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.5508923520032914e-05,
+      "loss": 0.8324,
+      "step": 7796
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.550635190042689e-05,
+      "loss": 0.9417,
+      "step": 7797
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.550378028082087e-05,
+      "loss": 0.791,
+      "step": 7798
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.550120866121484e-05,
+      "loss": 0.8743,
+      "step": 7799
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.549863704160881e-05,
+      "loss": 0.8837,
+      "step": 7800
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.5496065422002776e-05,
+      "loss": 0.9908,
+      "step": 7801
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.5493493802396746e-05,
+      "loss": 0.5524,
+      "step": 7802
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.549092218279072e-05,
+      "loss": 0.74,
+      "step": 7803
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.54883505631847e-05,
+      "loss": 0.6111,
+      "step": 7804
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.548577894357867e-05,
+      "loss": 0.8501,
+      "step": 7805
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.548320732397264e-05,
+      "loss": 0.8637,
+      "step": 7806
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.548063570436661e-05,
+      "loss": 0.7734,
+      "step": 7807
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.5478064084760585e-05,
+      "loss": 0.7279,
+      "step": 7808
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.547549246515456e-05,
+      "loss": 0.9107,
+      "step": 7809
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.547292084554853e-05,
+      "loss": 1.0275,
+      "step": 7810
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.54703492259425e-05,
+      "loss": 0.7277,
+      "step": 7811
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.546777760633647e-05,
+      "loss": 1.1967,
+      "step": 7812
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.5465205986730447e-05,
+      "loss": 0.7973,
+      "step": 7813
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.5462634367124416e-05,
+      "loss": 0.7961,
+      "step": 7814
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.546006274751839e-05,
+      "loss": 0.6096,
+      "step": 7815
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.545749112791236e-05,
+      "loss": 0.5954,
+      "step": 7816
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.545491950830633e-05,
+      "loss": 1.0253,
+      "step": 7817
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.54523478887003e-05,
+      "loss": 0.8122,
+      "step": 7818
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.544977626909428e-05,
+      "loss": 1.1053,
+      "step": 7819
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.544720464948825e-05,
+      "loss": 0.5985,
+      "step": 7820
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.5444633029882224e-05,
+      "loss": 0.8652,
+      "step": 7821
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.5442061410276194e-05,
+      "loss": 0.7905,
+      "step": 7822
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.5439489790670164e-05,
+      "loss": 0.4064,
+      "step": 7823
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.543691817106414e-05,
+      "loss": 0.819,
+      "step": 7824
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.543434655145811e-05,
+      "loss": 0.7403,
+      "step": 7825
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.543177493185208e-05,
+      "loss": 0.7197,
+      "step": 7826
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.5429203312246056e-05,
+      "loss": 0.7365,
+      "step": 7827
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.5426631692640026e-05,
+      "loss": 0.8521,
+      "step": 7828
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.5424060073034e-05,
+      "loss": 0.7629,
+      "step": 7829
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.542148845342797e-05,
+      "loss": 0.7194,
+      "step": 7830
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.541891683382194e-05,
+      "loss": 0.8199,
+      "step": 7831
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.541634521421591e-05,
+      "loss": 0.8078,
+      "step": 7832
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.541377359460989e-05,
+      "loss": 0.8047,
+      "step": 7833
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.541120197500386e-05,
+      "loss": 0.9261,
+      "step": 7834
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.5408630355397834e-05,
+      "loss": 0.8641,
+      "step": 7835
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.5406058735791803e-05,
+      "loss": 0.671,
+      "step": 7836
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.540348711618577e-05,
+      "loss": 0.5696,
+      "step": 7837
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.540091549657974e-05,
+      "loss": 0.9658,
+      "step": 7838
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.539834387697372e-05,
+      "loss": 1.3126,
+      "step": 7839
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.5395772257367696e-05,
+      "loss": 0.9721,
+      "step": 7840
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.5393200637761665e-05,
+      "loss": 0.4739,
+      "step": 7841
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.5390629018155635e-05,
+      "loss": 0.7009,
+      "step": 7842
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.5388057398549605e-05,
+      "loss": 0.8399,
+      "step": 7843
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.5385485778943575e-05,
+      "loss": 0.7779,
+      "step": 7844
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.538291415933756e-05,
+      "loss": 0.871,
+      "step": 7845
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.538034253973153e-05,
+      "loss": 0.9363,
+      "step": 7846
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.53777709201255e-05,
+      "loss": 0.6818,
+      "step": 7847
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.537519930051947e-05,
+      "loss": 0.5168,
+      "step": 7848
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.5372627680913436e-05,
+      "loss": 0.9404,
+      "step": 7849
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.537005606130741e-05,
+      "loss": 0.738,
+      "step": 7850
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.536748444170139e-05,
+      "loss": 0.6269,
+      "step": 7851
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.536491282209536e-05,
+      "loss": 0.772,
+      "step": 7852
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.536234120248933e-05,
+      "loss": 0.495,
+      "step": 7853
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.53597695828833e-05,
+      "loss": 0.781,
+      "step": 7854
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.5357197963277275e-05,
+      "loss": 0.8533,
+      "step": 7855
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.5354626343671245e-05,
+      "loss": 0.8668,
+      "step": 7856
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.535205472406522e-05,
+      "loss": 0.7167,
+      "step": 7857
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.534948310445919e-05,
+      "loss": 0.7255,
+      "step": 7858
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.534691148485316e-05,
+      "loss": 1.0127,
+      "step": 7859
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.534433986524714e-05,
+      "loss": 0.6993,
+      "step": 7860
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.5341768245641107e-05,
+      "loss": 0.4929,
+      "step": 7861
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.5339196626035076e-05,
+      "loss": 1.0044,
+      "step": 7862
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.533662500642905e-05,
+      "loss": 1.0395,
+      "step": 7863
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.533405338682302e-05,
+      "loss": 0.7472,
+      "step": 7864
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.533148176721699e-05,
+      "loss": 0.582,
+      "step": 7865
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.532891014761097e-05,
+      "loss": 0.7058,
+      "step": 7866
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.532633852800494e-05,
+      "loss": 0.9978,
+      "step": 7867
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.532376690839891e-05,
+      "loss": 1.1448,
+      "step": 7868
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.5321195288792884e-05,
+      "loss": 0.9558,
+      "step": 7869
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.5318623669186854e-05,
+      "loss": 0.6826,
+      "step": 7870
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.531605204958083e-05,
+      "loss": 0.8191,
+      "step": 7871
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.53134804299748e-05,
+      "loss": 0.6313,
+      "step": 7872
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.531090881036877e-05,
+      "loss": 0.9486,
+      "step": 7873
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.530833719076274e-05,
+      "loss": 1.2375,
+      "step": 7874
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.5305765571156716e-05,
+      "loss": 0.706,
+      "step": 7875
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.530319395155069e-05,
+      "loss": 0.7517,
+      "step": 7876
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.530062233194466e-05,
+      "loss": 0.5706,
+      "step": 7877
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.529805071233863e-05,
+      "loss": 0.8851,
+      "step": 7878
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.52954790927326e-05,
+      "loss": 0.5073,
+      "step": 7879
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.529290747312657e-05,
+      "loss": 0.724,
+      "step": 7880
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.529033585352055e-05,
+      "loss": 0.4933,
+      "step": 7881
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.5287764233914524e-05,
+      "loss": 0.953,
+      "step": 7882
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.5285192614308494e-05,
+      "loss": 0.6957,
+      "step": 7883
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.5282620994702464e-05,
+      "loss": 0.9148,
+      "step": 7884
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.528004937509643e-05,
+      "loss": 0.5506,
+      "step": 7885
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.527747775549041e-05,
+      "loss": 0.6443,
+      "step": 7886
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.5274906135884386e-05,
+      "loss": 0.7711,
+      "step": 7887
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.5272334516278356e-05,
+      "loss": 0.7719,
+      "step": 7888
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.5269762896672326e-05,
+      "loss": 0.9617,
+      "step": 7889
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.5267191277066295e-05,
+      "loss": 0.7056,
+      "step": 7890
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.5264619657460265e-05,
+      "loss": 1.0555,
+      "step": 7891
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.526204803785425e-05,
+      "loss": 0.8298,
+      "step": 7892
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.525947641824822e-05,
+      "loss": 0.6908,
+      "step": 7893
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.525690479864219e-05,
+      "loss": 0.7587,
+      "step": 7894
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.525433317903616e-05,
+      "loss": 0.9013,
+      "step": 7895
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.525176155943013e-05,
+      "loss": 1.0779,
+      "step": 7896
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.52491899398241e-05,
+      "loss": 0.7886,
+      "step": 7897
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.524661832021808e-05,
+      "loss": 0.6433,
+      "step": 7898
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.524404670061205e-05,
+      "loss": 0.5482,
+      "step": 7899
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.524147508100602e-05,
+      "loss": 0.5069,
+      "step": 7900
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.523890346139999e-05,
+      "loss": 0.8272,
+      "step": 7901
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.5236331841793965e-05,
+      "loss": 0.8558,
+      "step": 7902
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.5233760222187935e-05,
+      "loss": 1.0008,
+      "step": 7903
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.523118860258191e-05,
+      "loss": 0.8142,
+      "step": 7904
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.522861698297588e-05,
+      "loss": 0.7786,
+      "step": 7905
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.522604536336985e-05,
+      "loss": 0.7551,
+      "step": 7906
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.522347374376383e-05,
+      "loss": 0.6552,
+      "step": 7907
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.52209021241578e-05,
+      "loss": 1.2409,
+      "step": 7908
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.521833050455177e-05,
+      "loss": 0.8564,
+      "step": 7909
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.521575888494574e-05,
+      "loss": 0.7866,
+      "step": 7910
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.521318726533971e-05,
+      "loss": 0.7072,
+      "step": 7911
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.521061564573368e-05,
+      "loss": 0.6875,
+      "step": 7912
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.520804402612766e-05,
+      "loss": 0.9284,
+      "step": 7913
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.520547240652163e-05,
+      "loss": 0.7364,
+      "step": 7914
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.52029007869156e-05,
+      "loss": 0.7267,
+      "step": 7915
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.5200329167309575e-05,
+      "loss": 0.8717,
+      "step": 7916
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.5197757547703544e-05,
+      "loss": 0.9318,
+      "step": 7917
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.519518592809752e-05,
+      "loss": 0.6823,
+      "step": 7918
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.519261430849149e-05,
+      "loss": 0.8618,
+      "step": 7919
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.519004268888546e-05,
+      "loss": 1.2825,
+      "step": 7920
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.518747106927943e-05,
+      "loss": 0.8371,
+      "step": 7921
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.5184899449673406e-05,
+      "loss": 0.6648,
+      "step": 7922
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.518232783006738e-05,
+      "loss": 0.7204,
+      "step": 7923
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.517975621046135e-05,
+      "loss": 0.8137,
+      "step": 7924
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.517718459085532e-05,
+      "loss": 0.7801,
+      "step": 7925
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.517461297124929e-05,
+      "loss": 0.754,
+      "step": 7926
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.517204135164326e-05,
+      "loss": 0.7738,
+      "step": 7927
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.516946973203724e-05,
+      "loss": 1.0006,
+      "step": 7928
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.5166898112431215e-05,
+      "loss": 0.7447,
+      "step": 7929
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.5164326492825184e-05,
+      "loss": 1.0261,
+      "step": 7930
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.5161754873219154e-05,
+      "loss": 0.6119,
+      "step": 7931
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.5159183253613124e-05,
+      "loss": 0.6169,
+      "step": 7932
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.51566116340071e-05,
+      "loss": 0.7774,
+      "step": 7933
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.5154040014401077e-05,
+      "loss": 0.8395,
+      "step": 7934
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.5151468394795046e-05,
+      "loss": 0.79,
+      "step": 7935
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.5148896775189016e-05,
+      "loss": 0.6629,
+      "step": 7936
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.5146325155582986e-05,
+      "loss": 0.8158,
+      "step": 7937
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.514375353597696e-05,
+      "loss": 0.7112,
+      "step": 7938
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.514118191637093e-05,
+      "loss": 0.7928,
+      "step": 7939
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.513861029676491e-05,
+      "loss": 0.7671,
+      "step": 7940
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.513603867715888e-05,
+      "loss": 0.8814,
+      "step": 7941
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.513346705755285e-05,
+      "loss": 0.693,
+      "step": 7942
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.513089543794682e-05,
+      "loss": 0.8168,
+      "step": 7943
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.5128323818340794e-05,
+      "loss": 0.8998,
+      "step": 7944
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.5125752198734763e-05,
+      "loss": 0.8714,
+      "step": 7945
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.512318057912874e-05,
+      "loss": 0.6507,
+      "step": 7946
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.512060895952271e-05,
+      "loss": 0.8511,
+      "step": 7947
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.511803733991668e-05,
+      "loss": 0.539,
+      "step": 7948
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.5115465720310656e-05,
+      "loss": 0.5748,
+      "step": 7949
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.5112894100704625e-05,
+      "loss": 0.8461,
+      "step": 7950
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.5110322481098595e-05,
+      "loss": 0.8733,
+      "step": 7951
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.510775086149257e-05,
+      "loss": 1.1371,
+      "step": 7952
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.510517924188654e-05,
+      "loss": 0.9639,
+      "step": 7953
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.510260762228052e-05,
+      "loss": 0.933,
+      "step": 7954
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.510003600267449e-05,
+      "loss": 0.909,
+      "step": 7955
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.509746438306846e-05,
+      "loss": 0.899,
+      "step": 7956
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.509489276346243e-05,
+      "loss": 0.42,
+      "step": 7957
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.50923211438564e-05,
+      "loss": 0.8412,
+      "step": 7958
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.508974952425037e-05,
+      "loss": 1.0777,
+      "step": 7959
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.508717790464435e-05,
+      "loss": 0.9384,
+      "step": 7960
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.508460628503832e-05,
+      "loss": 0.8447,
+      "step": 7961
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.508203466543229e-05,
+      "loss": 0.5195,
+      "step": 7962
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.507946304582626e-05,
+      "loss": 0.6331,
+      "step": 7963
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.5076891426220235e-05,
+      "loss": 0.6932,
+      "step": 7964
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.507431980661421e-05,
+      "loss": 1.1057,
+      "step": 7965
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.507174818700818e-05,
+      "loss": 0.8207,
+      "step": 7966
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.506917656740215e-05,
+      "loss": 0.7101,
+      "step": 7967
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.506660494779612e-05,
+      "loss": 0.4942,
+      "step": 7968
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.506403332819009e-05,
+      "loss": 0.7588,
+      "step": 7969
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.506146170858407e-05,
+      "loss": 0.8128,
+      "step": 7970
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.505889008897804e-05,
+      "loss": 0.6748,
+      "step": 7971
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.505631846937201e-05,
+      "loss": 0.8967,
+      "step": 7972
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.505374684976598e-05,
+      "loss": 1.0107,
+      "step": 7973
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.505117523015995e-05,
+      "loss": 0.9329,
+      "step": 7974
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.504860361055393e-05,
+      "loss": 0.8114,
+      "step": 7975
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.5046031990947905e-05,
+      "loss": 0.7605,
+      "step": 7976
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.5043460371341875e-05,
+      "loss": 1.1851,
+      "step": 7977
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.5040888751735844e-05,
+      "loss": 1.0379,
+      "step": 7978
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.5038317132129814e-05,
+      "loss": 0.9943,
+      "step": 7979
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.503574551252379e-05,
+      "loss": 0.9726,
+      "step": 7980
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.503317389291776e-05,
+      "loss": 0.5427,
+      "step": 7981
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.503060227331174e-05,
+      "loss": 0.7171,
+      "step": 7982
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.5028030653705706e-05,
+      "loss": 0.7626,
+      "step": 7983
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.5025459034099676e-05,
+      "loss": 0.7454,
+      "step": 7984
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.502288741449365e-05,
+      "loss": 0.7694,
+      "step": 7985
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.502031579488762e-05,
+      "loss": 0.9532,
+      "step": 7986
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.501774417528159e-05,
+      "loss": 0.8681,
+      "step": 7987
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.501517255567557e-05,
+      "loss": 0.6795,
+      "step": 7988
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.501260093606954e-05,
+      "loss": 0.8825,
+      "step": 7989
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.501002931646351e-05,
+      "loss": 0.7124,
+      "step": 7990
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.5007457696857484e-05,
+      "loss": 0.8627,
+      "step": 7991
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.5004886077251454e-05,
+      "loss": 0.9757,
+      "step": 7992
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.5002314457645424e-05,
+      "loss": 0.7058,
+      "step": 7993
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.49997428380394e-05,
+      "loss": 0.8522,
+      "step": 7994
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.499717121843337e-05,
+      "loss": 0.8246,
+      "step": 7995
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.4994599598827346e-05,
+      "loss": 0.5776,
+      "step": 7996
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.4992027979221316e-05,
+      "loss": 0.8754,
+      "step": 7997
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.4989456359615286e-05,
+      "loss": 0.9234,
+      "step": 7998
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.4986884740009255e-05,
+      "loss": 0.7558,
+      "step": 7999
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.498431312040323e-05,
+      "loss": 0.7882,
+      "step": 8000
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.498174150079721e-05,
+      "loss": 0.6314,
+      "step": 8001
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.497916988119118e-05,
+      "loss": 0.4671,
+      "step": 8002
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.497659826158515e-05,
+      "loss": 0.5826,
+      "step": 8003
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.497402664197912e-05,
+      "loss": 1.011,
+      "step": 8004
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.497145502237309e-05,
+      "loss": 0.7844,
+      "step": 8005
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.496888340276706e-05,
+      "loss": 0.8341,
+      "step": 8006
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.496631178316104e-05,
+      "loss": 0.7015,
+      "step": 8007
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.496374016355501e-05,
+      "loss": 0.7758,
+      "step": 8008
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.496116854394898e-05,
+      "loss": 1.0336,
+      "step": 8009
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.495859692434295e-05,
+      "loss": 0.7943,
+      "step": 8010
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.4956025304736925e-05,
+      "loss": 0.9767,
+      "step": 8011
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.49534536851309e-05,
+      "loss": 0.9193,
+      "step": 8012
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.495088206552487e-05,
+      "loss": 0.8183,
+      "step": 8013
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.494831044591884e-05,
+      "loss": 0.7743,
+      "step": 8014
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.494573882631281e-05,
+      "loss": 0.9932,
+      "step": 8015
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.494316720670678e-05,
+      "loss": 0.9593,
+      "step": 8016
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.494059558710076e-05,
+      "loss": 0.9236,
+      "step": 8017
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.4938023967494733e-05,
+      "loss": 0.3997,
+      "step": 8018
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.49354523478887e-05,
+      "loss": 0.7463,
+      "step": 8019
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.493288072828267e-05,
+      "loss": 0.8394,
+      "step": 8020
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.493030910867664e-05,
+      "loss": 0.7054,
+      "step": 8021
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.492773748907062e-05,
+      "loss": 1.069,
+      "step": 8022
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.492516586946459e-05,
+      "loss": 0.8482,
+      "step": 8023
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.4922594249858565e-05,
+      "loss": 0.8267,
+      "step": 8024
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.4920022630252535e-05,
+      "loss": 1.077,
+      "step": 8025
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.4917451010646504e-05,
+      "loss": 0.7981,
+      "step": 8026
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.491487939104048e-05,
+      "loss": 0.8489,
+      "step": 8027
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.491230777143445e-05,
+      "loss": 0.7089,
+      "step": 8028
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.490973615182842e-05,
+      "loss": 0.6535,
+      "step": 8029
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.49071645322224e-05,
+      "loss": 0.78,
+      "step": 8030
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.4904592912616366e-05,
+      "loss": 0.932,
+      "step": 8031
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.490202129301034e-05,
+      "loss": 0.9118,
+      "step": 8032
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.489944967340431e-05,
+      "loss": 0.7006,
+      "step": 8033
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.489687805379828e-05,
+      "loss": 1.3148,
+      "step": 8034
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.489430643419225e-05,
+      "loss": 0.4858,
+      "step": 8035
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.489173481458623e-05,
+      "loss": 0.8842,
+      "step": 8036
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.48891631949802e-05,
+      "loss": 0.8445,
+      "step": 8037
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.4886591575374175e-05,
+      "loss": 1.0521,
+      "step": 8038
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.4884019955768144e-05,
+      "loss": 0.6635,
+      "step": 8039
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.4881448336162114e-05,
+      "loss": 1.0038,
+      "step": 8040
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.487887671655609e-05,
+      "loss": 0.7926,
+      "step": 8041
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.487630509695006e-05,
+      "loss": 0.7835,
+      "step": 8042
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.4873733477344037e-05,
+      "loss": 1.0124,
+      "step": 8043
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.4871161857738006e-05,
+      "loss": 0.6364,
+      "step": 8044
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.4868590238131976e-05,
+      "loss": 0.5948,
+      "step": 8045
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.4866018618525946e-05,
+      "loss": 0.6374,
+      "step": 8046
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.486344699891992e-05,
+      "loss": 0.7015,
+      "step": 8047
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.48608753793139e-05,
+      "loss": 0.7318,
+      "step": 8048
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.485830375970787e-05,
+      "loss": 1.0446,
+      "step": 8049
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.485573214010184e-05,
+      "loss": 1.2036,
+      "step": 8050
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.485316052049581e-05,
+      "loss": 0.831,
+      "step": 8051
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.485058890088978e-05,
+      "loss": 0.8295,
+      "step": 8052
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.4848017281283754e-05,
+      "loss": 0.8116,
+      "step": 8053
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.484544566167773e-05,
+      "loss": 0.8268,
+      "step": 8054
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.48428740420717e-05,
+      "loss": 0.7924,
+      "step": 8055
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.484030242246567e-05,
+      "loss": 0.8481,
+      "step": 8056
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.483773080285964e-05,
+      "loss": 0.4869,
+      "step": 8057
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.4835159183253616e-05,
+      "loss": 0.4748,
+      "step": 8058
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.483258756364759e-05,
+      "loss": 0.9407,
+      "step": 8059
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.483001594404156e-05,
+      "loss": 1.033,
+      "step": 8060
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.482744432443553e-05,
+      "loss": 0.9411,
+      "step": 8061
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.48248727048295e-05,
+      "loss": 1.0857,
+      "step": 8062
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.482230108522348e-05,
+      "loss": 0.9697,
+      "step": 8063
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.481972946561745e-05,
+      "loss": 0.5838,
+      "step": 8064
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.4817157846011424e-05,
+      "loss": 0.8562,
+      "step": 8065
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.4814586226405394e-05,
+      "loss": 0.5973,
+      "step": 8066
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.481201460679936e-05,
+      "loss": 0.7852,
+      "step": 8067
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.480944298719333e-05,
+      "loss": 0.9043,
+      "step": 8068
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.480687136758731e-05,
+      "loss": 0.5318,
+      "step": 8069
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.480429974798128e-05,
+      "loss": 0.8614,
+      "step": 8070
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.4801728128375256e-05,
+      "loss": 0.8422,
+      "step": 8071
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.4799156508769225e-05,
+      "loss": 0.7345,
+      "step": 8072
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.4796584889163195e-05,
+      "loss": 1.0286,
+      "step": 8073
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.479401326955717e-05,
+      "loss": 0.7232,
+      "step": 8074
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.479144164995114e-05,
+      "loss": 0.8936,
+      "step": 8075
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.478887003034511e-05,
+      "loss": 0.9868,
+      "step": 8076
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.478629841073909e-05,
+      "loss": 0.6985,
+      "step": 8077
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.478372679113306e-05,
+      "loss": 0.5487,
+      "step": 8078
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.478115517152703e-05,
+      "loss": 1.1542,
+      "step": 8079
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.4778583551921e-05,
+      "loss": 1.0862,
+      "step": 8080
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.477601193231497e-05,
+      "loss": 0.7681,
+      "step": 8081
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.477344031270894e-05,
+      "loss": 0.6189,
+      "step": 8082
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.477086869310292e-05,
+      "loss": 0.8413,
+      "step": 8083
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.476829707349689e-05,
+      "loss": 0.8053,
+      "step": 8084
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.4765725453890865e-05,
+      "loss": 0.8107,
+      "step": 8085
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.4763153834284835e-05,
+      "loss": 0.8182,
+      "step": 8086
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.4760582214678804e-05,
+      "loss": 1.1642,
+      "step": 8087
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.4758010595072774e-05,
+      "loss": 0.9168,
+      "step": 8088
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.475543897546675e-05,
+      "loss": 0.8914,
+      "step": 8089
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.475286735586073e-05,
+      "loss": 0.6694,
+      "step": 8090
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.47502957362547e-05,
+      "loss": 0.8063,
+      "step": 8091
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.4747724116648666e-05,
+      "loss": 0.7912,
+      "step": 8092
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.4745152497042636e-05,
+      "loss": 0.8781,
+      "step": 8093
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.4742580877436606e-05,
+      "loss": 0.951,
+      "step": 8094
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.474000925783059e-05,
+      "loss": 0.9148,
+      "step": 8095
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.473743763822456e-05,
+      "loss": 0.7497,
+      "step": 8096
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.473486601861853e-05,
+      "loss": 0.7078,
+      "step": 8097
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.47322943990125e-05,
+      "loss": 0.6583,
+      "step": 8098
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.472972277940647e-05,
+      "loss": 0.8067,
+      "step": 8099
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.4727151159800444e-05,
+      "loss": 0.7308,
+      "step": 8100
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.472457954019442e-05,
+      "loss": 0.9886,
+      "step": 8101
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.472200792058839e-05,
+      "loss": 0.8621,
+      "step": 8102
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.471943630098236e-05,
+      "loss": 0.8116,
+      "step": 8103
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.471686468137633e-05,
+      "loss": 0.7438,
+      "step": 8104
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.4714293061770306e-05,
+      "loss": 0.8384,
+      "step": 8105
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.4711721442164276e-05,
+      "loss": 0.6409,
+      "step": 8106
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.470914982255825e-05,
+      "loss": 0.5435,
+      "step": 8107
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.470657820295222e-05,
+      "loss": 0.8351,
+      "step": 8108
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.470400658334619e-05,
+      "loss": 0.6853,
+      "step": 8109
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.470143496374017e-05,
+      "loss": 0.7112,
+      "step": 8110
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.469886334413414e-05,
+      "loss": 0.6076,
+      "step": 8111
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.469629172452811e-05,
+      "loss": 0.9446,
+      "step": 8112
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.4693720104922084e-05,
+      "loss": 0.6428,
+      "step": 8113
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.4691148485316054e-05,
+      "loss": 0.5054,
+      "step": 8114
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.468857686571002e-05,
+      "loss": 0.6973,
+      "step": 8115
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.4686005246104e-05,
+      "loss": 0.7636,
+      "step": 8116
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.468343362649797e-05,
+      "loss": 0.8192,
+      "step": 8117
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.468086200689194e-05,
+      "loss": 0.6531,
+      "step": 8118
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.4678290387285916e-05,
+      "loss": 0.6335,
+      "step": 8119
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.4675718767679885e-05,
+      "loss": 1.0892,
+      "step": 8120
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.467314714807386e-05,
+      "loss": 0.8759,
+      "step": 8121
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.467057552846783e-05,
+      "loss": 0.7337,
+      "step": 8122
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.46680039088618e-05,
+      "loss": 0.8667,
+      "step": 8123
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.466543228925577e-05,
+      "loss": 0.6893,
+      "step": 8124
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.466286066964975e-05,
+      "loss": 0.7433,
+      "step": 8125
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.4660289050043724e-05,
+      "loss": 0.7244,
+      "step": 8126
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.4657717430437693e-05,
+      "loss": 0.5699,
+      "step": 8127
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.465514581083166e-05,
+      "loss": 0.6377,
+      "step": 8128
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.465257419122563e-05,
+      "loss": 0.9316,
+      "step": 8129
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 3.46500025716196e-05,
+      "loss": 0.8522,
+      "step": 8130
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 3.464743095201358e-05,
+      "loss": 0.9525,
+      "step": 8131
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 3.4644859332407555e-05,
+      "loss": 0.7758,
+      "step": 8132
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 3.4642287712801525e-05,
+      "loss": 0.594,
+      "step": 8133
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 3.4639716093195495e-05,
+      "loss": 1.0066,
+      "step": 8134
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 3.4637144473589464e-05,
+      "loss": 0.8969,
+      "step": 8135
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 3.463457285398344e-05,
+      "loss": 0.6854,
+      "step": 8136
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 3.463200123437742e-05,
+      "loss": 0.7319,
+      "step": 8137
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 3.462942961477139e-05,
+      "loss": 0.6376,
+      "step": 8138
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 3.462685799516536e-05,
+      "loss": 0.8977,
+      "step": 8139
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 3.4624286375559326e-05,
+      "loss": 0.6457,
+      "step": 8140
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 3.46217147559533e-05,
+      "loss": 0.9122,
+      "step": 8141
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 3.461914313634727e-05,
+      "loss": 1.1492,
+      "step": 8142
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 3.461657151674125e-05,
+      "loss": 0.8853,
+      "step": 8143
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 3.461399989713522e-05,
+      "loss": 0.8927,
+      "step": 8144
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 3.461142827752919e-05,
+      "loss": 0.8464,
+      "step": 8145
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 3.460885665792316e-05,
+      "loss": 0.8332,
+      "step": 8146
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 3.4606285038317135e-05,
+      "loss": 0.7645,
+      "step": 8147
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 3.4603713418711104e-05,
+      "loss": 1.0248,
+      "step": 8148
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 3.460114179910508e-05,
+      "loss": 0.9223,
+      "step": 8149
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 3.459857017949905e-05,
+      "loss": 0.6715,
+      "step": 8150
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 3.459599855989302e-05,
+      "loss": 0.5846,
+      "step": 8151
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 3.4593426940286997e-05,
+      "loss": 0.5265,
+      "step": 8152
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 3.4590855320680966e-05,
+      "loss": 0.9334,
+      "step": 8153
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 3.4588283701074936e-05,
+      "loss": 0.942,
+      "step": 8154
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 3.458571208146891e-05,
+      "loss": 0.6412,
+      "step": 8155
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 3.458314046186288e-05,
+      "loss": 0.7969,
+      "step": 8156
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 3.458056884225686e-05,
+      "loss": 0.7499,
+      "step": 8157
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 3.457799722265083e-05,
+      "loss": 0.5958,
+      "step": 8158
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 3.45754256030448e-05,
+      "loss": 0.5083,
+      "step": 8159
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 3.457285398343877e-05,
+      "loss": 0.8453,
+      "step": 8160
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 3.4570282363832744e-05,
+      "loss": 1.177,
+      "step": 8161
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 3.4567710744226714e-05,
+      "loss": 1.0874,
+      "step": 8162
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 3.456513912462069e-05,
+      "loss": 0.7078,
+      "step": 8163
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 3.456256750501466e-05,
+      "loss": 0.6124,
+      "step": 8164
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 3.455999588540863e-05,
+      "loss": 0.6181,
+      "step": 8165
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 3.45574242658026e-05,
+      "loss": 0.9189,
+      "step": 8166
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 3.4554852646196576e-05,
+      "loss": 0.634,
+      "step": 8167
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 3.455228102659055e-05,
+      "loss": 0.8126,
+      "step": 8168
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 3.454970940698452e-05,
+      "loss": 0.4328,
+      "step": 8169
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 3.454713778737849e-05,
+      "loss": 0.9546,
+      "step": 8170
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 3.454456616777246e-05,
+      "loss": 0.62,
+      "step": 8171
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 3.454199454816643e-05,
+      "loss": 0.7875,
+      "step": 8172
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 3.4539422928560414e-05,
+      "loss": 0.908,
+      "step": 8173
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 3.4536851308954384e-05,
+      "loss": 0.5732,
+      "step": 8174
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 3.4534279689348353e-05,
+      "loss": 0.9219,
+      "step": 8175
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 3.453170806974232e-05,
+      "loss": 0.9443,
+      "step": 8176
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 3.452913645013629e-05,
+      "loss": 1.0205,
+      "step": 8177
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 3.452656483053027e-05,
+      "loss": 0.6467,
+      "step": 8178
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 3.4523993210924246e-05,
+      "loss": 1.1419,
+      "step": 8179
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 3.4521421591318215e-05,
+      "loss": 0.8228,
+      "step": 8180
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 3.4518849971712185e-05,
+      "loss": 0.8526,
+      "step": 8181
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 3.4516278352106155e-05,
+      "loss": 1.1358,
+      "step": 8182
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 3.451370673250013e-05,
+      "loss": 0.8327,
+      "step": 8183
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 3.45111351128941e-05,
+      "loss": 0.5843,
+      "step": 8184
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 3.450856349328808e-05,
+      "loss": 0.5655,
+      "step": 8185
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 3.450599187368205e-05,
+      "loss": 0.7273,
+      "step": 8186
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 3.450342025407602e-05,
+      "loss": 0.7125,
+      "step": 8187
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 3.450084863446999e-05,
+      "loss": 0.5302,
+      "step": 8188
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 3.449827701486396e-05,
+      "loss": 0.9321,
+      "step": 8189
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 3.449570539525794e-05,
+      "loss": 0.6275,
+      "step": 8190
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 3.449313377565191e-05,
+      "loss": 0.5738,
+      "step": 8191
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 3.449056215604588e-05,
+      "loss": 0.7238,
+      "step": 8192
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 3.448799053643985e-05,
+      "loss": 0.4931,
+      "step": 8193
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 3.4485418916833825e-05,
+      "loss": 0.7567,
+      "step": 8194
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 3.4482847297227795e-05,
+      "loss": 0.5402,
+      "step": 8195
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 3.448027567762177e-05,
+      "loss": 0.6339,
+      "step": 8196
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 3.447770405801574e-05,
+      "loss": 0.755,
+      "step": 8197
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 3.447513243840971e-05,
+      "loss": 0.4447,
+      "step": 8198
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 3.447256081880369e-05,
+      "loss": 0.5544,
+      "step": 8199
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 3.4469989199197657e-05,
+      "loss": 1.1695,
+      "step": 8200
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 3.4467417579591626e-05,
+      "loss": 0.799,
+      "step": 8201
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 3.44648459599856e-05,
+      "loss": 1.0488,
+      "step": 8202
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 3.446227434037957e-05,
+      "loss": 0.8713,
+      "step": 8203
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 3.445970272077355e-05,
+      "loss": 0.999,
+      "step": 8204
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 3.445713110116752e-05,
+      "loss": 0.3393,
+      "step": 8205
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 3.445455948156149e-05,
+      "loss": 0.6056,
+      "step": 8206
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 3.445198786195546e-05,
+      "loss": 0.8361,
+      "step": 8207
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 3.4449416242349434e-05,
+      "loss": 0.7091,
+      "step": 8208
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 3.4446844622743404e-05,
+      "loss": 0.7265,
+      "step": 8209
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 3.444427300313738e-05,
+      "loss": 0.834,
+      "step": 8210
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 3.444170138353135e-05,
+      "loss": 0.7805,
+      "step": 8211
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 3.443912976392532e-05,
+      "loss": 0.6272,
+      "step": 8212
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 3.443655814431929e-05,
+      "loss": 0.7149,
+      "step": 8213
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 3.4433986524713266e-05,
+      "loss": 0.829,
+      "step": 8214
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 3.443141490510724e-05,
+      "loss": 0.5622,
+      "step": 8215
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 3.442884328550121e-05,
+      "loss": 0.6438,
+      "step": 8216
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 3.442627166589518e-05,
+      "loss": 0.6055,
+      "step": 8217
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 3.442370004628915e-05,
+      "loss": 0.7141,
+      "step": 8218
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 3.442112842668312e-05,
+      "loss": 0.4789,
+      "step": 8219
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 3.4418556807077105e-05,
+      "loss": 0.7935,
+      "step": 8220
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 3.4415985187471074e-05,
+      "loss": 0.8959,
+      "step": 8221
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 3.4413413567865044e-05,
+      "loss": 1.1045,
+      "step": 8222
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 3.4410841948259014e-05,
+      "loss": 0.6547,
+      "step": 8223
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 3.440827032865298e-05,
+      "loss": 0.7016,
+      "step": 8224
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 3.440569870904696e-05,
+      "loss": 0.7506,
+      "step": 8225
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 3.4403127089440936e-05,
+      "loss": 1.0793,
+      "step": 8226
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 3.4400555469834906e-05,
+      "loss": 0.8957,
+      "step": 8227
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 3.4397983850228876e-05,
+      "loss": 1.044,
+      "step": 8228
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 3.4395412230622845e-05,
+      "loss": 0.9766,
+      "step": 8229
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 3.439284061101682e-05,
+      "loss": 0.8739,
+      "step": 8230
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 3.439026899141079e-05,
+      "loss": 0.716,
+      "step": 8231
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 3.438769737180477e-05,
+      "loss": 0.8185,
+      "step": 8232
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 3.438512575219874e-05,
+      "loss": 0.965,
+      "step": 8233
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 3.438255413259271e-05,
+      "loss": 0.8087,
+      "step": 8234
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 3.4379982512986684e-05,
+      "loss": 0.7818,
+      "step": 8235
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 3.437741089338065e-05,
+      "loss": 0.5039,
+      "step": 8236
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 3.437483927377462e-05,
+      "loss": 0.9962,
+      "step": 8237
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 3.43722676541686e-05,
+      "loss": 0.6264,
+      "step": 8238
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 3.436969603456257e-05,
+      "loss": 0.8867,
+      "step": 8239
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 3.436712441495654e-05,
+      "loss": 1.0843,
+      "step": 8240
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 3.4364552795350515e-05,
+      "loss": 0.746,
+      "step": 8241
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 3.4361981175744485e-05,
+      "loss": 0.7486,
+      "step": 8242
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 3.4359409556138455e-05,
+      "loss": 0.7397,
+      "step": 8243
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 3.435683793653243e-05,
+      "loss": 0.6649,
+      "step": 8244
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 3.43542663169264e-05,
+      "loss": 0.78,
+      "step": 8245
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 3.435169469732038e-05,
+      "loss": 0.5456,
+      "step": 8246
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 3.434912307771435e-05,
+      "loss": 0.6591,
+      "step": 8247
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 3.434655145810832e-05,
+      "loss": 1.1459,
+      "step": 8248
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 3.4343979838502286e-05,
+      "loss": 0.8557,
+      "step": 8249
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 3.434140821889626e-05,
+      "loss": 0.621,
+      "step": 8250
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 3.433883659929024e-05,
+      "loss": 1.1383,
+      "step": 8251
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 3.433626497968421e-05,
+      "loss": 0.6459,
+      "step": 8252
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 3.433369336007818e-05,
+      "loss": 0.6547,
+      "step": 8253
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 3.433112174047215e-05,
+      "loss": 0.6924,
+      "step": 8254
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 3.432855012086612e-05,
+      "loss": 0.9629,
+      "step": 8255
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 3.4325978501260095e-05,
+      "loss": 0.8082,
+      "step": 8256
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 3.432340688165407e-05,
+      "loss": 0.4529,
+      "step": 8257
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 3.432083526204804e-05,
+      "loss": 0.8957,
+      "step": 8258
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 3.431826364244201e-05,
+      "loss": 0.8122,
+      "step": 8259
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 3.431569202283598e-05,
+      "loss": 1.084,
+      "step": 8260
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 3.4313120403229956e-05,
+      "loss": 0.8321,
+      "step": 8261
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 3.431054878362393e-05,
+      "loss": 0.9334,
+      "step": 8262
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 3.43079771640179e-05,
+      "loss": 1.0397,
+      "step": 8263
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 3.430540554441187e-05,
+      "loss": 0.8181,
+      "step": 8264
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 3.430283392480584e-05,
+      "loss": 0.9473,
+      "step": 8265
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 3.430026230519982e-05,
+      "loss": 0.6762,
+      "step": 8266
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 3.429769068559379e-05,
+      "loss": 0.5905,
+      "step": 8267
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 3.4295119065987765e-05,
+      "loss": 0.9907,
+      "step": 8268
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 3.4292547446381734e-05,
+      "loss": 0.6203,
+      "step": 8269
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 3.4289975826775704e-05,
+      "loss": 0.8462,
+      "step": 8270
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 3.4287404207169674e-05,
+      "loss": 0.7505,
+      "step": 8271
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 3.428483258756365e-05,
+      "loss": 0.6972,
+      "step": 8272
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 3.428226096795762e-05,
+      "loss": 0.8317,
+      "step": 8273
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 3.4279689348351596e-05,
+      "loss": 0.6619,
+      "step": 8274
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 3.4277117728745566e-05,
+      "loss": 0.519,
+      "step": 8275
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 3.4274546109139536e-05,
+      "loss": 0.9471,
+      "step": 8276
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 3.427197448953351e-05,
+      "loss": 0.8585,
+      "step": 8277
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 3.426940286992748e-05,
+      "loss": 0.8124,
+      "step": 8278
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 3.426683125032145e-05,
+      "loss": 0.8036,
+      "step": 8279
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 3.426425963071543e-05,
+      "loss": 0.8367,
+      "step": 8280
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 3.42616880111094e-05,
+      "loss": 0.8517,
+      "step": 8281
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 3.4259116391503374e-05,
+      "loss": 0.7912,
+      "step": 8282
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 3.4256544771897344e-05,
+      "loss": 0.6088,
+      "step": 8283
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 3.4253973152291313e-05,
+      "loss": 0.9446,
+      "step": 8284
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 3.425140153268528e-05,
+      "loss": 1.0965,
+      "step": 8285
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 3.424882991307926e-05,
+      "loss": 0.7443,
+      "step": 8286
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 3.424625829347323e-05,
+      "loss": 0.8252,
+      "step": 8287
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 3.4243686673867206e-05,
+      "loss": 1.0812,
+      "step": 8288
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 3.4241115054261175e-05,
+      "loss": 0.5321,
+      "step": 8289
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 3.4238543434655145e-05,
+      "loss": 0.8802,
+      "step": 8290
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 3.4235971815049115e-05,
+      "loss": 1.1207,
+      "step": 8291
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 3.423340019544309e-05,
+      "loss": 0.5252,
+      "step": 8292
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 3.423082857583707e-05,
+      "loss": 0.9332,
+      "step": 8293
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 3.422825695623104e-05,
+      "loss": 0.8481,
+      "step": 8294
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 3.422568533662501e-05,
+      "loss": 0.6254,
+      "step": 8295
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 3.422311371701898e-05,
+      "loss": 1.4053,
+      "step": 8296
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 3.4220542097412946e-05,
+      "loss": 1.045,
+      "step": 8297
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 3.421797047780693e-05,
+      "loss": 0.9059,
+      "step": 8298
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 3.42153988582009e-05,
+      "loss": 1.0902,
+      "step": 8299
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 3.421282723859487e-05,
+      "loss": 0.7653,
+      "step": 8300
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 3.421025561898884e-05,
+      "loss": 0.8988,
+      "step": 8301
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 3.420768399938281e-05,
+      "loss": 0.5093,
+      "step": 8302
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 3.4205112379776785e-05,
+      "loss": 0.5514,
+      "step": 8303
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 3.420254076017076e-05,
+      "loss": 0.8186,
+      "step": 8304
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 3.419996914056473e-05,
+      "loss": 0.7324,
+      "step": 8305
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 3.41973975209587e-05,
+      "loss": 0.7271,
+      "step": 8306
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 3.419482590135267e-05,
+      "loss": 0.9991,
+      "step": 8307
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 3.419225428174665e-05,
+      "loss": 0.5199,
+      "step": 8308
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 3.4189682662140617e-05,
+      "loss": 0.7329,
+      "step": 8309
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 3.418711104253459e-05,
+      "loss": 0.9582,
+      "step": 8310
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 3.418453942292856e-05,
+      "loss": 0.8079,
+      "step": 8311
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 3.418196780332253e-05,
+      "loss": 1.055,
+      "step": 8312
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 3.417939618371651e-05,
+      "loss": 0.6403,
+      "step": 8313
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 3.417682456411048e-05,
+      "loss": 1.0559,
+      "step": 8314
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 3.417425294450445e-05,
+      "loss": 1.0274,
+      "step": 8315
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 3.4171681324898425e-05,
+      "loss": 0.8104,
+      "step": 8316
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 3.4169109705292394e-05,
+      "loss": 0.7614,
+      "step": 8317
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 3.4166538085686364e-05,
+      "loss": 0.4643,
+      "step": 8318
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 3.416396646608034e-05,
+      "loss": 1.095,
+      "step": 8319
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 3.416139484647431e-05,
+      "loss": 0.7397,
+      "step": 8320
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 3.415882322686828e-05,
+      "loss": 0.6878,
+      "step": 8321
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 3.4156251607262256e-05,
+      "loss": 0.8192,
+      "step": 8322
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 3.4153679987656226e-05,
+      "loss": 0.7397,
+      "step": 8323
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 3.41511083680502e-05,
+      "loss": 0.5942,
+      "step": 8324
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 3.414853674844417e-05,
+      "loss": 0.8455,
+      "step": 8325
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 3.414596512883814e-05,
+      "loss": 1.0169,
+      "step": 8326
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 3.414339350923211e-05,
+      "loss": 0.9049,
+      "step": 8327
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 3.414082188962609e-05,
+      "loss": 0.7079,
+      "step": 8328
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 3.4138250270020064e-05,
+      "loss": 0.6704,
+      "step": 8329
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 3.4135678650414034e-05,
+      "loss": 0.7655,
+      "step": 8330
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 3.4133107030808004e-05,
+      "loss": 0.4776,
+      "step": 8331
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 3.4130535411201974e-05,
+      "loss": 1.0032,
+      "step": 8332
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 3.412796379159594e-05,
+      "loss": 1.0416,
+      "step": 8333
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 3.412539217198992e-05,
+      "loss": 0.7285,
+      "step": 8334
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 3.4122820552383896e-05,
+      "loss": 0.7644,
+      "step": 8335
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 3.4120248932777866e-05,
+      "loss": 0.8454,
+      "step": 8336
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 3.4117677313171836e-05,
+      "loss": 0.9606,
+      "step": 8337
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 3.4115105693565805e-05,
+      "loss": 0.8456,
+      "step": 8338
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 3.411253407395978e-05,
+      "loss": 0.7054,
+      "step": 8339
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 3.410996245435376e-05,
+      "loss": 0.7928,
+      "step": 8340
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 3.410739083474773e-05,
+      "loss": 0.8492,
+      "step": 8341
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 3.41048192151417e-05,
+      "loss": 0.6009,
+      "step": 8342
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 3.410224759553567e-05,
+      "loss": 1.112,
+      "step": 8343
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 3.409967597592964e-05,
+      "loss": 0.7749,
+      "step": 8344
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 3.409710435632362e-05,
+      "loss": 0.8783,
+      "step": 8345
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 3.409453273671759e-05,
+      "loss": 0.7685,
+      "step": 8346
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 3.409196111711156e-05,
+      "loss": 0.8432,
+      "step": 8347
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 3.408938949750553e-05,
+      "loss": 0.9753,
+      "step": 8348
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 3.40868178778995e-05,
+      "loss": 0.7643,
+      "step": 8349
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 3.4084246258293475e-05,
+      "loss": 0.7804,
+      "step": 8350
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 3.408167463868745e-05,
+      "loss": 0.753,
+      "step": 8351
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 3.407910301908142e-05,
+      "loss": 0.6874,
+      "step": 8352
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 3.407653139947539e-05,
+      "loss": 0.8161,
+      "step": 8353
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 3.407395977986936e-05,
+      "loss": 0.7633,
+      "step": 8354
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 3.407138816026334e-05,
+      "loss": 0.9176,
+      "step": 8355
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 3.406881654065731e-05,
+      "loss": 0.8274,
+      "step": 8356
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 3.4066244921051283e-05,
+      "loss": 0.5465,
+      "step": 8357
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 3.406367330144525e-05,
+      "loss": 0.6583,
+      "step": 8358
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 3.406110168183922e-05,
+      "loss": 0.5138,
+      "step": 8359
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 3.40585300622332e-05,
+      "loss": 0.7656,
+      "step": 8360
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 3.405595844262717e-05,
+      "loss": 0.8119,
+      "step": 8361
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 3.405338682302114e-05,
+      "loss": 0.7095,
+      "step": 8362
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 3.4050815203415115e-05,
+      "loss": 0.788,
+      "step": 8363
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 3.4048243583809085e-05,
+      "loss": 0.6292,
+      "step": 8364
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 3.4045671964203054e-05,
+      "loss": 0.9397,
+      "step": 8365
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 3.404310034459703e-05,
+      "loss": 0.5853,
+      "step": 8366
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 3.4040528724991e-05,
+      "loss": 0.8209,
+      "step": 8367
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 3.403795710538497e-05,
+      "loss": 0.7222,
+      "step": 8368
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 3.403538548577895e-05,
+      "loss": 0.6656,
+      "step": 8369
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 3.4032813866172916e-05,
+      "loss": 0.8846,
+      "step": 8370
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 3.403024224656689e-05,
+      "loss": 0.9483,
+      "step": 8371
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 3.402767062696086e-05,
+      "loss": 0.6202,
+      "step": 8372
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 3.402509900735483e-05,
+      "loss": 0.6409,
+      "step": 8373
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 3.40225273877488e-05,
+      "loss": 0.6617,
+      "step": 8374
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 3.401995576814278e-05,
+      "loss": 0.5992,
+      "step": 8375
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 3.4017384148536755e-05,
+      "loss": 0.8004,
+      "step": 8376
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 3.4014812528930725e-05,
+      "loss": 0.5108,
+      "step": 8377
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 3.4012240909324694e-05,
+      "loss": 0.8721,
+      "step": 8378
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 3.4009669289718664e-05,
+      "loss": 0.8129,
+      "step": 8379
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 3.4007097670112634e-05,
+      "loss": 0.8414,
+      "step": 8380
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 3.400452605050661e-05,
+      "loss": 1.0028,
+      "step": 8381
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 3.4001954430900587e-05,
+      "loss": 0.9065,
+      "step": 8382
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 3.3999382811294556e-05,
+      "loss": 0.7106,
+      "step": 8383
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 3.3996811191688526e-05,
+      "loss": 0.793,
+      "step": 8384
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 3.3994239572082496e-05,
+      "loss": 0.7793,
+      "step": 8385
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 3.399166795247647e-05,
+      "loss": 0.9976,
+      "step": 8386
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 3.398909633287045e-05,
+      "loss": 1.0831,
+      "step": 8387
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 3.398652471326442e-05,
+      "loss": 0.7717,
+      "step": 8388
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 3.398395309365839e-05,
+      "loss": 0.6908,
+      "step": 8389
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 3.398138147405236e-05,
+      "loss": 0.734,
+      "step": 8390
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 3.3978809854446334e-05,
+      "loss": 0.8779,
+      "step": 8391
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 3.3976238234840304e-05,
+      "loss": 0.9725,
+      "step": 8392
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 3.397366661523428e-05,
+      "loss": 0.6182,
+      "step": 8393
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 3.397109499562825e-05,
+      "loss": 0.5811,
+      "step": 8394
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 3.396852337602222e-05,
+      "loss": 0.7049,
+      "step": 8395
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 3.396595175641619e-05,
+      "loss": 0.9627,
+      "step": 8396
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 3.3963380136810166e-05,
+      "loss": 0.8205,
+      "step": 8397
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 3.3960808517204135e-05,
+      "loss": 0.5369,
+      "step": 8398
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 3.395823689759811e-05,
+      "loss": 0.7794,
+      "step": 8399
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 3.395566527799208e-05,
+      "loss": 1.5813,
+      "step": 8400
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 3.395309365838605e-05,
+      "loss": 0.6496,
+      "step": 8401
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 3.395052203878003e-05,
+      "loss": 0.6392,
+      "step": 8402
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 3.3947950419174e-05,
+      "loss": 1.006,
+      "step": 8403
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 3.394537879956797e-05,
+      "loss": 0.6575,
+      "step": 8404
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 3.3942807179961944e-05,
+      "loss": 0.9961,
+      "step": 8405
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 3.394023556035591e-05,
+      "loss": 1.3627,
+      "step": 8406
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 3.393766394074989e-05,
+      "loss": 0.9053,
+      "step": 8407
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 3.393509232114386e-05,
+      "loss": 0.7084,
+      "step": 8408
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 3.393252070153783e-05,
+      "loss": 0.7524,
+      "step": 8409
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 3.39299490819318e-05,
+      "loss": 0.6868,
+      "step": 8410
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 3.3927377462325775e-05,
+      "loss": 0.8447,
+      "step": 8411
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 3.3924805842719745e-05,
+      "loss": 0.5226,
+      "step": 8412
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 3.392223422311372e-05,
+      "loss": 0.5111,
+      "step": 8413
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 3.391966260350769e-05,
+      "loss": 0.7953,
+      "step": 8414
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 3.391709098390166e-05,
+      "loss": 0.5571,
+      "step": 8415
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 3.391451936429563e-05,
+      "loss": 0.6783,
+      "step": 8416
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 3.391194774468961e-05,
+      "loss": 0.6125,
+      "step": 8417
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 3.390937612508358e-05,
+      "loss": 0.8853,
+      "step": 8418
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 3.390680450547755e-05,
+      "loss": 0.6203,
+      "step": 8419
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 3.390423288587152e-05,
+      "loss": 0.6129,
+      "step": 8420
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 3.390166126626549e-05,
+      "loss": 0.7603,
+      "step": 8421
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 3.389908964665946e-05,
+      "loss": 0.7288,
+      "step": 8422
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 3.3896518027053445e-05,
+      "loss": 0.4715,
+      "step": 8423
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 3.3893946407447415e-05,
+      "loss": 1.0067,
+      "step": 8424
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 3.3891374787841385e-05,
+      "loss": 0.5549,
+      "step": 8425
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 3.3888803168235354e-05,
+      "loss": 1.069,
+      "step": 8426
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 3.3886231548629324e-05,
+      "loss": 0.4402,
+      "step": 8427
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 3.38836599290233e-05,
+      "loss": 0.6629,
+      "step": 8428
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 3.388108830941728e-05,
+      "loss": 0.9946,
+      "step": 8429
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 3.387851668981125e-05,
+      "loss": 0.6266,
+      "step": 8430
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 3.3875945070205216e-05,
+      "loss": 1.2217,
+      "step": 8431
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 3.3873373450599186e-05,
+      "loss": 0.6125,
+      "step": 8432
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 3.387080183099316e-05,
+      "loss": 0.996,
+      "step": 8433
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 3.386823021138713e-05,
+      "loss": 0.6678,
+      "step": 8434
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 3.386565859178111e-05,
+      "loss": 0.68,
+      "step": 8435
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 3.386308697217508e-05,
+      "loss": 0.961,
+      "step": 8436
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 3.386051535256905e-05,
+      "loss": 1.0005,
+      "step": 8437
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 3.3857943732963024e-05,
+      "loss": 0.7177,
+      "step": 8438
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 3.3855372113356994e-05,
+      "loss": 0.6735,
+      "step": 8439
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 3.3852800493750964e-05,
+      "loss": 0.5885,
+      "step": 8440
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 3.385022887414494e-05,
+      "loss": 0.7642,
+      "step": 8441
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 3.384765725453891e-05,
+      "loss": 0.923,
+      "step": 8442
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 3.384508563493288e-05,
+      "loss": 0.6991,
+      "step": 8443
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 3.3842514015326856e-05,
+      "loss": 0.7576,
+      "step": 8444
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 3.3839942395720826e-05,
+      "loss": 0.65,
+      "step": 8445
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 3.3837370776114796e-05,
+      "loss": 0.7569,
+      "step": 8446
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 3.383479915650877e-05,
+      "loss": 0.5497,
+      "step": 8447
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 3.383222753690274e-05,
+      "loss": 0.571,
+      "step": 8448
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 3.382965591729672e-05,
+      "loss": 0.7315,
+      "step": 8449
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 3.382708429769069e-05,
+      "loss": 0.6736,
+      "step": 8450
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 3.382451267808466e-05,
+      "loss": 0.9713,
+      "step": 8451
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 3.382194105847863e-05,
+      "loss": 0.6125,
+      "step": 8452
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 3.3819369438872604e-05,
+      "loss": 0.9082,
+      "step": 8453
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 3.381679781926658e-05,
+      "loss": 0.8167,
+      "step": 8454
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 3.381422619966055e-05,
+      "loss": 0.9473,
+      "step": 8455
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 3.381165458005452e-05,
+      "loss": 0.7426,
+      "step": 8456
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 3.380908296044849e-05,
+      "loss": 0.839,
+      "step": 8457
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 3.380651134084246e-05,
+      "loss": 0.6174,
+      "step": 8458
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 3.3803939721236435e-05,
+      "loss": 0.662,
+      "step": 8459
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 3.380136810163041e-05,
+      "loss": 0.8847,
+      "step": 8460
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 3.379879648202438e-05,
+      "loss": 0.8323,
+      "step": 8461
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 3.379622486241835e-05,
+      "loss": 0.8056,
+      "step": 8462
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 3.379365324281232e-05,
+      "loss": 0.8203,
+      "step": 8463
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 3.37910816232063e-05,
+      "loss": 0.8475,
+      "step": 8464
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 3.3788510003600274e-05,
+      "loss": 0.8732,
+      "step": 8465
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 3.3785938383994243e-05,
+      "loss": 0.4246,
+      "step": 8466
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 3.378336676438821e-05,
+      "loss": 0.6482,
+      "step": 8467
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 3.378079514478218e-05,
+      "loss": 0.7867,
+      "step": 8468
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 3.377822352517615e-05,
+      "loss": 0.7413,
+      "step": 8469
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 3.377565190557013e-05,
+      "loss": 0.8874,
+      "step": 8470
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 3.3773080285964105e-05,
+      "loss": 0.7088,
+      "step": 8471
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 3.3770508666358075e-05,
+      "loss": 0.8905,
+      "step": 8472
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 3.3767937046752045e-05,
+      "loss": 0.9255,
+      "step": 8473
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 3.3765365427146014e-05,
+      "loss": 0.7784,
+      "step": 8474
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 3.376279380753999e-05,
+      "loss": 0.8812,
+      "step": 8475
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 3.376022218793396e-05,
+      "loss": 0.7378,
+      "step": 8476
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 3.375765056832794e-05,
+      "loss": 1.0078,
+      "step": 8477
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 3.375507894872191e-05,
+      "loss": 0.8884,
+      "step": 8478
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 3.3752507329115876e-05,
+      "loss": 0.6346,
+      "step": 8479
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 3.374993570950985e-05,
+      "loss": 0.6766,
+      "step": 8480
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 3.374736408990382e-05,
+      "loss": 0.9877,
+      "step": 8481
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 3.374479247029779e-05,
+      "loss": 0.6814,
+      "step": 8482
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 3.374222085069177e-05,
+      "loss": 0.7615,
+      "step": 8483
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 3.373964923108574e-05,
+      "loss": 0.5352,
+      "step": 8484
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 3.3737077611479715e-05,
+      "loss": 0.6282,
+      "step": 8485
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 3.3734505991873685e-05,
+      "loss": 0.652,
+      "step": 8486
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 3.3731934372267654e-05,
+      "loss": 1.0031,
+      "step": 8487
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 3.3729362752661624e-05,
+      "loss": 1.0272,
+      "step": 8488
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 3.37267911330556e-05,
+      "loss": 0.6865,
+      "step": 8489
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 3.372421951344957e-05,
+      "loss": 1.0164,
+      "step": 8490
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 3.3721647893843547e-05,
+      "loss": 0.5703,
+      "step": 8491
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 3.3719076274237516e-05,
+      "loss": 0.8306,
+      "step": 8492
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 3.3716504654631486e-05,
+      "loss": 1.0444,
+      "step": 8493
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 3.371393303502546e-05,
+      "loss": 0.4882,
+      "step": 8494
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 3.371136141541943e-05,
+      "loss": 0.5866,
+      "step": 8495
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 3.370878979581341e-05,
+      "loss": 0.6425,
+      "step": 8496
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 3.370621817620738e-05,
+      "loss": 0.7348,
+      "step": 8497
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 3.370364655660135e-05,
+      "loss": 0.6724,
+      "step": 8498
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 3.370107493699532e-05,
+      "loss": 0.465,
+      "step": 8499
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 3.3698503317389294e-05,
+      "loss": 0.7445,
+      "step": 8500
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 3.369593169778327e-05,
+      "loss": 1.0365,
+      "step": 8501
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 3.369336007817724e-05,
+      "loss": 0.8099,
+      "step": 8502
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 3.369078845857121e-05,
+      "loss": 0.8948,
+      "step": 8503
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 3.368821683896518e-05,
+      "loss": 0.791,
+      "step": 8504
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 3.368564521935915e-05,
+      "loss": 0.8427,
+      "step": 8505
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 3.3683073599753126e-05,
+      "loss": 0.8131,
+      "step": 8506
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 3.36805019801471e-05,
+      "loss": 1.0535,
+      "step": 8507
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 3.367793036054107e-05,
+      "loss": 0.795,
+      "step": 8508
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 3.367535874093504e-05,
+      "loss": 0.796,
+      "step": 8509
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 3.367278712132901e-05,
+      "loss": 0.8723,
+      "step": 8510
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 3.367021550172299e-05,
+      "loss": 0.5947,
+      "step": 8511
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 3.3667643882116964e-05,
+      "loss": 0.6382,
+      "step": 8512
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 3.3665072262510934e-05,
+      "loss": 0.6938,
+      "step": 8513
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 3.3662500642904904e-05,
+      "loss": 0.9445,
+      "step": 8514
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 3.365992902329887e-05,
+      "loss": 0.6825,
+      "step": 8515
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 3.365735740369285e-05,
+      "loss": 0.6389,
+      "step": 8516
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 3.365478578408682e-05,
+      "loss": 0.7889,
+      "step": 8517
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 3.3652214164480796e-05,
+      "loss": 1.1191,
+      "step": 8518
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 3.3649642544874765e-05,
+      "loss": 0.5044,
+      "step": 8519
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 3.3647070925268735e-05,
+      "loss": 1.1137,
+      "step": 8520
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 3.3644499305662705e-05,
+      "loss": 0.7126,
+      "step": 8521
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 3.364192768605668e-05,
+      "loss": 0.4704,
+      "step": 8522
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 3.363935606645065e-05,
+      "loss": 0.7567,
+      "step": 8523
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 3.363678444684463e-05,
+      "loss": 0.8416,
+      "step": 8524
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 3.36342128272386e-05,
+      "loss": 0.9806,
+      "step": 8525
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 3.363164120763257e-05,
+      "loss": 0.7291,
+      "step": 8526
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 3.362906958802654e-05,
+      "loss": 0.5858,
+      "step": 8527
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 3.362649796842051e-05,
+      "loss": 0.9267,
+      "step": 8528
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 3.362392634881448e-05,
+      "loss": 0.6486,
+      "step": 8529
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 3.362135472920846e-05,
+      "loss": 0.6709,
+      "step": 8530
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 3.361878310960243e-05,
+      "loss": 0.7047,
+      "step": 8531
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 3.3616211489996405e-05,
+      "loss": 0.8128,
+      "step": 8532
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 3.3613639870390375e-05,
+      "loss": 0.6333,
+      "step": 8533
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 3.3611068250784345e-05,
+      "loss": 0.706,
+      "step": 8534
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 3.3608496631178314e-05,
+      "loss": 0.8833,
+      "step": 8535
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 3.360592501157229e-05,
+      "loss": 0.621,
+      "step": 8536
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 3.360335339196626e-05,
+      "loss": 0.8058,
+      "step": 8537
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 3.360078177236024e-05,
+      "loss": 0.7518,
+      "step": 8538
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 3.359821015275421e-05,
+      "loss": 0.7019,
+      "step": 8539
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 3.3595638533148176e-05,
+      "loss": 0.7714,
+      "step": 8540
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 3.3593066913542146e-05,
+      "loss": 0.7404,
+      "step": 8541
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 3.359049529393612e-05,
+      "loss": 0.8917,
+      "step": 8542
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 3.35879236743301e-05,
+      "loss": 0.5989,
+      "step": 8543
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 3.358535205472407e-05,
+      "loss": 0.8926,
+      "step": 8544
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 3.358278043511804e-05,
+      "loss": 0.7935,
+      "step": 8545
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 3.358020881551201e-05,
+      "loss": 0.5534,
+      "step": 8546
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 3.357763719590598e-05,
+      "loss": 0.8279,
+      "step": 8547
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 3.357506557629996e-05,
+      "loss": 0.7312,
+      "step": 8548
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 3.357249395669393e-05,
+      "loss": 0.7768,
+      "step": 8549
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 3.35699223370879e-05,
+      "loss": 0.8598,
+      "step": 8550
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 3.356735071748187e-05,
+      "loss": 1.3293,
+      "step": 8551
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 3.356477909787584e-05,
+      "loss": 0.8858,
+      "step": 8552
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 3.3562207478269816e-05,
+      "loss": 0.9193,
+      "step": 8553
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 3.355963585866379e-05,
+      "loss": 0.7568,
+      "step": 8554
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 3.355706423905776e-05,
+      "loss": 0.994,
+      "step": 8555
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 3.355449261945173e-05,
+      "loss": 1.2496,
+      "step": 8556
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 3.35519209998457e-05,
+      "loss": 0.9267,
+      "step": 8557
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 3.354934938023968e-05,
+      "loss": 0.5535,
+      "step": 8558
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 3.354677776063365e-05,
+      "loss": 0.6094,
+      "step": 8559
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 3.3544206141027624e-05,
+      "loss": 0.6455,
+      "step": 8560
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 3.3541634521421594e-05,
+      "loss": 0.8282,
+      "step": 8561
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.3539062901815564e-05,
+      "loss": 0.6705,
+      "step": 8562
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.353649128220954e-05,
+      "loss": 0.7275,
+      "step": 8563
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.353391966260351e-05,
+      "loss": 1.0247,
+      "step": 8564
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.353134804299748e-05,
+      "loss": 1.1724,
+      "step": 8565
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.3528776423391456e-05,
+      "loss": 0.9893,
+      "step": 8566
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.3526204803785426e-05,
+      "loss": 0.8235,
+      "step": 8567
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.3523633184179395e-05,
+      "loss": 0.635,
+      "step": 8568
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.352106156457337e-05,
+      "loss": 0.6899,
+      "step": 8569
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.351848994496734e-05,
+      "loss": 0.9357,
+      "step": 8570
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.351591832536131e-05,
+      "loss": 0.7937,
+      "step": 8571
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.351334670575529e-05,
+      "loss": 0.8674,
+      "step": 8572
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.351077508614926e-05,
+      "loss": 0.5076,
+      "step": 8573
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.3508203466543234e-05,
+      "loss": 1.1125,
+      "step": 8574
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.35056318469372e-05,
+      "loss": 0.9462,
+      "step": 8575
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.350306022733117e-05,
+      "loss": 0.8526,
+      "step": 8576
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.350048860772514e-05,
+      "loss": 0.6734,
+      "step": 8577
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.349791698811912e-05,
+      "loss": 0.6103,
+      "step": 8578
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.3495345368513096e-05,
+      "loss": 0.7017,
+      "step": 8579
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.3492773748907065e-05,
+      "loss": 0.7865,
+      "step": 8580
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.3490202129301035e-05,
+      "loss": 0.9059,
+      "step": 8581
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.3487630509695005e-05,
+      "loss": 0.7754,
+      "step": 8582
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.3485058890088974e-05,
+      "loss": 0.5313,
+      "step": 8583
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.348248727048295e-05,
+      "loss": 0.6524,
+      "step": 8584
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.347991565087693e-05,
+      "loss": 0.6906,
+      "step": 8585
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.34773440312709e-05,
+      "loss": 1.0206,
+      "step": 8586
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.347477241166487e-05,
+      "loss": 0.4776,
+      "step": 8587
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.3472200792058836e-05,
+      "loss": 0.6019,
+      "step": 8588
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.346962917245281e-05,
+      "loss": 0.8115,
+      "step": 8589
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.346705755284679e-05,
+      "loss": 0.9963,
+      "step": 8590
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.346448593324076e-05,
+      "loss": 0.6846,
+      "step": 8591
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.346191431363473e-05,
+      "loss": 1.0548,
+      "step": 8592
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.34593426940287e-05,
+      "loss": 0.6766,
+      "step": 8593
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.345677107442267e-05,
+      "loss": 0.5335,
+      "step": 8594
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.3454199454816645e-05,
+      "loss": 0.9607,
+      "step": 8595
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.345162783521062e-05,
+      "loss": 0.7634,
+      "step": 8596
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.344905621560459e-05,
+      "loss": 1.2745,
+      "step": 8597
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.344648459599856e-05,
+      "loss": 1.0339,
+      "step": 8598
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.344391297639253e-05,
+      "loss": 1.2407,
+      "step": 8599
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.3441341356786507e-05,
+      "loss": 0.5903,
+      "step": 8600
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.3438769737180476e-05,
+      "loss": 0.8767,
+      "step": 8601
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.343619811757445e-05,
+      "loss": 0.6139,
+      "step": 8602
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.343362649796842e-05,
+      "loss": 0.8591,
+      "step": 8603
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.343105487836239e-05,
+      "loss": 0.9324,
+      "step": 8604
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.342848325875637e-05,
+      "loss": 0.7929,
+      "step": 8605
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.342591163915034e-05,
+      "loss": 0.7829,
+      "step": 8606
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.342334001954431e-05,
+      "loss": 0.6552,
+      "step": 8607
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.3420768399938284e-05,
+      "loss": 0.8239,
+      "step": 8608
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.3418196780332254e-05,
+      "loss": 0.617,
+      "step": 8609
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.341562516072623e-05,
+      "loss": 0.9216,
+      "step": 8610
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.34130535411202e-05,
+      "loss": 1.0076,
+      "step": 8611
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.341048192151417e-05,
+      "loss": 0.7721,
+      "step": 8612
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.340791030190814e-05,
+      "loss": 0.5709,
+      "step": 8613
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.3405338682302116e-05,
+      "loss": 0.7952,
+      "step": 8614
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.3402767062696086e-05,
+      "loss": 0.5895,
+      "step": 8615
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.340019544309006e-05,
+      "loss": 1.0045,
+      "step": 8616
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.339762382348403e-05,
+      "loss": 0.6992,
+      "step": 8617
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.3395052203878e-05,
+      "loss": 0.9189,
+      "step": 8618
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.339248058427197e-05,
+      "loss": 0.603,
+      "step": 8619
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.338990896466595e-05,
+      "loss": 1.0278,
+      "step": 8620
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.3387337345059924e-05,
+      "loss": 0.9924,
+      "step": 8621
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.3384765725453894e-05,
+      "loss": 0.7386,
+      "step": 8622
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.3382194105847863e-05,
+      "loss": 0.7591,
+      "step": 8623
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.337962248624183e-05,
+      "loss": 1.0574,
+      "step": 8624
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.33770508666358e-05,
+      "loss": 0.5196,
+      "step": 8625
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.3374479247029786e-05,
+      "loss": 0.5683,
+      "step": 8626
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.3371907627423756e-05,
+      "loss": 1.0167,
+      "step": 8627
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.3369336007817725e-05,
+      "loss": 0.8719,
+      "step": 8628
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.3366764388211695e-05,
+      "loss": 0.5882,
+      "step": 8629
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.3364192768605665e-05,
+      "loss": 1.0198,
+      "step": 8630
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.336162114899964e-05,
+      "loss": 1.0158,
+      "step": 8631
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.335904952939362e-05,
+      "loss": 1.045,
+      "step": 8632
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.335647790978759e-05,
+      "loss": 0.7863,
+      "step": 8633
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.335390629018156e-05,
+      "loss": 0.5446,
+      "step": 8634
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.335133467057553e-05,
+      "loss": 1.0751,
+      "step": 8635
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.33487630509695e-05,
+      "loss": 0.6696,
+      "step": 8636
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.334619143136347e-05,
+      "loss": 1.0589,
+      "step": 8637
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.334361981175745e-05,
+      "loss": 0.5042,
+      "step": 8638
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.334104819215142e-05,
+      "loss": 1.0076,
+      "step": 8639
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.333847657254539e-05,
+      "loss": 0.6106,
+      "step": 8640
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.3335904952939365e-05,
+      "loss": 0.732,
+      "step": 8641
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.3333333333333335e-05,
+      "loss": 0.5909,
+      "step": 8642
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.333076171372731e-05,
+      "loss": 0.7498,
+      "step": 8643
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.332819009412128e-05,
+      "loss": 1.255,
+      "step": 8644
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.332561847451525e-05,
+      "loss": 1.0259,
+      "step": 8645
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.332304685490922e-05,
+      "loss": 0.8772,
+      "step": 8646
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.33204752353032e-05,
+      "loss": 0.7881,
+      "step": 8647
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.3317903615697167e-05,
+      "loss": 0.6897,
+      "step": 8648
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.331533199609114e-05,
+      "loss": 0.6844,
+      "step": 8649
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.331276037648511e-05,
+      "loss": 0.5866,
+      "step": 8650
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.331018875687908e-05,
+      "loss": 0.7753,
+      "step": 8651
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.330761713727306e-05,
+      "loss": 0.6969,
+      "step": 8652
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.330504551766703e-05,
+      "loss": 0.8928,
+      "step": 8653
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.3302473898061e-05,
+      "loss": 0.6548,
+      "step": 8654
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.3299902278454975e-05,
+      "loss": 0.9018,
+      "step": 8655
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.3297330658848944e-05,
+      "loss": 0.4016,
+      "step": 8656
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.329475903924292e-05,
+      "loss": 1.1603,
+      "step": 8657
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.329218741963689e-05,
+      "loss": 0.5541,
+      "step": 8658
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.328961580003086e-05,
+      "loss": 0.833,
+      "step": 8659
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.328704418042483e-05,
+      "loss": 0.7927,
+      "step": 8660
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.3284472560818806e-05,
+      "loss": 0.7855,
+      "step": 8661
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.3281900941212776e-05,
+      "loss": 0.51,
+      "step": 8662
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.327932932160675e-05,
+      "loss": 0.8171,
+      "step": 8663
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.327675770200072e-05,
+      "loss": 0.8895,
+      "step": 8664
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.327418608239469e-05,
+      "loss": 0.5593,
+      "step": 8665
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.327161446278866e-05,
+      "loss": 0.9291,
+      "step": 8666
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.326904284318264e-05,
+      "loss": 0.6057,
+      "step": 8667
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.3266471223576615e-05,
+      "loss": 0.74,
+      "step": 8668
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.3263899603970584e-05,
+      "loss": 0.8356,
+      "step": 8669
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.3261327984364554e-05,
+      "loss": 0.8561,
+      "step": 8670
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.3258756364758524e-05,
+      "loss": 0.9908,
+      "step": 8671
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.325618474515249e-05,
+      "loss": 0.602,
+      "step": 8672
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.3253613125546476e-05,
+      "loss": 1.181,
+      "step": 8673
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.3251041505940446e-05,
+      "loss": 0.8561,
+      "step": 8674
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.3248469886334416e-05,
+      "loss": 0.7055,
+      "step": 8675
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.3245898266728386e-05,
+      "loss": 0.5986,
+      "step": 8676
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.3243326647122355e-05,
+      "loss": 1.0143,
+      "step": 8677
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.324075502751633e-05,
+      "loss": 1.0524,
+      "step": 8678
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.323818340791031e-05,
+      "loss": 0.7145,
+      "step": 8679
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.323561178830428e-05,
+      "loss": 0.8227,
+      "step": 8680
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.323304016869825e-05,
+      "loss": 0.9587,
+      "step": 8681
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.323046854909222e-05,
+      "loss": 0.5085,
+      "step": 8682
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.3227896929486194e-05,
+      "loss": 0.7129,
+      "step": 8683
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.322532530988016e-05,
+      "loss": 0.907,
+      "step": 8684
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.322275369027414e-05,
+      "loss": 0.6671,
+      "step": 8685
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.322018207066811e-05,
+      "loss": 0.7326,
+      "step": 8686
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.321761045106208e-05,
+      "loss": 0.6972,
+      "step": 8687
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.3215038831456056e-05,
+      "loss": 0.7093,
+      "step": 8688
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.3212467211850025e-05,
+      "loss": 0.6134,
+      "step": 8689
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.3209895592243995e-05,
+      "loss": 0.7892,
+      "step": 8690
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.320732397263797e-05,
+      "loss": 0.8205,
+      "step": 8691
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.320475235303194e-05,
+      "loss": 0.8898,
+      "step": 8692
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.320218073342591e-05,
+      "loss": 0.8598,
+      "step": 8693
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.319960911381989e-05,
+      "loss": 0.8179,
+      "step": 8694
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.319703749421386e-05,
+      "loss": 1.0401,
+      "step": 8695
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.319446587460783e-05,
+      "loss": 0.7732,
+      "step": 8696
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.31918942550018e-05,
+      "loss": 0.9823,
+      "step": 8697
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.318932263539577e-05,
+      "loss": 0.6619,
+      "step": 8698
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.318675101578975e-05,
+      "loss": 0.7887,
+      "step": 8699
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.318417939618372e-05,
+      "loss": 0.7509,
+      "step": 8700
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.318160777657769e-05,
+      "loss": 0.9558,
+      "step": 8701
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.317903615697166e-05,
+      "loss": 0.5956,
+      "step": 8702
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.3176464537365635e-05,
+      "loss": 0.8137,
+      "step": 8703
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.317389291775961e-05,
+      "loss": 0.8053,
+      "step": 8704
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.317132129815358e-05,
+      "loss": 0.5972,
+      "step": 8705
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.316874967854755e-05,
+      "loss": 0.7332,
+      "step": 8706
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.316617805894152e-05,
+      "loss": 0.7089,
+      "step": 8707
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.316360643933549e-05,
+      "loss": 0.7581,
+      "step": 8708
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.3161034819729466e-05,
+      "loss": 0.9307,
+      "step": 8709
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.315846320012344e-05,
+      "loss": 0.8611,
+      "step": 8710
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.315589158051741e-05,
+      "loss": 0.7125,
+      "step": 8711
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.315331996091138e-05,
+      "loss": 0.8784,
+      "step": 8712
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.315074834130535e-05,
+      "loss": 0.9596,
+      "step": 8713
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.314817672169933e-05,
+      "loss": 0.964,
+      "step": 8714
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.3145605102093305e-05,
+      "loss": 0.6407,
+      "step": 8715
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.3143033482487275e-05,
+      "loss": 0.7213,
+      "step": 8716
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.3140461862881244e-05,
+      "loss": 0.8546,
+      "step": 8717
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.3137890243275214e-05,
+      "loss": 0.7634,
+      "step": 8718
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.3135318623669184e-05,
+      "loss": 1.285,
+      "step": 8719
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.313274700406316e-05,
+      "loss": 0.7823,
+      "step": 8720
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.3130175384457137e-05,
+      "loss": 0.8951,
+      "step": 8721
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.3127603764851106e-05,
+      "loss": 0.8507,
+      "step": 8722
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.3125032145245076e-05,
+      "loss": 0.8676,
+      "step": 8723
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 3.3122460525639046e-05,
+      "loss": 0.596,
+      "step": 8724
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 3.311988890603302e-05,
+      "loss": 0.8675,
+      "step": 8725
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 3.311731728642699e-05,
+      "loss": 0.8921,
+      "step": 8726
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 3.311474566682097e-05,
+      "loss": 0.8992,
+      "step": 8727
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 3.311217404721494e-05,
+      "loss": 1.1634,
+      "step": 8728
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 3.310960242760891e-05,
+      "loss": 0.8479,
+      "step": 8729
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 3.3107030808002884e-05,
+      "loss": 0.6175,
+      "step": 8730
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 3.3104459188396854e-05,
+      "loss": 0.5758,
+      "step": 8731
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 3.3101887568790823e-05,
+      "loss": 0.7737,
+      "step": 8732
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 3.30993159491848e-05,
+      "loss": 0.7676,
+      "step": 8733
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 3.309674432957877e-05,
+      "loss": 1.1305,
+      "step": 8734
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 3.3094172709972746e-05,
+      "loss": 0.5307,
+      "step": 8735
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 3.3091601090366716e-05,
+      "loss": 0.9755,
+      "step": 8736
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 3.3089029470760685e-05,
+      "loss": 0.6244,
+      "step": 8737
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 3.3086457851154655e-05,
+      "loss": 0.9283,
+      "step": 8738
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 3.308388623154863e-05,
+      "loss": 0.9769,
+      "step": 8739
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 3.30813146119426e-05,
+      "loss": 0.6811,
+      "step": 8740
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 3.307874299233658e-05,
+      "loss": 0.6557,
+      "step": 8741
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 3.307617137273055e-05,
+      "loss": 0.6001,
+      "step": 8742
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 3.307359975312452e-05,
+      "loss": 0.9774,
+      "step": 8743
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 3.307102813351849e-05,
+      "loss": 1.0259,
+      "step": 8744
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 3.306845651391246e-05,
+      "loss": 1.0526,
+      "step": 8745
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 3.306588489430644e-05,
+      "loss": 0.9483,
+      "step": 8746
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 3.306331327470041e-05,
+      "loss": 0.6675,
+      "step": 8747
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 3.306074165509438e-05,
+      "loss": 0.7689,
+      "step": 8748
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 3.305817003548835e-05,
+      "loss": 0.8856,
+      "step": 8749
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 3.305559841588232e-05,
+      "loss": 0.6704,
+      "step": 8750
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 3.30530267962763e-05,
+      "loss": 0.9491,
+      "step": 8751
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 3.305045517667027e-05,
+      "loss": 0.7297,
+      "step": 8752
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 3.304788355706424e-05,
+      "loss": 0.7118,
+      "step": 8753
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 3.304531193745821e-05,
+      "loss": 0.4786,
+      "step": 8754
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 3.304274031785218e-05,
+      "loss": 0.5701,
+      "step": 8755
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 3.304016869824616e-05,
+      "loss": 0.8989,
+      "step": 8756
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 3.303759707864013e-05,
+      "loss": 0.6763,
+      "step": 8757
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 3.30350254590341e-05,
+      "loss": 0.7545,
+      "step": 8758
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 3.303245383942807e-05,
+      "loss": 0.7284,
+      "step": 8759
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 3.302988221982204e-05,
+      "loss": 0.7476,
+      "step": 8760
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 3.302731060021602e-05,
+      "loss": 0.9438,
+      "step": 8761
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 3.302473898060999e-05,
+      "loss": 0.6662,
+      "step": 8762
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 3.3022167361003965e-05,
+      "loss": 0.8274,
+      "step": 8763
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 3.3019595741397935e-05,
+      "loss": 0.9222,
+      "step": 8764
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 3.3017024121791904e-05,
+      "loss": 0.6268,
+      "step": 8765
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 3.301445250218588e-05,
+      "loss": 0.8568,
+      "step": 8766
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 3.301188088257985e-05,
+      "loss": 0.5697,
+      "step": 8767
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 3.300930926297382e-05,
+      "loss": 0.7908,
+      "step": 8768
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 3.30067376433678e-05,
+      "loss": 0.5599,
+      "step": 8769
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 3.3004166023761766e-05,
+      "loss": 0.7157,
+      "step": 8770
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 3.3001594404155736e-05,
+      "loss": 0.6815,
+      "step": 8771
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 3.299902278454971e-05,
+      "loss": 1.0542,
+      "step": 8772
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 3.299645116494368e-05,
+      "loss": 0.8786,
+      "step": 8773
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 3.299387954533765e-05,
+      "loss": 0.6947,
+      "step": 8774
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 3.299130792573163e-05,
+      "loss": 0.8889,
+      "step": 8775
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 3.29887363061256e-05,
+      "loss": 1.2896,
+      "step": 8776
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 3.2986164686519574e-05,
+      "loss": 0.9608,
+      "step": 8777
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 3.2983593066913544e-05,
+      "loss": 0.3666,
+      "step": 8778
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 3.2981021447307514e-05,
+      "loss": 0.4555,
+      "step": 8779
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 3.2978449827701484e-05,
+      "loss": 0.7125,
+      "step": 8780
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 3.297587820809546e-05,
+      "loss": 1.0732,
+      "step": 8781
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 3.2973306588489436e-05,
+      "loss": 0.5021,
+      "step": 8782
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 3.2970734968883406e-05,
+      "loss": 0.9809,
+      "step": 8783
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 3.2968163349277376e-05,
+      "loss": 0.8784,
+      "step": 8784
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 3.2965591729671346e-05,
+      "loss": 0.8962,
+      "step": 8785
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 3.2963020110065315e-05,
+      "loss": 0.9787,
+      "step": 8786
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 3.296044849045929e-05,
+      "loss": 0.6699,
+      "step": 8787
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 3.295787687085327e-05,
+      "loss": 0.723,
+      "step": 8788
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 3.295530525124724e-05,
+      "loss": 0.8276,
+      "step": 8789
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 3.295273363164121e-05,
+      "loss": 0.808,
+      "step": 8790
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 3.295016201203518e-05,
+      "loss": 0.7853,
+      "step": 8791
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 3.2947590392429154e-05,
+      "loss": 0.8415,
+      "step": 8792
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 3.294501877282313e-05,
+      "loss": 0.8314,
+      "step": 8793
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 3.29424471532171e-05,
+      "loss": 0.5931,
+      "step": 8794
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 3.293987553361107e-05,
+      "loss": 0.6083,
+      "step": 8795
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 3.293730391400504e-05,
+      "loss": 0.9822,
+      "step": 8796
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 3.293473229439901e-05,
+      "loss": 0.8116,
+      "step": 8797
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 3.293216067479299e-05,
+      "loss": 0.7953,
+      "step": 8798
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 3.292958905518696e-05,
+      "loss": 1.2135,
+      "step": 8799
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 3.292701743558093e-05,
+      "loss": 1.0058,
+      "step": 8800
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 3.29244458159749e-05,
+      "loss": 1.0689,
+      "step": 8801
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 3.292187419636887e-05,
+      "loss": 0.8695,
+      "step": 8802
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 3.291930257676285e-05,
+      "loss": 0.8661,
+      "step": 8803
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 3.2916730957156824e-05,
+      "loss": 0.8581,
+      "step": 8804
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 3.2914159337550793e-05,
+      "loss": 0.5882,
+      "step": 8805
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 3.291158771794476e-05,
+      "loss": 0.5868,
+      "step": 8806
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 3.290901609833873e-05,
+      "loss": 0.9244,
+      "step": 8807
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 3.290644447873271e-05,
+      "loss": 0.665,
+      "step": 8808
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 3.290387285912668e-05,
+      "loss": 0.6669,
+      "step": 8809
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 3.2901301239520655e-05,
+      "loss": 1.1129,
+      "step": 8810
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 3.2898729619914625e-05,
+      "loss": 0.6176,
+      "step": 8811
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 3.2896158000308595e-05,
+      "loss": 0.7506,
+      "step": 8812
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 3.289358638070257e-05,
+      "loss": 0.6597,
+      "step": 8813
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 3.289101476109654e-05,
+      "loss": 0.9271,
+      "step": 8814
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 3.288844314149051e-05,
+      "loss": 0.7615,
+      "step": 8815
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 3.288587152188449e-05,
+      "loss": 0.8755,
+      "step": 8816
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 3.288329990227846e-05,
+      "loss": 0.6564,
+      "step": 8817
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 3.2880728282672426e-05,
+      "loss": 0.5577,
+      "step": 8818
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 3.28781566630664e-05,
+      "loss": 0.9727,
+      "step": 8819
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 3.287558504346037e-05,
+      "loss": 0.8235,
+      "step": 8820
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 3.287301342385434e-05,
+      "loss": 1.0747,
+      "step": 8821
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 3.287044180424832e-05,
+      "loss": 0.7709,
+      "step": 8822
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 3.286787018464229e-05,
+      "loss": 0.7722,
+      "step": 8823
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 3.2865298565036265e-05,
+      "loss": 0.7907,
+      "step": 8824
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 3.2862726945430235e-05,
+      "loss": 0.8367,
+      "step": 8825
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 3.2860155325824204e-05,
+      "loss": 0.8757,
+      "step": 8826
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 3.2857583706218174e-05,
+      "loss": 0.6388,
+      "step": 8827
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 3.285501208661215e-05,
+      "loss": 0.6491,
+      "step": 8828
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 3.285244046700613e-05,
+      "loss": 0.8425,
+      "step": 8829
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 3.2849868847400097e-05,
+      "loss": 1.0238,
+      "step": 8830
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 3.2847297227794066e-05,
+      "loss": 0.9092,
+      "step": 8831
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 3.2844725608188036e-05,
+      "loss": 0.6266,
+      "step": 8832
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 3.2842153988582006e-05,
+      "loss": 0.8069,
+      "step": 8833
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 3.283958236897598e-05,
+      "loss": 0.6267,
+      "step": 8834
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 3.283701074936996e-05,
+      "loss": 0.4642,
+      "step": 8835
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 3.283443912976393e-05,
+      "loss": 0.6781,
+      "step": 8836
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 3.28318675101579e-05,
+      "loss": 0.7478,
+      "step": 8837
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 3.282929589055187e-05,
+      "loss": 0.7237,
+      "step": 8838
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 3.2826724270945844e-05,
+      "loss": 0.6246,
+      "step": 8839
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 3.282415265133982e-05,
+      "loss": 0.6298,
+      "step": 8840
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 3.282158103173379e-05,
+      "loss": 0.8777,
+      "step": 8841
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 3.281900941212776e-05,
+      "loss": 0.7768,
+      "step": 8842
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 3.281643779252173e-05,
+      "loss": 0.7448,
+      "step": 8843
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 3.2813866172915706e-05,
+      "loss": 0.6325,
+      "step": 8844
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 3.2811294553309676e-05,
+      "loss": 0.7954,
+      "step": 8845
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 3.280872293370365e-05,
+      "loss": 0.7776,
+      "step": 8846
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 3.280615131409762e-05,
+      "loss": 0.6063,
+      "step": 8847
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 3.280357969449159e-05,
+      "loss": 0.6418,
+      "step": 8848
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 3.280100807488556e-05,
+      "loss": 0.8107,
+      "step": 8849
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 3.279843645527954e-05,
+      "loss": 0.7759,
+      "step": 8850
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 3.279586483567351e-05,
+      "loss": 0.7974,
+      "step": 8851
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 3.2793293216067484e-05,
+      "loss": 1.0118,
+      "step": 8852
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 3.2790721596461454e-05,
+      "loss": 1.0863,
+      "step": 8853
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 3.278814997685542e-05,
+      "loss": 0.9723,
+      "step": 8854
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 3.27855783572494e-05,
+      "loss": 0.6731,
+      "step": 8855
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 3.278300673764337e-05,
+      "loss": 0.815,
+      "step": 8856
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 3.278043511803734e-05,
+      "loss": 0.7954,
+      "step": 8857
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 3.2777863498431315e-05,
+      "loss": 0.6627,
+      "step": 8858
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 3.2775291878825285e-05,
+      "loss": 0.713,
+      "step": 8859
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 3.277272025921926e-05,
+      "loss": 0.605,
+      "step": 8860
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 3.277014863961323e-05,
+      "loss": 0.869,
+      "step": 8861
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 3.27675770200072e-05,
+      "loss": 0.8736,
+      "step": 8862
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 3.276500540040117e-05,
+      "loss": 0.9762,
+      "step": 8863
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 3.276243378079515e-05,
+      "loss": 0.6718,
+      "step": 8864
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 3.275986216118912e-05,
+      "loss": 0.7588,
+      "step": 8865
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 3.275729054158309e-05,
+      "loss": 0.8296,
+      "step": 8866
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 3.275471892197706e-05,
+      "loss": 0.6762,
+      "step": 8867
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 3.275214730237103e-05,
+      "loss": 0.8017,
+      "step": 8868
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 3.2749575682765e-05,
+      "loss": 0.6517,
+      "step": 8869
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 3.274700406315898e-05,
+      "loss": 0.659,
+      "step": 8870
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 3.2744432443552955e-05,
+      "loss": 0.8236,
+      "step": 8871
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 3.2741860823946925e-05,
+      "loss": 0.5615,
+      "step": 8872
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 3.2739289204340895e-05,
+      "loss": 0.9061,
+      "step": 8873
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 3.2736717584734864e-05,
+      "loss": 1.0262,
+      "step": 8874
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 3.2734145965128834e-05,
+      "loss": 0.848,
+      "step": 8875
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 3.273157434552282e-05,
+      "loss": 0.5518,
+      "step": 8876
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 3.272900272591679e-05,
+      "loss": 0.8366,
+      "step": 8877
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 3.272643110631076e-05,
+      "loss": 0.6551,
+      "step": 8878
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 3.2723859486704726e-05,
+      "loss": 0.7604,
+      "step": 8879
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 3.2721287867098696e-05,
+      "loss": 0.7545,
+      "step": 8880
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 3.271871624749267e-05,
+      "loss": 0.5154,
+      "step": 8881
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 3.271614462788665e-05,
+      "loss": 0.4194,
+      "step": 8882
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 3.271357300828062e-05,
+      "loss": 0.717,
+      "step": 8883
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 3.271100138867459e-05,
+      "loss": 0.9819,
+      "step": 8884
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 3.270842976906856e-05,
+      "loss": 0.7506,
+      "step": 8885
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3.2705858149462534e-05,
+      "loss": 0.5588,
+      "step": 8886
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3.2703286529856504e-05,
+      "loss": 0.8825,
+      "step": 8887
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3.270071491025048e-05,
+      "loss": 0.8831,
+      "step": 8888
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3.269814329064445e-05,
+      "loss": 0.7994,
+      "step": 8889
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3.269557167103842e-05,
+      "loss": 0.7447,
+      "step": 8890
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3.2693000051432396e-05,
+      "loss": 0.6489,
+      "step": 8891
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3.2690428431826366e-05,
+      "loss": 0.7751,
+      "step": 8892
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3.2687856812220336e-05,
+      "loss": 1.0527,
+      "step": 8893
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3.268528519261431e-05,
+      "loss": 0.6417,
+      "step": 8894
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3.268271357300828e-05,
+      "loss": 0.678,
+      "step": 8895
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3.268014195340225e-05,
+      "loss": 0.9043,
+      "step": 8896
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3.267757033379623e-05,
+      "loss": 0.6092,
+      "step": 8897
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3.26749987141902e-05,
+      "loss": 0.7846,
+      "step": 8898
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3.267242709458417e-05,
+      "loss": 0.6517,
+      "step": 8899
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3.2669855474978144e-05,
+      "loss": 0.6931,
+      "step": 8900
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3.2667283855372114e-05,
+      "loss": 0.7576,
+      "step": 8901
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3.266471223576609e-05,
+      "loss": 0.8118,
+      "step": 8902
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3.266214061616006e-05,
+      "loss": 0.6642,
+      "step": 8903
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3.265956899655403e-05,
+      "loss": 0.6661,
+      "step": 8904
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3.2656997376948e-05,
+      "loss": 1.0942,
+      "step": 8905
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3.2654425757341976e-05,
+      "loss": 0.7902,
+      "step": 8906
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3.265185413773595e-05,
+      "loss": 1.2256,
+      "step": 8907
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3.264928251812992e-05,
+      "loss": 0.4526,
+      "step": 8908
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3.264671089852389e-05,
+      "loss": 0.7512,
+      "step": 8909
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3.264413927891786e-05,
+      "loss": 0.832,
+      "step": 8910
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3.264156765931183e-05,
+      "loss": 0.8581,
+      "step": 8911
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3.263899603970581e-05,
+      "loss": 0.9962,
+      "step": 8912
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3.2636424420099784e-05,
+      "loss": 1.1071,
+      "step": 8913
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3.2633852800493753e-05,
+      "loss": 0.5609,
+      "step": 8914
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3.263128118088772e-05,
+      "loss": 0.5718,
+      "step": 8915
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3.262870956128169e-05,
+      "loss": 1.0095,
+      "step": 8916
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3.262613794167567e-05,
+      "loss": 1.0541,
+      "step": 8917
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3.2623566322069646e-05,
+      "loss": 0.9265,
+      "step": 8918
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3.2620994702463615e-05,
+      "loss": 0.8019,
+      "step": 8919
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3.2618423082857585e-05,
+      "loss": 0.5501,
+      "step": 8920
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3.2615851463251555e-05,
+      "loss": 0.6524,
+      "step": 8921
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3.2613279843645524e-05,
+      "loss": 0.5591,
+      "step": 8922
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3.26107082240395e-05,
+      "loss": 0.5755,
+      "step": 8923
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3.260813660443348e-05,
+      "loss": 0.8148,
+      "step": 8924
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3.260556498482745e-05,
+      "loss": 0.7861,
+      "step": 8925
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3.260299336522142e-05,
+      "loss": 0.6732,
+      "step": 8926
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3.2600421745615386e-05,
+      "loss": 0.8049,
+      "step": 8927
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3.259785012600936e-05,
+      "loss": 0.6952,
+      "step": 8928
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3.259527850640333e-05,
+      "loss": 0.4717,
+      "step": 8929
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3.259270688679731e-05,
+      "loss": 0.728,
+      "step": 8930
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3.259013526719128e-05,
+      "loss": 0.9163,
+      "step": 8931
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3.258756364758525e-05,
+      "loss": 1.0435,
+      "step": 8932
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3.2584992027979225e-05,
+      "loss": 0.7601,
+      "step": 8933
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3.2582420408373195e-05,
+      "loss": 0.84,
+      "step": 8934
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3.2579848788767164e-05,
+      "loss": 0.6839,
+      "step": 8935
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3.257727716916114e-05,
+      "loss": 0.9334,
+      "step": 8936
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3.257470554955511e-05,
+      "loss": 0.9787,
+      "step": 8937
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3.257213392994909e-05,
+      "loss": 0.8651,
+      "step": 8938
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3.2569562310343057e-05,
+      "loss": 0.8883,
+      "step": 8939
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 3.2566990690737026e-05,
+      "loss": 0.5344,
+      "step": 8940
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 3.2564419071130996e-05,
+      "loss": 0.7869,
+      "step": 8941
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 3.256184745152497e-05,
+      "loss": 0.7085,
+      "step": 8942
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 3.255927583191894e-05,
+      "loss": 0.6458,
+      "step": 8943
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 3.255670421231292e-05,
+      "loss": 0.6813,
+      "step": 8944
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 3.255413259270689e-05,
+      "loss": 0.6537,
+      "step": 8945
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 3.255156097310086e-05,
+      "loss": 1.1299,
+      "step": 8946
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 3.2548989353494834e-05,
+      "loss": 0.7084,
+      "step": 8947
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 3.2546417733888804e-05,
+      "loss": 0.8338,
+      "step": 8948
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 3.254384611428278e-05,
+      "loss": 0.8373,
+      "step": 8949
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 3.254127449467675e-05,
+      "loss": 0.6783,
+      "step": 8950
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 3.253870287507072e-05,
+      "loss": 0.8212,
+      "step": 8951
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 3.253613125546469e-05,
+      "loss": 0.8748,
+      "step": 8952
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 3.2533559635858666e-05,
+      "loss": 0.8146,
+      "step": 8953
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 3.253098801625264e-05,
+      "loss": 0.5739,
+      "step": 8954
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 3.252841639664661e-05,
+      "loss": 0.9134,
+      "step": 8955
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 3.252584477704058e-05,
+      "loss": 0.8372,
+      "step": 8956
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 3.252327315743455e-05,
+      "loss": 0.7008,
+      "step": 8957
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 3.252070153782852e-05,
+      "loss": 0.7578,
+      "step": 8958
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 3.25181299182225e-05,
+      "loss": 0.6792,
+      "step": 8959
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 3.2515558298616474e-05,
+      "loss": 0.7227,
+      "step": 8960
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 3.2512986679010444e-05,
+      "loss": 0.9645,
+      "step": 8961
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 3.2510415059404413e-05,
+      "loss": 0.5661,
+      "step": 8962
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 3.250784343979838e-05,
+      "loss": 0.8889,
+      "step": 8963
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 3.250527182019236e-05,
+      "loss": 0.5483,
+      "step": 8964
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 3.2502700200586336e-05,
+      "loss": 0.6849,
+      "step": 8965
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 3.2500128580980306e-05,
+      "loss": 0.6719,
+      "step": 8966
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 3.2497556961374275e-05,
+      "loss": 0.8689,
+      "step": 8967
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 3.2494985341768245e-05,
+      "loss": 0.956,
+      "step": 8968
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 3.249241372216222e-05,
+      "loss": 0.3393,
+      "step": 8969
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 3.248984210255619e-05,
+      "loss": 1.0775,
+      "step": 8970
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 3.248727048295017e-05,
+      "loss": 0.5775,
+      "step": 8971
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 3.248469886334414e-05,
+      "loss": 0.7858,
+      "step": 8972
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 3.248212724373811e-05,
+      "loss": 0.9716,
+      "step": 8973
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 3.247955562413208e-05,
+      "loss": 0.4941,
+      "step": 8974
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 3.247698400452605e-05,
+      "loss": 0.6807,
+      "step": 8975
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 3.247441238492002e-05,
+      "loss": 0.5766,
+      "step": 8976
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 3.2471840765314e-05,
+      "loss": 0.5746,
+      "step": 8977
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 3.246926914570797e-05,
+      "loss": 0.9089,
+      "step": 8978
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 3.246669752610194e-05,
+      "loss": 0.78,
+      "step": 8979
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 3.2464125906495915e-05,
+      "loss": 0.893,
+      "step": 8980
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 3.2461554286889885e-05,
+      "loss": 0.842,
+      "step": 8981
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 3.2458982667283855e-05,
+      "loss": 0.7867,
+      "step": 8982
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 3.245641104767783e-05,
+      "loss": 0.6304,
+      "step": 8983
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 3.24538394280718e-05,
+      "loss": 1.0398,
+      "step": 8984
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 3.245126780846578e-05,
+      "loss": 0.7384,
+      "step": 8985
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 3.244869618885975e-05,
+      "loss": 0.7213,
+      "step": 8986
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 3.2446124569253717e-05,
+      "loss": 1.183,
+      "step": 8987
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 3.2443552949647686e-05,
+      "loss": 1.0182,
+      "step": 8988
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 3.244098133004166e-05,
+      "loss": 0.8078,
+      "step": 8989
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 3.243840971043563e-05,
+      "loss": 0.9465,
+      "step": 8990
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 3.243583809082961e-05,
+      "loss": 0.9399,
+      "step": 8991
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 3.243326647122358e-05,
+      "loss": 0.8488,
+      "step": 8992
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 3.243069485161755e-05,
+      "loss": 0.6276,
+      "step": 8993
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 3.242812323201152e-05,
+      "loss": 1.1775,
+      "step": 8994
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 3.2425551612405494e-05,
+      "loss": 0.513,
+      "step": 8995
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 3.242297999279947e-05,
+      "loss": 0.811,
+      "step": 8996
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 3.242040837319344e-05,
+      "loss": 1.0793,
+      "step": 8997
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 3.241783675358741e-05,
+      "loss": 1.0467,
+      "step": 8998
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 3.241526513398138e-05,
+      "loss": 0.7731,
+      "step": 8999
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 3.241269351437535e-05,
+      "loss": 0.5685,
+      "step": 9000
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 3.241012189476933e-05,
+      "loss": 0.6111,
+      "step": 9001
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 3.24075502751633e-05,
+      "loss": 0.827,
+      "step": 9002
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 3.240497865555727e-05,
+      "loss": 0.7431,
+      "step": 9003
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 3.240240703595124e-05,
+      "loss": 0.5848,
+      "step": 9004
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 3.239983541634521e-05,
+      "loss": 0.9422,
+      "step": 9005
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 3.239726379673919e-05,
+      "loss": 0.7565,
+      "step": 9006
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 3.2394692177133165e-05,
+      "loss": 0.3738,
+      "step": 9007
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 3.2392120557527134e-05,
+      "loss": 0.6887,
+      "step": 9008
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 3.2389548937921104e-05,
+      "loss": 0.8154,
+      "step": 9009
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 3.2386977318315074e-05,
+      "loss": 0.8434,
+      "step": 9010
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 3.238440569870905e-05,
+      "loss": 0.8745,
+      "step": 9011
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 3.238183407910302e-05,
+      "loss": 0.9513,
+      "step": 9012
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 3.2379262459496996e-05,
+      "loss": 0.7088,
+      "step": 9013
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 3.2376690839890966e-05,
+      "loss": 0.7082,
+      "step": 9014
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 3.2374119220284936e-05,
+      "loss": 0.7142,
+      "step": 9015
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 3.237154760067891e-05,
+      "loss": 0.7743,
+      "step": 9016
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 3.236897598107288e-05,
+      "loss": 0.4546,
+      "step": 9017
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 3.236640436146685e-05,
+      "loss": 0.5471,
+      "step": 9018
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 3.236383274186083e-05,
+      "loss": 0.5734,
+      "step": 9019
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 3.23612611222548e-05,
+      "loss": 0.9487,
+      "step": 9020
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 3.235868950264877e-05,
+      "loss": 0.9546,
+      "step": 9021
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 3.2356117883042744e-05,
+      "loss": 0.7278,
+      "step": 9022
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 3.235354626343671e-05,
+      "loss": 0.7793,
+      "step": 9023
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 3.235097464383068e-05,
+      "loss": 0.7799,
+      "step": 9024
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 3.234840302422466e-05,
+      "loss": 0.7876,
+      "step": 9025
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 3.234583140461863e-05,
+      "loss": 0.7001,
+      "step": 9026
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 3.2343259785012606e-05,
+      "loss": 0.7871,
+      "step": 9027
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 3.2340688165406575e-05,
+      "loss": 0.7272,
+      "step": 9028
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 3.2338116545800545e-05,
+      "loss": 0.7018,
+      "step": 9029
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 3.2335544926194515e-05,
+      "loss": 0.8523,
+      "step": 9030
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 3.233297330658849e-05,
+      "loss": 0.5493,
+      "step": 9031
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 3.233040168698247e-05,
+      "loss": 0.9243,
+      "step": 9032
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 3.232783006737644e-05,
+      "loss": 0.8783,
+      "step": 9033
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 3.232525844777041e-05,
+      "loss": 0.8767,
+      "step": 9034
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 3.232268682816438e-05,
+      "loss": 0.5658,
+      "step": 9035
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 3.2320115208558346e-05,
+      "loss": 0.8912,
+      "step": 9036
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 3.231754358895232e-05,
+      "loss": 0.8301,
+      "step": 9037
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 3.23149719693463e-05,
+      "loss": 0.4467,
+      "step": 9038
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 3.231240034974027e-05,
+      "loss": 0.8217,
+      "step": 9039
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 3.230982873013424e-05,
+      "loss": 0.9946,
+      "step": 9040
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 3.230725711052821e-05,
+      "loss": 0.9273,
+      "step": 9041
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 3.2304685490922185e-05,
+      "loss": 0.814,
+      "step": 9042
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 3.230211387131616e-05,
+      "loss": 0.8342,
+      "step": 9043
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 3.229954225171013e-05,
+      "loss": 0.7979,
+      "step": 9044
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 3.22969706321041e-05,
+      "loss": 0.9405,
+      "step": 9045
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 3.229439901249807e-05,
+      "loss": 0.7569,
+      "step": 9046
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 3.229182739289204e-05,
+      "loss": 0.8827,
+      "step": 9047
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.2289255773286016e-05,
+      "loss": 0.9958,
+      "step": 9048
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.228668415367999e-05,
+      "loss": 0.8078,
+      "step": 9049
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.228411253407396e-05,
+      "loss": 0.718,
+      "step": 9050
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.228154091446793e-05,
+      "loss": 0.8029,
+      "step": 9051
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.22789692948619e-05,
+      "loss": 0.6866,
+      "step": 9052
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.227639767525588e-05,
+      "loss": 0.9603,
+      "step": 9053
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.227382605564985e-05,
+      "loss": 0.9666,
+      "step": 9054
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.2271254436043825e-05,
+      "loss": 1.019,
+      "step": 9055
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.2268682816437794e-05,
+      "loss": 0.859,
+      "step": 9056
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.2266111196831764e-05,
+      "loss": 0.8243,
+      "step": 9057
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.226353957722574e-05,
+      "loss": 0.867,
+      "step": 9058
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.226096795761971e-05,
+      "loss": 0.8031,
+      "step": 9059
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.225839633801368e-05,
+      "loss": 0.8484,
+      "step": 9060
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.2255824718407656e-05,
+      "loss": 0.6616,
+      "step": 9061
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.2253253098801626e-05,
+      "loss": 0.7283,
+      "step": 9062
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.22506814791956e-05,
+      "loss": 0.5561,
+      "step": 9063
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.224810985958957e-05,
+      "loss": 0.7957,
+      "step": 9064
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.224553823998354e-05,
+      "loss": 1.1625,
+      "step": 9065
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.224296662037751e-05,
+      "loss": 0.6947,
+      "step": 9066
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.224039500077149e-05,
+      "loss": 0.7117,
+      "step": 9067
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.223782338116546e-05,
+      "loss": 0.6777,
+      "step": 9068
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.2235251761559434e-05,
+      "loss": 0.7676,
+      "step": 9069
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.2232680141953404e-05,
+      "loss": 0.7711,
+      "step": 9070
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.2230108522347373e-05,
+      "loss": 0.8963,
+      "step": 9071
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.222753690274134e-05,
+      "loss": 0.7914,
+      "step": 9072
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.222496528313532e-05,
+      "loss": 0.7441,
+      "step": 9073
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.2222393663529296e-05,
+      "loss": 0.8372,
+      "step": 9074
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.2219822043923266e-05,
+      "loss": 1.0129,
+      "step": 9075
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.2217250424317235e-05,
+      "loss": 1.008,
+      "step": 9076
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.2214678804711205e-05,
+      "loss": 0.8005,
+      "step": 9077
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.2212107185105175e-05,
+      "loss": 0.6842,
+      "step": 9078
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.220953556549916e-05,
+      "loss": 1.0006,
+      "step": 9079
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.220696394589313e-05,
+      "loss": 0.7393,
+      "step": 9080
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.22043923262871e-05,
+      "loss": 0.8789,
+      "step": 9081
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.220182070668107e-05,
+      "loss": 0.6139,
+      "step": 9082
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.219924908707504e-05,
+      "loss": 0.6046,
+      "step": 9083
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.219667746746901e-05,
+      "loss": 0.8886,
+      "step": 9084
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.219410584786299e-05,
+      "loss": 0.4286,
+      "step": 9085
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.219153422825696e-05,
+      "loss": 1.0383,
+      "step": 9086
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.218896260865093e-05,
+      "loss": 0.5396,
+      "step": 9087
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.21863909890449e-05,
+      "loss": 0.9322,
+      "step": 9088
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.2183819369438875e-05,
+      "loss": 0.7376,
+      "step": 9089
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.2181247749832845e-05,
+      "loss": 0.7673,
+      "step": 9090
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.217867613022682e-05,
+      "loss": 0.5597,
+      "step": 9091
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.217610451062079e-05,
+      "loss": 0.7972,
+      "step": 9092
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.217353289101476e-05,
+      "loss": 0.6379,
+      "step": 9093
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.217096127140874e-05,
+      "loss": 0.9134,
+      "step": 9094
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.216838965180271e-05,
+      "loss": 0.7221,
+      "step": 9095
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.216581803219668e-05,
+      "loss": 0.7661,
+      "step": 9096
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.216324641259065e-05,
+      "loss": 0.8119,
+      "step": 9097
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.216067479298462e-05,
+      "loss": 0.5014,
+      "step": 9098
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.215810317337859e-05,
+      "loss": 0.6852,
+      "step": 9099
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.215553155377257e-05,
+      "loss": 0.6482,
+      "step": 9100
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.215295993416654e-05,
+      "loss": 0.5664,
+      "step": 9101
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.2150388314560515e-05,
+      "loss": 0.6762,
+      "step": 9102
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.2147816694954485e-05,
+      "loss": 0.8514,
+      "step": 9103
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.2145245075348454e-05,
+      "loss": 0.8763,
+      "step": 9104
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.214267345574243e-05,
+      "loss": 0.9493,
+      "step": 9105
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.21401018361364e-05,
+      "loss": 0.5026,
+      "step": 9106
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.213753021653037e-05,
+      "loss": 0.7283,
+      "step": 9107
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.213495859692435e-05,
+      "loss": 0.9918,
+      "step": 9108
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.2132386977318316e-05,
+      "loss": 0.6432,
+      "step": 9109
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.212981535771229e-05,
+      "loss": 0.6693,
+      "step": 9110
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.212724373810626e-05,
+      "loss": 0.7157,
+      "step": 9111
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.212467211850023e-05,
+      "loss": 0.7354,
+      "step": 9112
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.21221004988942e-05,
+      "loss": 1.0325,
+      "step": 9113
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.211952887928818e-05,
+      "loss": 0.7484,
+      "step": 9114
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.211695725968215e-05,
+      "loss": 0.7027,
+      "step": 9115
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.2114385640076124e-05,
+      "loss": 0.7401,
+      "step": 9116
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.2111814020470094e-05,
+      "loss": 0.9421,
+      "step": 9117
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.2109242400864064e-05,
+      "loss": 0.7869,
+      "step": 9118
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.2106670781258034e-05,
+      "loss": 0.7266,
+      "step": 9119
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.210409916165201e-05,
+      "loss": 1.0335,
+      "step": 9120
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.2101527542045986e-05,
+      "loss": 0.8478,
+      "step": 9121
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.2098955922439956e-05,
+      "loss": 0.6988,
+      "step": 9122
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.2096384302833926e-05,
+      "loss": 0.8332,
+      "step": 9123
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.2093812683227896e-05,
+      "loss": 0.7048,
+      "step": 9124
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.2091241063621865e-05,
+      "loss": 0.6411,
+      "step": 9125
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.208866944401585e-05,
+      "loss": 0.8427,
+      "step": 9126
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.208609782440982e-05,
+      "loss": 0.78,
+      "step": 9127
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.208352620480379e-05,
+      "loss": 1.0113,
+      "step": 9128
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.208095458519776e-05,
+      "loss": 0.7747,
+      "step": 9129
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.207838296559173e-05,
+      "loss": 0.6319,
+      "step": 9130
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.2075811345985704e-05,
+      "loss": 0.8228,
+      "step": 9131
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.207323972637968e-05,
+      "loss": 0.8409,
+      "step": 9132
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.207066810677365e-05,
+      "loss": 0.7919,
+      "step": 9133
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.206809648716762e-05,
+      "loss": 0.8192,
+      "step": 9134
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.206552486756159e-05,
+      "loss": 0.767,
+      "step": 9135
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.2062953247955566e-05,
+      "loss": 0.8857,
+      "step": 9136
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.2060381628349535e-05,
+      "loss": 0.9638,
+      "step": 9137
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.205781000874351e-05,
+      "loss": 0.8195,
+      "step": 9138
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.205523838913748e-05,
+      "loss": 0.9266,
+      "step": 9139
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.205266676953145e-05,
+      "loss": 0.8663,
+      "step": 9140
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.205009514992543e-05,
+      "loss": 0.6832,
+      "step": 9141
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.20475235303194e-05,
+      "loss": 0.8168,
+      "step": 9142
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.204495191071337e-05,
+      "loss": 0.8106,
+      "step": 9143
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.2042380291107343e-05,
+      "loss": 0.9119,
+      "step": 9144
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.203980867150131e-05,
+      "loss": 0.7814,
+      "step": 9145
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.203723705189528e-05,
+      "loss": 0.897,
+      "step": 9146
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.203466543228926e-05,
+      "loss": 0.6685,
+      "step": 9147
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.203209381268323e-05,
+      "loss": 0.9316,
+      "step": 9148
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.20295221930772e-05,
+      "loss": 0.544,
+      "step": 9149
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.2026950573471175e-05,
+      "loss": 0.8671,
+      "step": 9150
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.2024378953865145e-05,
+      "loss": 0.6588,
+      "step": 9151
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.202180733425912e-05,
+      "loss": 0.6315,
+      "step": 9152
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.201923571465309e-05,
+      "loss": 0.6915,
+      "step": 9153
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.201666409504706e-05,
+      "loss": 0.7227,
+      "step": 9154
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.201409247544103e-05,
+      "loss": 0.6935,
+      "step": 9155
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 3.201152085583501e-05,
+      "loss": 0.5762,
+      "step": 9156
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 3.200894923622898e-05,
+      "loss": 0.7541,
+      "step": 9157
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 3.200637761662295e-05,
+      "loss": 0.9998,
+      "step": 9158
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 3.200380599701692e-05,
+      "loss": 0.9416,
+      "step": 9159
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 3.200123437741089e-05,
+      "loss": 0.7289,
+      "step": 9160
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 3.199866275780486e-05,
+      "loss": 0.9198,
+      "step": 9161
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 3.199609113819884e-05,
+      "loss": 0.8447,
+      "step": 9162
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 3.1993519518592815e-05,
+      "loss": 0.7412,
+      "step": 9163
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 3.1990947898986785e-05,
+      "loss": 0.6857,
+      "step": 9164
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 3.1988376279380754e-05,
+      "loss": 1.0207,
+      "step": 9165
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 3.1985804659774724e-05,
+      "loss": 0.7698,
+      "step": 9166
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 3.19832330401687e-05,
+      "loss": 0.5543,
+      "step": 9167
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 3.198066142056268e-05,
+      "loss": 0.7834,
+      "step": 9168
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 3.1978089800956647e-05,
+      "loss": 0.6942,
+      "step": 9169
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 3.1975518181350616e-05,
+      "loss": 0.5812,
+      "step": 9170
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 3.1972946561744586e-05,
+      "loss": 0.3976,
+      "step": 9171
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 3.1970374942138556e-05,
+      "loss": 0.742,
+      "step": 9172
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 3.196780332253253e-05,
+      "loss": 0.6772,
+      "step": 9173
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 3.196523170292651e-05,
+      "loss": 0.9547,
+      "step": 9174
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 3.196266008332048e-05,
+      "loss": 0.8441,
+      "step": 9175
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 3.196008846371445e-05,
+      "loss": 0.9939,
+      "step": 9176
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 3.195751684410842e-05,
+      "loss": 0.5974,
+      "step": 9177
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 3.1954945224502394e-05,
+      "loss": 0.8551,
+      "step": 9178
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 3.1952373604896364e-05,
+      "loss": 0.6118,
+      "step": 9179
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 3.194980198529034e-05,
+      "loss": 1.0363,
+      "step": 9180
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 3.194723036568431e-05,
+      "loss": 0.6874,
+      "step": 9181
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 3.194465874607828e-05,
+      "loss": 0.6753,
+      "step": 9182
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 3.1942087126472256e-05,
+      "loss": 0.6378,
+      "step": 9183
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 3.1939515506866226e-05,
+      "loss": 1.1078,
+      "step": 9184
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 3.1936943887260195e-05,
+      "loss": 1.5136,
+      "step": 9185
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 3.193437226765417e-05,
+      "loss": 1.3213,
+      "step": 9186
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 3.193180064804814e-05,
+      "loss": 1.0159,
+      "step": 9187
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 3.192922902844212e-05,
+      "loss": 0.9786,
+      "step": 9188
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 3.192665740883609e-05,
+      "loss": 0.5833,
+      "step": 9189
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 3.192408578923006e-05,
+      "loss": 1.3369,
+      "step": 9190
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 3.192151416962403e-05,
+      "loss": 0.8345,
+      "step": 9191
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 3.1918942550018004e-05,
+      "loss": 0.8689,
+      "step": 9192
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 3.191637093041197e-05,
+      "loss": 0.678,
+      "step": 9193
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 3.191379931080595e-05,
+      "loss": 0.8571,
+      "step": 9194
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 3.191122769119992e-05,
+      "loss": 0.9937,
+      "step": 9195
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 3.190865607159389e-05,
+      "loss": 0.8545,
+      "step": 9196
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 3.190608445198786e-05,
+      "loss": 0.9545,
+      "step": 9197
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 3.1903512832381835e-05,
+      "loss": 0.9672,
+      "step": 9198
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 3.190094121277581e-05,
+      "loss": 0.9153,
+      "step": 9199
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 3.189836959316978e-05,
+      "loss": 0.5353,
+      "step": 9200
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 3.189579797356375e-05,
+      "loss": 0.8586,
+      "step": 9201
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 3.189322635395772e-05,
+      "loss": 0.7387,
+      "step": 9202
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 3.189065473435169e-05,
+      "loss": 0.9428,
+      "step": 9203
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 3.1888083114745674e-05,
+      "loss": 0.7595,
+      "step": 9204
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 3.188551149513964e-05,
+      "loss": 0.8517,
+      "step": 9205
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 3.188293987553361e-05,
+      "loss": 0.7776,
+      "step": 9206
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 3.188036825592758e-05,
+      "loss": 0.7817,
+      "step": 9207
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 3.187779663632155e-05,
+      "loss": 0.7618,
+      "step": 9208
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 3.187522501671553e-05,
+      "loss": 0.5382,
+      "step": 9209
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 3.1872653397109505e-05,
+      "loss": 0.7049,
+      "step": 9210
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 3.1870081777503475e-05,
+      "loss": 0.7509,
+      "step": 9211
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 3.1867510157897445e-05,
+      "loss": 0.5845,
+      "step": 9212
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 3.1864938538291414e-05,
+      "loss": 0.7765,
+      "step": 9213
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 3.186236691868539e-05,
+      "loss": 0.7343,
+      "step": 9214
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 3.185979529907936e-05,
+      "loss": 1.0231,
+      "step": 9215
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 3.185722367947334e-05,
+      "loss": 0.8389,
+      "step": 9216
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 3.185465205986731e-05,
+      "loss": 1.0632,
+      "step": 9217
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 3.1852080440261276e-05,
+      "loss": 0.7509,
+      "step": 9218
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 3.184950882065525e-05,
+      "loss": 0.6893,
+      "step": 9219
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 3.184693720104922e-05,
+      "loss": 0.9373,
+      "step": 9220
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 3.184436558144319e-05,
+      "loss": 0.6921,
+      "step": 9221
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 3.184179396183717e-05,
+      "loss": 0.6209,
+      "step": 9222
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 3.183922234223114e-05,
+      "loss": 0.6107,
+      "step": 9223
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 3.183665072262511e-05,
+      "loss": 0.6749,
+      "step": 9224
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 3.1834079103019084e-05,
+      "loss": 0.6164,
+      "step": 9225
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 3.1831507483413054e-05,
+      "loss": 0.8872,
+      "step": 9226
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 3.1828935863807024e-05,
+      "loss": 0.8918,
+      "step": 9227
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 3.1826364244201e-05,
+      "loss": 1.1128,
+      "step": 9228
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 3.182379262459497e-05,
+      "loss": 0.7414,
+      "step": 9229
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 3.1821221004988946e-05,
+      "loss": 0.6794,
+      "step": 9230
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 3.1818649385382916e-05,
+      "loss": 0.6031,
+      "step": 9231
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 3.1816077765776886e-05,
+      "loss": 0.7231,
+      "step": 9232
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 3.1813506146170855e-05,
+      "loss": 1.0856,
+      "step": 9233
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 3.181093452656483e-05,
+      "loss": 0.8864,
+      "step": 9234
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 3.180836290695881e-05,
+      "loss": 1.0533,
+      "step": 9235
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 3.180579128735278e-05,
+      "loss": 1.038,
+      "step": 9236
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 3.180321966774675e-05,
+      "loss": 0.6624,
+      "step": 9237
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 3.180064804814072e-05,
+      "loss": 0.7423,
+      "step": 9238
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 3.179807642853469e-05,
+      "loss": 1.2555,
+      "step": 9239
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 3.1795504808928664e-05,
+      "loss": 0.6702,
+      "step": 9240
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 3.179293318932264e-05,
+      "loss": 1.109,
+      "step": 9241
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 3.179036156971661e-05,
+      "loss": 0.7793,
+      "step": 9242
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 3.178778995011058e-05,
+      "loss": 0.8,
+      "step": 9243
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 3.178521833050455e-05,
+      "loss": 0.7167,
+      "step": 9244
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 3.1782646710898526e-05,
+      "loss": 0.7715,
+      "step": 9245
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 3.17800750912925e-05,
+      "loss": 0.474,
+      "step": 9246
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 3.177750347168647e-05,
+      "loss": 0.7068,
+      "step": 9247
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 3.177493185208044e-05,
+      "loss": 0.7492,
+      "step": 9248
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 3.177236023247441e-05,
+      "loss": 0.5539,
+      "step": 9249
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 3.176978861286838e-05,
+      "loss": 1.1731,
+      "step": 9250
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 3.1767216993262364e-05,
+      "loss": 0.9635,
+      "step": 9251
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 3.1764645373656334e-05,
+      "loss": 0.7581,
+      "step": 9252
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 3.1762073754050303e-05,
+      "loss": 0.6512,
+      "step": 9253
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 3.175950213444427e-05,
+      "loss": 0.7889,
+      "step": 9254
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 3.175693051483824e-05,
+      "loss": 0.9675,
+      "step": 9255
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 3.175435889523222e-05,
+      "loss": 0.8168,
+      "step": 9256
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 3.1751787275626196e-05,
+      "loss": 0.6932,
+      "step": 9257
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 3.1749215656020165e-05,
+      "loss": 0.8332,
+      "step": 9258
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 3.1746644036414135e-05,
+      "loss": 0.894,
+      "step": 9259
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 3.1744072416808105e-05,
+      "loss": 0.892,
+      "step": 9260
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 3.174150079720208e-05,
+      "loss": 0.9759,
+      "step": 9261
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 3.173892917759605e-05,
+      "loss": 1.018,
+      "step": 9262
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 3.173635755799003e-05,
+      "loss": 0.7867,
+      "step": 9263
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 3.1733785938384e-05,
+      "loss": 0.6232,
+      "step": 9264
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 3.173121431877797e-05,
+      "loss": 0.6823,
+      "step": 9265
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 3.172864269917194e-05,
+      "loss": 1.0876,
+      "step": 9266
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 3.172607107956591e-05,
+      "loss": 0.83,
+      "step": 9267
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 3.172349945995988e-05,
+      "loss": 0.8115,
+      "step": 9268
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 3.172092784035386e-05,
+      "loss": 0.6657,
+      "step": 9269
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 3.171835622074783e-05,
+      "loss": 0.7298,
+      "step": 9270
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 3.17157846011418e-05,
+      "loss": 1.1408,
+      "step": 9271
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 3.1713212981535775e-05,
+      "loss": 0.8271,
+      "step": 9272
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 3.1710641361929745e-05,
+      "loss": 0.9624,
+      "step": 9273
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 3.1708069742323714e-05,
+      "loss": 0.8769,
+      "step": 9274
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 3.170549812271769e-05,
+      "loss": 0.7117,
+      "step": 9275
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 3.170292650311166e-05,
+      "loss": 0.7934,
+      "step": 9276
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 3.170035488350564e-05,
+      "loss": 0.9272,
+      "step": 9277
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 3.1697783263899607e-05,
+      "loss": 0.6998,
+      "step": 9278
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 3.1695211644293576e-05,
+      "loss": 0.9294,
+      "step": 9279
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 3.1692640024687546e-05,
+      "loss": 0.6136,
+      "step": 9280
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 3.169006840508152e-05,
+      "loss": 0.6881,
+      "step": 9281
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 3.16874967854755e-05,
+      "loss": 0.9631,
+      "step": 9282
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 3.168492516586947e-05,
+      "loss": 0.6051,
+      "step": 9283
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 3.168235354626344e-05,
+      "loss": 0.6662,
+      "step": 9284
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 3.167978192665741e-05,
+      "loss": 0.6968,
+      "step": 9285
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 3.167721030705138e-05,
+      "loss": 1.2352,
+      "step": 9286
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 3.1674638687445354e-05,
+      "loss": 0.9716,
+      "step": 9287
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 3.167206706783933e-05,
+      "loss": 0.7184,
+      "step": 9288
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 3.16694954482333e-05,
+      "loss": 0.8142,
+      "step": 9289
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 3.166692382862727e-05,
+      "loss": 0.8304,
+      "step": 9290
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 3.166435220902124e-05,
+      "loss": 0.6582,
+      "step": 9291
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 3.1661780589415216e-05,
+      "loss": 1.0176,
+      "step": 9292
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 3.165920896980919e-05,
+      "loss": 0.8457,
+      "step": 9293
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 3.165663735020316e-05,
+      "loss": 0.7383,
+      "step": 9294
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 3.165406573059713e-05,
+      "loss": 0.8187,
+      "step": 9295
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 3.16514941109911e-05,
+      "loss": 0.8278,
+      "step": 9296
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 3.164892249138507e-05,
+      "loss": 0.8708,
+      "step": 9297
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 3.164635087177905e-05,
+      "loss": 0.5527,
+      "step": 9298
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 3.1643779252173024e-05,
+      "loss": 1.035,
+      "step": 9299
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 3.1641207632566994e-05,
+      "loss": 0.776,
+      "step": 9300
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 3.1638636012960964e-05,
+      "loss": 0.8939,
+      "step": 9301
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 3.163606439335493e-05,
+      "loss": 0.7492,
+      "step": 9302
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 3.163349277374891e-05,
+      "loss": 0.8441,
+      "step": 9303
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 3.163092115414288e-05,
+      "loss": 1.1504,
+      "step": 9304
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 3.1628349534536856e-05,
+      "loss": 0.9811,
+      "step": 9305
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 3.1625777914930825e-05,
+      "loss": 1.0336,
+      "step": 9306
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 3.1623206295324795e-05,
+      "loss": 0.8248,
+      "step": 9307
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 3.162063467571877e-05,
+      "loss": 0.6327,
+      "step": 9308
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 3.161806305611274e-05,
+      "loss": 0.6674,
+      "step": 9309
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 3.161549143650671e-05,
+      "loss": 0.849,
+      "step": 9310
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 3.161291981690069e-05,
+      "loss": 1.2347,
+      "step": 9311
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 3.161034819729466e-05,
+      "loss": 0.5513,
+      "step": 9312
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 3.1607776577688634e-05,
+      "loss": 0.9358,
+      "step": 9313
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 3.16052049580826e-05,
+      "loss": 0.6257,
+      "step": 9314
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 3.160263333847657e-05,
+      "loss": 0.6343,
+      "step": 9315
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 3.160006171887054e-05,
+      "loss": 0.8652,
+      "step": 9316
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 3.159749009926452e-05,
+      "loss": 0.9061,
+      "step": 9317
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 3.159491847965849e-05,
+      "loss": 0.7612,
+      "step": 9318
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 3.1592346860052465e-05,
+      "loss": 0.7254,
+      "step": 9319
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 3.1589775240446435e-05,
+      "loss": 0.6721,
+      "step": 9320
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 3.1587203620840405e-05,
+      "loss": 0.7745,
+      "step": 9321
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 3.1584632001234374e-05,
+      "loss": 0.6941,
+      "step": 9322
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 3.158206038162835e-05,
+      "loss": 0.8081,
+      "step": 9323
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 3.157948876202233e-05,
+      "loss": 0.6887,
+      "step": 9324
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 3.15769171424163e-05,
+      "loss": 1.161,
+      "step": 9325
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 3.157434552281027e-05,
+      "loss": 0.7245,
+      "step": 9326
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 3.1571773903204236e-05,
+      "loss": 0.8635,
+      "step": 9327
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 3.1569202283598206e-05,
+      "loss": 0.7343,
+      "step": 9328
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 3.156663066399219e-05,
+      "loss": 0.8782,
+      "step": 9329
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 3.156405904438616e-05,
+      "loss": 0.6418,
+      "step": 9330
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 3.156148742478013e-05,
+      "loss": 0.5948,
+      "step": 9331
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 3.15589158051741e-05,
+      "loss": 1.1777,
+      "step": 9332
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 3.155634418556807e-05,
+      "loss": 0.7934,
+      "step": 9333
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 3.1553772565962044e-05,
+      "loss": 0.9504,
+      "step": 9334
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 3.155120094635602e-05,
+      "loss": 1.1276,
+      "step": 9335
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 3.154862932674999e-05,
+      "loss": 0.7082,
+      "step": 9336
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 3.154605770714396e-05,
+      "loss": 0.8591,
+      "step": 9337
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 3.154348608753793e-05,
+      "loss": 0.8217,
+      "step": 9338
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 3.1540914467931906e-05,
+      "loss": 0.7342,
+      "step": 9339
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 3.1538342848325876e-05,
+      "loss": 0.8764,
+      "step": 9340
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 3.153577122871985e-05,
+      "loss": 0.6303,
+      "step": 9341
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 3.153319960911382e-05,
+      "loss": 0.7336,
+      "step": 9342
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 3.153062798950779e-05,
+      "loss": 0.8601,
+      "step": 9343
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 3.152805636990177e-05,
+      "loss": 0.8351,
+      "step": 9344
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 3.152548475029574e-05,
+      "loss": 0.6552,
+      "step": 9345
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 3.152291313068971e-05,
+      "loss": 1.0607,
+      "step": 9346
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 3.1520341511083684e-05,
+      "loss": 0.9593,
+      "step": 9347
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 3.1517769891477654e-05,
+      "loss": 1.1586,
+      "step": 9348
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 3.1515198271871624e-05,
+      "loss": 0.5938,
+      "step": 9349
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 3.15126266522656e-05,
+      "loss": 0.8094,
+      "step": 9350
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 3.151005503265957e-05,
+      "loss": 1.154,
+      "step": 9351
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 3.150748341305354e-05,
+      "loss": 0.7391,
+      "step": 9352
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 3.1504911793447516e-05,
+      "loss": 0.7669,
+      "step": 9353
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 3.1502340173841486e-05,
+      "loss": 0.8831,
+      "step": 9354
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 3.149976855423546e-05,
+      "loss": 0.6972,
+      "step": 9355
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 3.149719693462943e-05,
+      "loss": 0.7388,
+      "step": 9356
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 3.14946253150234e-05,
+      "loss": 0.8147,
+      "step": 9357
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 3.149205369541737e-05,
+      "loss": 0.4232,
+      "step": 9358
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 3.148948207581135e-05,
+      "loss": 0.9181,
+      "step": 9359
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 3.1486910456205324e-05,
+      "loss": 0.7481,
+      "step": 9360
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 3.1484338836599294e-05,
+      "loss": 0.8621,
+      "step": 9361
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 3.148176721699326e-05,
+      "loss": 0.7168,
+      "step": 9362
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 3.147919559738723e-05,
+      "loss": 0.6621,
+      "step": 9363
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 3.14766239777812e-05,
+      "loss": 0.6316,
+      "step": 9364
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 3.147405235817518e-05,
+      "loss": 0.7193,
+      "step": 9365
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 3.1471480738569156e-05,
+      "loss": 1.0185,
+      "step": 9366
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 3.1468909118963125e-05,
+      "loss": 0.9868,
+      "step": 9367
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 3.1466337499357095e-05,
+      "loss": 0.6116,
+      "step": 9368
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 3.1463765879751065e-05,
+      "loss": 1.0735,
+      "step": 9369
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 3.146119426014504e-05,
+      "loss": 0.7089,
+      "step": 9370
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 3.145862264053902e-05,
+      "loss": 0.7246,
+      "step": 9371
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 3.145605102093299e-05,
+      "loss": 0.5793,
+      "step": 9372
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 3.145347940132696e-05,
+      "loss": 0.8583,
+      "step": 9373
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 3.145090778172093e-05,
+      "loss": 0.4691,
+      "step": 9374
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 3.1448336162114896e-05,
+      "loss": 0.6723,
+      "step": 9375
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 3.144576454250887e-05,
+      "loss": 0.6088,
+      "step": 9376
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 3.144319292290285e-05,
+      "loss": 0.5316,
+      "step": 9377
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 3.144062130329682e-05,
+      "loss": 0.4921,
+      "step": 9378
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 3.143804968369079e-05,
+      "loss": 0.5206,
+      "step": 9379
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 3.143547806408476e-05,
+      "loss": 0.6132,
+      "step": 9380
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 3.1432906444478735e-05,
+      "loss": 0.765,
+      "step": 9381
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 3.1430334824872705e-05,
+      "loss": 0.6528,
+      "step": 9382
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 3.142776320526668e-05,
+      "loss": 0.949,
+      "step": 9383
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 3.142519158566065e-05,
+      "loss": 0.7902,
+      "step": 9384
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 3.142261996605462e-05,
+      "loss": 0.8622,
+      "step": 9385
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 3.14200483464486e-05,
+      "loss": 0.5744,
+      "step": 9386
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 3.1417476726842567e-05,
+      "loss": 0.9573,
+      "step": 9387
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 3.1414905107236536e-05,
+      "loss": 0.8316,
+      "step": 9388
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 3.141233348763051e-05,
+      "loss": 0.7196,
+      "step": 9389
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 3.140976186802448e-05,
+      "loss": 0.7834,
+      "step": 9390
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 3.140719024841846e-05,
+      "loss": 0.8986,
+      "step": 9391
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 3.140461862881243e-05,
+      "loss": 0.9063,
+      "step": 9392
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 3.14020470092064e-05,
+      "loss": 0.8074,
+      "step": 9393
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 3.139947538960037e-05,
+      "loss": 0.7723,
+      "step": 9394
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 3.1396903769994344e-05,
+      "loss": 0.8227,
+      "step": 9395
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 3.1394332150388314e-05,
+      "loss": 0.7883,
+      "step": 9396
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 3.139176053078229e-05,
+      "loss": 0.6427,
+      "step": 9397
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 3.138918891117626e-05,
+      "loss": 0.7227,
+      "step": 9398
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 3.138661729157023e-05,
+      "loss": 1.0485,
+      "step": 9399
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 3.1384045671964206e-05,
+      "loss": 1.0097,
+      "step": 9400
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 3.1381474052358176e-05,
+      "loss": 1.0266,
+      "step": 9401
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 3.137890243275215e-05,
+      "loss": 0.8387,
+      "step": 9402
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 3.137633081314612e-05,
+      "loss": 1.2551,
+      "step": 9403
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 3.137375919354009e-05,
+      "loss": 0.6079,
+      "step": 9404
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 3.137118757393406e-05,
+      "loss": 0.5571,
+      "step": 9405
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 3.136861595432804e-05,
+      "loss": 0.6874,
+      "step": 9406
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 3.1366044334722014e-05,
+      "loss": 0.6973,
+      "step": 9407
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 3.1363472715115984e-05,
+      "loss": 0.9108,
+      "step": 9408
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 3.1360901095509954e-05,
+      "loss": 0.9364,
+      "step": 9409
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 3.1358329475903923e-05,
+      "loss": 0.718,
+      "step": 9410
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 3.135575785629789e-05,
+      "loss": 0.6976,
+      "step": 9411
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 3.135318623669187e-05,
+      "loss": 0.6798,
+      "step": 9412
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 3.1350614617085846e-05,
+      "loss": 0.9454,
+      "step": 9413
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 3.1348042997479816e-05,
+      "loss": 0.6973,
+      "step": 9414
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 3.1345471377873785e-05,
+      "loss": 0.6751,
+      "step": 9415
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 3.1342899758267755e-05,
+      "loss": 0.8905,
+      "step": 9416
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 3.134032813866173e-05,
+      "loss": 0.5882,
+      "step": 9417
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 3.133775651905571e-05,
+      "loss": 0.5495,
+      "step": 9418
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 3.133518489944968e-05,
+      "loss": 0.5518,
+      "step": 9419
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 3.133261327984365e-05,
+      "loss": 0.6111,
+      "step": 9420
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 3.133004166023762e-05,
+      "loss": 0.7174,
+      "step": 9421
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 3.132747004063159e-05,
+      "loss": 0.8869,
+      "step": 9422
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 3.132489842102556e-05,
+      "loss": 0.8219,
+      "step": 9423
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 3.132232680141954e-05,
+      "loss": 0.5311,
+      "step": 9424
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 3.131975518181351e-05,
+      "loss": 0.8484,
+      "step": 9425
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 3.131718356220748e-05,
+      "loss": 1.1063,
+      "step": 9426
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 3.131461194260145e-05,
+      "loss": 0.584,
+      "step": 9427
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 3.1312040322995425e-05,
+      "loss": 0.618,
+      "step": 9428
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 3.1309468703389395e-05,
+      "loss": 0.5383,
+      "step": 9429
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 3.130689708378337e-05,
+      "loss": 0.7818,
+      "step": 9430
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 3.130432546417734e-05,
+      "loss": 0.896,
+      "step": 9431
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 3.130175384457131e-05,
+      "loss": 0.5295,
+      "step": 9432
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 3.129918222496529e-05,
+      "loss": 0.7616,
+      "step": 9433
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 3.129661060535926e-05,
+      "loss": 0.9042,
+      "step": 9434
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 3.1294038985753227e-05,
+      "loss": 0.7691,
+      "step": 9435
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 3.12914673661472e-05,
+      "loss": 0.7578,
+      "step": 9436
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 3.128889574654117e-05,
+      "loss": 0.7662,
+      "step": 9437
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 3.128632412693515e-05,
+      "loss": 1.0533,
+      "step": 9438
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 3.128375250732912e-05,
+      "loss": 0.8097,
+      "step": 9439
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 3.128118088772309e-05,
+      "loss": 0.6723,
+      "step": 9440
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 3.127860926811706e-05,
+      "loss": 1.0066,
+      "step": 9441
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 3.1276037648511035e-05,
+      "loss": 0.7816,
+      "step": 9442
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 3.1273466028905004e-05,
+      "loss": 0.5783,
+      "step": 9443
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 3.127089440929898e-05,
+      "loss": 0.7774,
+      "step": 9444
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 3.126832278969295e-05,
+      "loss": 0.7888,
+      "step": 9445
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 3.126575117008692e-05,
+      "loss": 0.7449,
+      "step": 9446
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 3.126317955048089e-05,
+      "loss": 0.7731,
+      "step": 9447
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 3.1260607930874866e-05,
+      "loss": 0.918,
+      "step": 9448
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 3.125803631126884e-05,
+      "loss": 0.6174,
+      "step": 9449
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 3.125546469166281e-05,
+      "loss": 0.9391,
+      "step": 9450
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 3.125289307205678e-05,
+      "loss": 1.0944,
+      "step": 9451
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 3.125032145245075e-05,
+      "loss": 0.7624,
+      "step": 9452
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 3.124774983284472e-05,
+      "loss": 0.9117,
+      "step": 9453
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 3.1245178213238705e-05,
+      "loss": 0.7558,
+      "step": 9454
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 3.1242606593632675e-05,
+      "loss": 1.026,
+      "step": 9455
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 3.1240034974026644e-05,
+      "loss": 0.9602,
+      "step": 9456
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 3.1237463354420614e-05,
+      "loss": 0.9983,
+      "step": 9457
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 3.1234891734814584e-05,
+      "loss": 0.7651,
+      "step": 9458
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 3.123232011520856e-05,
+      "loss": 0.7098,
+      "step": 9459
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 3.1229748495602536e-05,
+      "loss": 0.8152,
+      "step": 9460
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 3.1227176875996506e-05,
+      "loss": 0.8702,
+      "step": 9461
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 3.1224605256390476e-05,
+      "loss": 0.7846,
+      "step": 9462
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 3.1222033636784446e-05,
+      "loss": 0.7678,
+      "step": 9463
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 3.121946201717842e-05,
+      "loss": 0.5537,
+      "step": 9464
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 3.121689039757239e-05,
+      "loss": 1.0293,
+      "step": 9465
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 3.121431877796637e-05,
+      "loss": 0.9229,
+      "step": 9466
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 3.121174715836034e-05,
+      "loss": 0.7465,
+      "step": 9467
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 3.120917553875431e-05,
+      "loss": 0.9761,
+      "step": 9468
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 3.1206603919148284e-05,
+      "loss": 1.1216,
+      "step": 9469
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 3.1204032299542254e-05,
+      "loss": 0.8099,
+      "step": 9470
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 3.120146067993622e-05,
+      "loss": 0.4867,
+      "step": 9471
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 3.11988890603302e-05,
+      "loss": 0.9352,
+      "step": 9472
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 3.119631744072417e-05,
+      "loss": 0.9451,
+      "step": 9473
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 3.119374582111814e-05,
+      "loss": 0.6519,
+      "step": 9474
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 3.1191174201512116e-05,
+      "loss": 0.43,
+      "step": 9475
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 3.1188602581906085e-05,
+      "loss": 0.8618,
+      "step": 9476
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 3.1186030962300055e-05,
+      "loss": 0.6599,
+      "step": 9477
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 3.118345934269403e-05,
+      "loss": 0.6881,
+      "step": 9478
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 3.1180887723088e-05,
+      "loss": 0.827,
+      "step": 9479
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.117831610348198e-05,
+      "loss": 0.7718,
+      "step": 9480
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.117574448387595e-05,
+      "loss": 0.8426,
+      "step": 9481
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.117317286426992e-05,
+      "loss": 1.188,
+      "step": 9482
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.117060124466389e-05,
+      "loss": 0.805,
+      "step": 9483
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.116802962505786e-05,
+      "loss": 0.4488,
+      "step": 9484
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.116545800545184e-05,
+      "loss": 0.4995,
+      "step": 9485
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.116288638584581e-05,
+      "loss": 0.5005,
+      "step": 9486
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.116031476623978e-05,
+      "loss": 0.8497,
+      "step": 9487
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.115774314663375e-05,
+      "loss": 0.3626,
+      "step": 9488
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.115517152702772e-05,
+      "loss": 0.4121,
+      "step": 9489
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.1152599907421695e-05,
+      "loss": 0.6533,
+      "step": 9490
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.115002828781567e-05,
+      "loss": 0.6824,
+      "step": 9491
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.114745666820964e-05,
+      "loss": 0.951,
+      "step": 9492
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.114488504860361e-05,
+      "loss": 1.0529,
+      "step": 9493
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.114231342899758e-05,
+      "loss": 0.9665,
+      "step": 9494
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.113974180939156e-05,
+      "loss": 0.9238,
+      "step": 9495
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.113717018978553e-05,
+      "loss": 0.9001,
+      "step": 9496
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.11345985701795e-05,
+      "loss": 0.7664,
+      "step": 9497
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.113202695057347e-05,
+      "loss": 0.9942,
+      "step": 9498
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.112945533096744e-05,
+      "loss": 1.1604,
+      "step": 9499
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.112688371136141e-05,
+      "loss": 0.4996,
+      "step": 9500
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.112431209175539e-05,
+      "loss": 0.3918,
+      "step": 9501
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.1121740472149365e-05,
+      "loss": 1.0693,
+      "step": 9502
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.1119168852543335e-05,
+      "loss": 0.8893,
+      "step": 9503
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.1116597232937304e-05,
+      "loss": 1.3382,
+      "step": 9504
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.1114025613331274e-05,
+      "loss": 0.8647,
+      "step": 9505
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.111145399372525e-05,
+      "loss": 0.5698,
+      "step": 9506
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.110888237411922e-05,
+      "loss": 0.6829,
+      "step": 9507
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.1106310754513197e-05,
+      "loss": 0.6748,
+      "step": 9508
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.1103739134907166e-05,
+      "loss": 0.8246,
+      "step": 9509
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.1101167515301136e-05,
+      "loss": 0.7593,
+      "step": 9510
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.109859589569511e-05,
+      "loss": 0.6351,
+      "step": 9511
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.109602427608908e-05,
+      "loss": 0.9037,
+      "step": 9512
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.109345265648305e-05,
+      "loss": 1.0128,
+      "step": 9513
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.109088103687703e-05,
+      "loss": 0.6654,
+      "step": 9514
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.1088309417271e-05,
+      "loss": 0.9424,
+      "step": 9515
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.1085737797664974e-05,
+      "loss": 1.0267,
+      "step": 9516
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.1083166178058944e-05,
+      "loss": 0.8438,
+      "step": 9517
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.1080594558452914e-05,
+      "loss": 0.5695,
+      "step": 9518
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.1078022938846883e-05,
+      "loss": 0.5271,
+      "step": 9519
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.107545131924086e-05,
+      "loss": 0.7826,
+      "step": 9520
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.107287969963483e-05,
+      "loss": 0.5595,
+      "step": 9521
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.1070308080028806e-05,
+      "loss": 1.0499,
+      "step": 9522
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.1067736460422776e-05,
+      "loss": 1.0729,
+      "step": 9523
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.1065164840816745e-05,
+      "loss": 0.9335,
+      "step": 9524
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.1062593221210715e-05,
+      "loss": 0.8757,
+      "step": 9525
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.106002160160469e-05,
+      "loss": 0.693,
+      "step": 9526
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.105744998199867e-05,
+      "loss": 0.5061,
+      "step": 9527
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.105487836239264e-05,
+      "loss": 0.7037,
+      "step": 9528
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.105230674278661e-05,
+      "loss": 0.6928,
+      "step": 9529
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.104973512318058e-05,
+      "loss": 0.564,
+      "step": 9530
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.104716350357455e-05,
+      "loss": 1.0275,
+      "step": 9531
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.104459188396853e-05,
+      "loss": 0.8105,
+      "step": 9532
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.10420202643625e-05,
+      "loss": 0.758,
+      "step": 9533
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.103944864475647e-05,
+      "loss": 1.186,
+      "step": 9534
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.103687702515044e-05,
+      "loss": 0.5137,
+      "step": 9535
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.103430540554441e-05,
+      "loss": 0.8218,
+      "step": 9536
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.1031733785938385e-05,
+      "loss": 0.7834,
+      "step": 9537
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.102916216633236e-05,
+      "loss": 0.7975,
+      "step": 9538
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.102659054672633e-05,
+      "loss": 0.42,
+      "step": 9539
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.10240189271203e-05,
+      "loss": 0.9557,
+      "step": 9540
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.102144730751427e-05,
+      "loss": 0.7838,
+      "step": 9541
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.101887568790825e-05,
+      "loss": 0.9867,
+      "step": 9542
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.101630406830222e-05,
+      "loss": 0.6413,
+      "step": 9543
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.101373244869619e-05,
+      "loss": 0.8252,
+      "step": 9544
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.101116082909016e-05,
+      "loss": 1.0104,
+      "step": 9545
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.100858920948413e-05,
+      "loss": 1.0489,
+      "step": 9546
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.100601758987811e-05,
+      "loss": 0.7028,
+      "step": 9547
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.100344597027208e-05,
+      "loss": 0.7371,
+      "step": 9548
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.1000874350666055e-05,
+      "loss": 0.6396,
+      "step": 9549
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.0998302731060025e-05,
+      "loss": 0.7281,
+      "step": 9550
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.0995731111453995e-05,
+      "loss": 0.8946,
+      "step": 9551
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.0993159491847964e-05,
+      "loss": 0.6237,
+      "step": 9552
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.099058787224194e-05,
+      "loss": 0.9106,
+      "step": 9553
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.098801625263591e-05,
+      "loss": 0.9143,
+      "step": 9554
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.098544463302989e-05,
+      "loss": 0.7524,
+      "step": 9555
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.098287301342386e-05,
+      "loss": 0.6707,
+      "step": 9556
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.0980301393817826e-05,
+      "loss": 0.7242,
+      "step": 9557
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.09777297742118e-05,
+      "loss": 0.6592,
+      "step": 9558
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.097515815460577e-05,
+      "loss": 0.984,
+      "step": 9559
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.097258653499974e-05,
+      "loss": 0.5129,
+      "step": 9560
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.097001491539372e-05,
+      "loss": 0.7818,
+      "step": 9561
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.096744329578769e-05,
+      "loss": 1.046,
+      "step": 9562
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.0964871676181665e-05,
+      "loss": 0.7981,
+      "step": 9563
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.0962300056575634e-05,
+      "loss": 0.9494,
+      "step": 9564
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.0959728436969604e-05,
+      "loss": 0.6226,
+      "step": 9565
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.0957156817363574e-05,
+      "loss": 0.9994,
+      "step": 9566
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.095458519775755e-05,
+      "loss": 0.847,
+      "step": 9567
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.095201357815152e-05,
+      "loss": 0.7344,
+      "step": 9568
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.0949441958545496e-05,
+      "loss": 0.5203,
+      "step": 9569
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.0946870338939466e-05,
+      "loss": 0.7266,
+      "step": 9570
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.0944298719333436e-05,
+      "loss": 0.8332,
+      "step": 9571
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.0941727099727406e-05,
+      "loss": 0.6275,
+      "step": 9572
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.093915548012138e-05,
+      "loss": 0.6762,
+      "step": 9573
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.093658386051536e-05,
+      "loss": 0.8526,
+      "step": 9574
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.093401224090933e-05,
+      "loss": 0.802,
+      "step": 9575
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.09314406213033e-05,
+      "loss": 0.7766,
+      "step": 9576
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.092886900169727e-05,
+      "loss": 0.9863,
+      "step": 9577
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.092629738209124e-05,
+      "loss": 0.5097,
+      "step": 9578
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.092372576248522e-05,
+      "loss": 0.7209,
+      "step": 9579
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.092115414287919e-05,
+      "loss": 0.7614,
+      "step": 9580
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.091858252327316e-05,
+      "loss": 0.8356,
+      "step": 9581
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.091601090366713e-05,
+      "loss": 0.7598,
+      "step": 9582
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.09134392840611e-05,
+      "loss": 1.6791,
+      "step": 9583
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.0910867664455076e-05,
+      "loss": 0.7477,
+      "step": 9584
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.090829604484905e-05,
+      "loss": 0.8213,
+      "step": 9585
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.090572442524302e-05,
+      "loss": 0.7018,
+      "step": 9586
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.090315280563699e-05,
+      "loss": 0.962,
+      "step": 9587
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.090058118603096e-05,
+      "loss": 1.0712,
+      "step": 9588
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.089800956642494e-05,
+      "loss": 0.8254,
+      "step": 9589
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.089543794681891e-05,
+      "loss": 0.7805,
+      "step": 9590
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.0892866327212884e-05,
+      "loss": 0.489,
+      "step": 9591
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.0890294707606853e-05,
+      "loss": 0.7939,
+      "step": 9592
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.088772308800082e-05,
+      "loss": 0.3673,
+      "step": 9593
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.08851514683948e-05,
+      "loss": 0.995,
+      "step": 9594
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.088257984878877e-05,
+      "loss": 0.9158,
+      "step": 9595
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.088000822918274e-05,
+      "loss": 0.9961,
+      "step": 9596
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.0877436609576715e-05,
+      "loss": 0.6238,
+      "step": 9597
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.0874864989970685e-05,
+      "loss": 0.9452,
+      "step": 9598
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.0872293370364655e-05,
+      "loss": 0.5573,
+      "step": 9599
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.086972175075863e-05,
+      "loss": 0.9103,
+      "step": 9600
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.08671501311526e-05,
+      "loss": 0.9733,
+      "step": 9601
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.086457851154657e-05,
+      "loss": 0.8767,
+      "step": 9602
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.086200689194055e-05,
+      "loss": 0.9921,
+      "step": 9603
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.085943527233452e-05,
+      "loss": 0.8565,
+      "step": 9604
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.085686365272849e-05,
+      "loss": 0.9655,
+      "step": 9605
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.085429203312246e-05,
+      "loss": 0.7198,
+      "step": 9606
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.085172041351643e-05,
+      "loss": 0.6942,
+      "step": 9607
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.08491487939104e-05,
+      "loss": 0.9047,
+      "step": 9608
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.084657717430438e-05,
+      "loss": 0.7402,
+      "step": 9609
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.0844005554698355e-05,
+      "loss": 0.5588,
+      "step": 9610
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.0841433935092325e-05,
+      "loss": 0.6964,
+      "step": 9611
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.0838862315486295e-05,
+      "loss": 1.0354,
+      "step": 9612
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.0836290695880264e-05,
+      "loss": 0.9692,
+      "step": 9613
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.0833719076274234e-05,
+      "loss": 0.4664,
+      "step": 9614
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.083114745666821e-05,
+      "loss": 0.8391,
+      "step": 9615
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.082857583706219e-05,
+      "loss": 0.7637,
+      "step": 9616
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.0826004217456157e-05,
+      "loss": 0.8032,
+      "step": 9617
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.0823432597850126e-05,
+      "loss": 0.7408,
+      "step": 9618
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.0820860978244096e-05,
+      "loss": 0.6843,
+      "step": 9619
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.081828935863807e-05,
+      "loss": 0.6798,
+      "step": 9620
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.081571773903205e-05,
+      "loss": 0.9203,
+      "step": 9621
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.081314611942602e-05,
+      "loss": 0.5797,
+      "step": 9622
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.081057449981999e-05,
+      "loss": 0.698,
+      "step": 9623
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.080800288021396e-05,
+      "loss": 0.9369,
+      "step": 9624
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.080543126060793e-05,
+      "loss": 0.7425,
+      "step": 9625
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.0802859641001904e-05,
+      "loss": 0.8552,
+      "step": 9626
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.080028802139588e-05,
+      "loss": 0.6051,
+      "step": 9627
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.079771640178985e-05,
+      "loss": 0.6251,
+      "step": 9628
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.079514478218382e-05,
+      "loss": 0.6108,
+      "step": 9629
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.079257316257779e-05,
+      "loss": 0.9345,
+      "step": 9630
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.0790001542971766e-05,
+      "loss": 1.0006,
+      "step": 9631
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.0787429923365736e-05,
+      "loss": 0.9723,
+      "step": 9632
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.078485830375971e-05,
+      "loss": 1.1667,
+      "step": 9633
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.078228668415368e-05,
+      "loss": 0.8437,
+      "step": 9634
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.077971506454765e-05,
+      "loss": 0.9944,
+      "step": 9635
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.077714344494163e-05,
+      "loss": 0.8545,
+      "step": 9636
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.07745718253356e-05,
+      "loss": 0.6675,
+      "step": 9637
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.077200020572957e-05,
+      "loss": 0.6488,
+      "step": 9638
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.0769428586123544e-05,
+      "loss": 0.8261,
+      "step": 9639
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.0766856966517514e-05,
+      "loss": 0.694,
+      "step": 9640
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.076428534691149e-05,
+      "loss": 0.7722,
+      "step": 9641
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 3.076171372730546e-05,
+      "loss": 0.8526,
+      "step": 9642
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 3.075914210769943e-05,
+      "loss": 0.6892,
+      "step": 9643
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 3.07565704880934e-05,
+      "loss": 0.599,
+      "step": 9644
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 3.0753998868487375e-05,
+      "loss": 0.6301,
+      "step": 9645
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 3.0751427248881345e-05,
+      "loss": 0.6919,
+      "step": 9646
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 3.074885562927532e-05,
+      "loss": 0.771,
+      "step": 9647
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 3.074628400966929e-05,
+      "loss": 0.5893,
+      "step": 9648
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 3.074371239006326e-05,
+      "loss": 0.5707,
+      "step": 9649
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 3.074114077045723e-05,
+      "loss": 0.9363,
+      "step": 9650
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 3.073856915085121e-05,
+      "loss": 0.7019,
+      "step": 9651
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 3.0735997531245184e-05,
+      "loss": 0.6906,
+      "step": 9652
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 3.073342591163915e-05,
+      "loss": 1.0408,
+      "step": 9653
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 3.073085429203312e-05,
+      "loss": 0.6671,
+      "step": 9654
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 3.072828267242709e-05,
+      "loss": 0.407,
+      "step": 9655
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 3.072571105282106e-05,
+      "loss": 0.752,
+      "step": 9656
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 3.0723139433215046e-05,
+      "loss": 0.721,
+      "step": 9657
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 3.0720567813609015e-05,
+      "loss": 0.8895,
+      "step": 9658
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 3.0717996194002985e-05,
+      "loss": 0.5523,
+      "step": 9659
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 3.0715424574396955e-05,
+      "loss": 0.6679,
+      "step": 9660
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 3.0712852954790924e-05,
+      "loss": 0.7488,
+      "step": 9661
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 3.07102813351849e-05,
+      "loss": 0.8879,
+      "step": 9662
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 3.070770971557888e-05,
+      "loss": 0.9508,
+      "step": 9663
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 3.070513809597285e-05,
+      "loss": 1.047,
+      "step": 9664
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 3.070256647636682e-05,
+      "loss": 0.8017,
+      "step": 9665
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 3.0699994856760786e-05,
+      "loss": 0.9701,
+      "step": 9666
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 3.069742323715476e-05,
+      "loss": 0.7167,
+      "step": 9667
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 3.069485161754873e-05,
+      "loss": 0.5714,
+      "step": 9668
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 3.069227999794271e-05,
+      "loss": 0.8474,
+      "step": 9669
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 3.068970837833668e-05,
+      "loss": 0.7915,
+      "step": 9670
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 3.068713675873065e-05,
+      "loss": 0.6173,
+      "step": 9671
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 3.0684565139124625e-05,
+      "loss": 0.6966,
+      "step": 9672
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 3.0681993519518594e-05,
+      "loss": 1.2077,
+      "step": 9673
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 3.0679421899912564e-05,
+      "loss": 0.9228,
+      "step": 9674
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 3.067685028030654e-05,
+      "loss": 0.8443,
+      "step": 9675
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 3.067427866070051e-05,
+      "loss": 0.738,
+      "step": 9676
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 3.067170704109448e-05,
+      "loss": 0.6271,
+      "step": 9677
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 3.0669135421488456e-05,
+      "loss": 0.8429,
+      "step": 9678
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 3.0666563801882426e-05,
+      "loss": 0.7045,
+      "step": 9679
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 3.0663992182276396e-05,
+      "loss": 0.6833,
+      "step": 9680
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 3.066142056267037e-05,
+      "loss": 0.6624,
+      "step": 9681
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 3.065884894306434e-05,
+      "loss": 0.9885,
+      "step": 9682
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 3.065627732345832e-05,
+      "loss": 0.4849,
+      "step": 9683
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 3.065370570385229e-05,
+      "loss": 0.7068,
+      "step": 9684
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 3.065113408424626e-05,
+      "loss": 0.8224,
+      "step": 9685
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 3.064856246464023e-05,
+      "loss": 0.7927,
+      "step": 9686
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 3.0645990845034204e-05,
+      "loss": 0.9457,
+      "step": 9687
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 3.064341922542818e-05,
+      "loss": 0.5287,
+      "step": 9688
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 3.064084760582215e-05,
+      "loss": 0.5996,
+      "step": 9689
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 3.063827598621612e-05,
+      "loss": 0.9851,
+      "step": 9690
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 3.063570436661009e-05,
+      "loss": 0.9007,
+      "step": 9691
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 3.063313274700406e-05,
+      "loss": 0.5754,
+      "step": 9692
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 3.0630561127398036e-05,
+      "loss": 0.6025,
+      "step": 9693
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 3.062798950779201e-05,
+      "loss": 0.6741,
+      "step": 9694
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 3.062541788818598e-05,
+      "loss": 0.9306,
+      "step": 9695
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 3.062284626857995e-05,
+      "loss": 0.6504,
+      "step": 9696
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 3.062027464897392e-05,
+      "loss": 0.9628,
+      "step": 9697
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 3.06177030293679e-05,
+      "loss": 1.0886,
+      "step": 9698
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 3.0615131409761874e-05,
+      "loss": 0.9558,
+      "step": 9699
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 3.0612559790155844e-05,
+      "loss": 0.6721,
+      "step": 9700
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 3.0609988170549813e-05,
+      "loss": 1.0002,
+      "step": 9701
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 3.060741655094378e-05,
+      "loss": 1.127,
+      "step": 9702
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 3.060484493133775e-05,
+      "loss": 0.813,
+      "step": 9703
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 3.0602273311731736e-05,
+      "loss": 0.6551,
+      "step": 9704
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 3.0599701692125706e-05,
+      "loss": 0.8684,
+      "step": 9705
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 3.0597130072519675e-05,
+      "loss": 0.7111,
+      "step": 9706
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 3.0594558452913645e-05,
+      "loss": 0.7499,
+      "step": 9707
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 3.0591986833307615e-05,
+      "loss": 0.8077,
+      "step": 9708
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 3.058941521370159e-05,
+      "loss": 0.8435,
+      "step": 9709
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 3.058684359409557e-05,
+      "loss": 0.7269,
+      "step": 9710
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 3.058427197448954e-05,
+      "loss": 0.6158,
+      "step": 9711
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 3.058170035488351e-05,
+      "loss": 0.6923,
+      "step": 9712
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 3.057912873527748e-05,
+      "loss": 0.398,
+      "step": 9713
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 3.057655711567145e-05,
+      "loss": 0.7591,
+      "step": 9714
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 3.057398549606542e-05,
+      "loss": 0.925,
+      "step": 9715
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 3.05714138764594e-05,
+      "loss": 1.1667,
+      "step": 9716
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 3.056884225685337e-05,
+      "loss": 0.8088,
+      "step": 9717
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 3.056627063724734e-05,
+      "loss": 0.7186,
+      "step": 9718
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 3.0563699017641315e-05,
+      "loss": 0.7688,
+      "step": 9719
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 3.0561127398035285e-05,
+      "loss": 1.1002,
+      "step": 9720
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 3.0558555778429255e-05,
+      "loss": 0.6065,
+      "step": 9721
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 3.055598415882323e-05,
+      "loss": 0.8216,
+      "step": 9722
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 3.05534125392172e-05,
+      "loss": 0.7058,
+      "step": 9723
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 3.055084091961117e-05,
+      "loss": 0.988,
+      "step": 9724
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 3.054826930000515e-05,
+      "loss": 0.7533,
+      "step": 9725
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 3.0545697680399117e-05,
+      "loss": 0.8932,
+      "step": 9726
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 3.0543126060793086e-05,
+      "loss": 0.7307,
+      "step": 9727
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 3.054055444118706e-05,
+      "loss": 0.8497,
+      "step": 9728
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 3.053798282158103e-05,
+      "loss": 0.7266,
+      "step": 9729
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 3.053541120197501e-05,
+      "loss": 1.1532,
+      "step": 9730
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 3.053283958236898e-05,
+      "loss": 0.8154,
+      "step": 9731
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 3.053026796276295e-05,
+      "loss": 0.7297,
+      "step": 9732
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 3.052769634315692e-05,
+      "loss": 1.0219,
+      "step": 9733
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 3.0525124723550894e-05,
+      "loss": 0.5036,
+      "step": 9734
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 3.052255310394487e-05,
+      "loss": 0.807,
+      "step": 9735
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 3.051998148433884e-05,
+      "loss": 0.9694,
+      "step": 9736
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 3.051740986473281e-05,
+      "loss": 0.5029,
+      "step": 9737
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 3.051483824512678e-05,
+      "loss": 0.5856,
+      "step": 9738
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 3.0512266625520753e-05,
+      "loss": 0.7046,
+      "step": 9739
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 3.050969500591473e-05,
+      "loss": 1.0333,
+      "step": 9740
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 3.05071233863087e-05,
+      "loss": 0.6251,
+      "step": 9741
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 3.0504551766702672e-05,
+      "loss": 0.7577,
+      "step": 9742
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 3.0501980147096642e-05,
+      "loss": 0.9578,
+      "step": 9743
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 3.0499408527490615e-05,
+      "loss": 0.6245,
+      "step": 9744
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 3.0496836907884585e-05,
+      "loss": 0.6749,
+      "step": 9745
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 3.049426528827856e-05,
+      "loss": 1.2469,
+      "step": 9746
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 3.0491693668672534e-05,
+      "loss": 0.7522,
+      "step": 9747
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 3.0489122049066504e-05,
+      "loss": 0.4971,
+      "step": 9748
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 3.0486550429460477e-05,
+      "loss": 0.7271,
+      "step": 9749
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 3.0483978809854447e-05,
+      "loss": 0.7912,
+      "step": 9750
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 3.0481407190248416e-05,
+      "loss": 0.8112,
+      "step": 9751
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 3.0478835570642396e-05,
+      "loss": 1.0891,
+      "step": 9752
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 3.0476263951036366e-05,
+      "loss": 0.5572,
+      "step": 9753
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 3.0473692331430335e-05,
+      "loss": 0.7256,
+      "step": 9754
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 3.047112071182431e-05,
+      "loss": 1.082,
+      "step": 9755
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 3.0468549092218278e-05,
+      "loss": 0.5145,
+      "step": 9756
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 3.046597747261225e-05,
+      "loss": 0.7068,
+      "step": 9757
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 3.0463405853006228e-05,
+      "loss": 0.7059,
+      "step": 9758
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 3.0460834233400197e-05,
+      "loss": 0.7536,
+      "step": 9759
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 3.045826261379417e-05,
+      "loss": 0.5689,
+      "step": 9760
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 3.045569099418814e-05,
+      "loss": 1.2763,
+      "step": 9761
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 3.0453119374582113e-05,
+      "loss": 1.0546,
+      "step": 9762
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 3.0450547754976083e-05,
+      "loss": 1.1284,
+      "step": 9763
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 3.044797613537006e-05,
+      "loss": 0.6148,
+      "step": 9764
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 3.0445404515764032e-05,
+      "loss": 0.9251,
+      "step": 9765
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 3.0442832896158002e-05,
+      "loss": 1.1124,
+      "step": 9766
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 3.0440261276551972e-05,
+      "loss": 0.7688,
+      "step": 9767
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 3.0437689656945945e-05,
+      "loss": 0.9217,
+      "step": 9768
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 3.0435118037339915e-05,
+      "loss": 0.9674,
+      "step": 9769
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 3.043254641773389e-05,
+      "loss": 0.8611,
+      "step": 9770
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 3.0429974798127864e-05,
+      "loss": 0.8835,
+      "step": 9771
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 3.0427403178521834e-05,
+      "loss": 0.8011,
+      "step": 9772
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 3.0424831558915807e-05,
+      "loss": 0.8603,
+      "step": 9773
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 3.0422259939309777e-05,
+      "loss": 0.9122,
+      "step": 9774
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 3.041968831970375e-05,
+      "loss": 0.6849,
+      "step": 9775
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 3.0417116700097726e-05,
+      "loss": 1.322,
+      "step": 9776
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 3.0414545080491696e-05,
+      "loss": 0.5378,
+      "step": 9777
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 3.041197346088567e-05,
+      "loss": 0.5964,
+      "step": 9778
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 3.040940184127964e-05,
+      "loss": 0.9676,
+      "step": 9779
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 3.040683022167361e-05,
+      "loss": 0.6951,
+      "step": 9780
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 3.040425860206758e-05,
+      "loss": 1.107,
+      "step": 9781
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 3.0401686982461558e-05,
+      "loss": 0.9007,
+      "step": 9782
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 3.039911536285553e-05,
+      "loss": 1.1785,
+      "step": 9783
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 3.03965437432495e-05,
+      "loss": 1.1471,
+      "step": 9784
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 3.039397212364347e-05,
+      "loss": 0.6833,
+      "step": 9785
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 3.0391400504037443e-05,
+      "loss": 0.6371,
+      "step": 9786
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 3.0388828884431413e-05,
+      "loss": 0.889,
+      "step": 9787
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 3.038625726482539e-05,
+      "loss": 0.8652,
+      "step": 9788
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 3.0383685645219363e-05,
+      "loss": 0.953,
+      "step": 9789
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 3.0381114025613332e-05,
+      "loss": 0.7617,
+      "step": 9790
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 3.0378542406007305e-05,
+      "loss": 1.0923,
+      "step": 9791
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 3.0375970786401275e-05,
+      "loss": 0.8539,
+      "step": 9792
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 3.0373399166795248e-05,
+      "loss": 0.8175,
+      "step": 9793
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 3.0370827547189225e-05,
+      "loss": 1.0304,
+      "step": 9794
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 3.0368255927583194e-05,
+      "loss": 0.3858,
+      "step": 9795
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 3.0365684307977167e-05,
+      "loss": 1.1712,
+      "step": 9796
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 3.0363112688371137e-05,
+      "loss": 0.779,
+      "step": 9797
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 3.0360541068765107e-05,
+      "loss": 0.8396,
+      "step": 9798
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 3.035796944915908e-05,
+      "loss": 0.6693,
+      "step": 9799
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 3.0355397829553056e-05,
+      "loss": 1.1614,
+      "step": 9800
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 3.0352826209947026e-05,
+      "loss": 0.7148,
+      "step": 9801
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 3.0350254590341e-05,
+      "loss": 0.7477,
+      "step": 9802
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 3.034768297073497e-05,
+      "loss": 0.7794,
+      "step": 9803
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 3.0345111351128942e-05,
+      "loss": 0.5156,
+      "step": 9804
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 3.034253973152291e-05,
+      "loss": 1.0688,
+      "step": 9805
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 3.0339968111916888e-05,
+      "loss": 0.7197,
+      "step": 9806
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 3.033739649231086e-05,
+      "loss": 0.8288,
+      "step": 9807
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 3.033482487270483e-05,
+      "loss": 0.8733,
+      "step": 9808
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 3.0332253253098804e-05,
+      "loss": 0.8966,
+      "step": 9809
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 3.0329681633492773e-05,
+      "loss": 0.8938,
+      "step": 9810
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 3.0327110013886746e-05,
+      "loss": 0.935,
+      "step": 9811
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 3.0324538394280723e-05,
+      "loss": 0.8264,
+      "step": 9812
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 3.0321966774674693e-05,
+      "loss": 0.5201,
+      "step": 9813
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 3.0319395155068662e-05,
+      "loss": 0.6242,
+      "step": 9814
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 3.0316823535462635e-05,
+      "loss": 0.8685,
+      "step": 9815
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 3.0314251915856605e-05,
+      "loss": 0.8348,
+      "step": 9816
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 3.0311680296250578e-05,
+      "loss": 0.7158,
+      "step": 9817
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 3.0309108676644555e-05,
+      "loss": 0.7999,
+      "step": 9818
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 3.0306537057038524e-05,
+      "loss": 0.7385,
+      "step": 9819
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 3.0303965437432497e-05,
+      "loss": 0.6624,
+      "step": 9820
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 3.0301393817826467e-05,
+      "loss": 0.7071,
+      "step": 9821
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 3.029882219822044e-05,
+      "loss": 0.8178,
+      "step": 9822
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 3.029625057861441e-05,
+      "loss": 0.6419,
+      "step": 9823
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 3.0293678959008386e-05,
+      "loss": 1.0925,
+      "step": 9824
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 3.029110733940236e-05,
+      "loss": 0.8273,
+      "step": 9825
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 3.028853571979633e-05,
+      "loss": 0.8527,
+      "step": 9826
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 3.0285964100190302e-05,
+      "loss": 0.7462,
+      "step": 9827
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 3.0283392480584272e-05,
+      "loss": 0.8414,
+      "step": 9828
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 3.028082086097824e-05,
+      "loss": 1.0056,
+      "step": 9829
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 3.027824924137222e-05,
+      "loss": 0.7752,
+      "step": 9830
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 3.027567762176619e-05,
+      "loss": 0.8425,
+      "step": 9831
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 3.027310600216016e-05,
+      "loss": 0.7464,
+      "step": 9832
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 3.0270534382554134e-05,
+      "loss": 1.0048,
+      "step": 9833
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 3.0267962762948103e-05,
+      "loss": 0.6981,
+      "step": 9834
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 3.0265391143342076e-05,
+      "loss": 0.8676,
+      "step": 9835
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 3.0262819523736053e-05,
+      "loss": 0.4935,
+      "step": 9836
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 3.0260247904130023e-05,
+      "loss": 0.5173,
+      "step": 9837
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 3.0257676284523996e-05,
+      "loss": 0.6176,
+      "step": 9838
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 3.0255104664917965e-05,
+      "loss": 0.6841,
+      "step": 9839
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 3.025253304531194e-05,
+      "loss": 0.7571,
+      "step": 9840
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 3.0249961425705908e-05,
+      "loss": 0.9719,
+      "step": 9841
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 3.0247389806099885e-05,
+      "loss": 0.7193,
+      "step": 9842
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 3.0244818186493858e-05,
+      "loss": 0.403,
+      "step": 9843
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 3.0242246566887827e-05,
+      "loss": 1.0145,
+      "step": 9844
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 3.0239674947281797e-05,
+      "loss": 0.742,
+      "step": 9845
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 3.023710332767577e-05,
+      "loss": 0.9716,
+      "step": 9846
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 3.023453170806974e-05,
+      "loss": 0.4603,
+      "step": 9847
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 3.0231960088463716e-05,
+      "loss": 0.5764,
+      "step": 9848
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 3.022938846885769e-05,
+      "loss": 0.8662,
+      "step": 9849
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 3.022681684925166e-05,
+      "loss": 1.0894,
+      "step": 9850
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 3.0224245229645632e-05,
+      "loss": 0.6097,
+      "step": 9851
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 3.0221673610039602e-05,
+      "loss": 0.9251,
+      "step": 9852
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 3.0219101990433578e-05,
+      "loss": 0.748,
+      "step": 9853
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 3.021653037082755e-05,
+      "loss": 1.191,
+      "step": 9854
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 3.021395875122152e-05,
+      "loss": 0.6849,
+      "step": 9855
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 3.0211387131615494e-05,
+      "loss": 0.8511,
+      "step": 9856
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 3.0208815512009464e-05,
+      "loss": 0.695,
+      "step": 9857
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.0206243892403437e-05,
+      "loss": 0.6702,
+      "step": 9858
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.0203672272797413e-05,
+      "loss": 1.0659,
+      "step": 9859
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.0201100653191383e-05,
+      "loss": 1.0282,
+      "step": 9860
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.0198529033585353e-05,
+      "loss": 0.7166,
+      "step": 9861
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.0195957413979326e-05,
+      "loss": 0.4524,
+      "step": 9862
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.0193385794373295e-05,
+      "loss": 0.8597,
+      "step": 9863
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.019081417476727e-05,
+      "loss": 0.8197,
+      "step": 9864
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.0188242555161245e-05,
+      "loss": 0.681,
+      "step": 9865
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.0185670935555215e-05,
+      "loss": 0.9654,
+      "step": 9866
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.0183099315949188e-05,
+      "loss": 0.798,
+      "step": 9867
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.0180527696343157e-05,
+      "loss": 0.4763,
+      "step": 9868
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.017795607673713e-05,
+      "loss": 0.7173,
+      "step": 9869
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.01753844571311e-05,
+      "loss": 1.1282,
+      "step": 9870
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.0172812837525077e-05,
+      "loss": 0.9682,
+      "step": 9871
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.017024121791905e-05,
+      "loss": 0.8434,
+      "step": 9872
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.016766959831302e-05,
+      "loss": 0.939,
+      "step": 9873
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.0165097978706992e-05,
+      "loss": 0.6881,
+      "step": 9874
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.0162526359100962e-05,
+      "loss": 0.8429,
+      "step": 9875
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.0159954739494932e-05,
+      "loss": 1.0329,
+      "step": 9876
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.015738311988891e-05,
+      "loss": 0.7977,
+      "step": 9877
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.015481150028288e-05,
+      "loss": 0.8194,
+      "step": 9878
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.015223988067685e-05,
+      "loss": 0.7734,
+      "step": 9879
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.0149668261070824e-05,
+      "loss": 0.7587,
+      "step": 9880
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.0147096641464794e-05,
+      "loss": 0.6806,
+      "step": 9881
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.0144525021858767e-05,
+      "loss": 0.8886,
+      "step": 9882
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.0141953402252743e-05,
+      "loss": 0.608,
+      "step": 9883
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.0139381782646713e-05,
+      "loss": 0.814,
+      "step": 9884
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.0136810163040686e-05,
+      "loss": 0.6713,
+      "step": 9885
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.0134238543434656e-05,
+      "loss": 0.4378,
+      "step": 9886
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.013166692382863e-05,
+      "loss": 1.2329,
+      "step": 9887
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.01290953042226e-05,
+      "loss": 0.7005,
+      "step": 9888
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.0126523684616575e-05,
+      "loss": 0.6571,
+      "step": 9889
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.0123952065010548e-05,
+      "loss": 1.1007,
+      "step": 9890
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.0121380445404518e-05,
+      "loss": 0.8551,
+      "step": 9891
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.0118808825798487e-05,
+      "loss": 1.1969,
+      "step": 9892
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.011623720619246e-05,
+      "loss": 0.8341,
+      "step": 9893
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.011366558658643e-05,
+      "loss": 0.762,
+      "step": 9894
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.0111093966980407e-05,
+      "loss": 0.7277,
+      "step": 9895
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.010852234737438e-05,
+      "loss": 0.9163,
+      "step": 9896
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.010595072776835e-05,
+      "loss": 0.9842,
+      "step": 9897
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.0103379108162323e-05,
+      "loss": 1.113,
+      "step": 9898
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.0100807488556292e-05,
+      "loss": 0.5749,
+      "step": 9899
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.0098235868950265e-05,
+      "loss": 0.8463,
+      "step": 9900
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.0095664249344242e-05,
+      "loss": 0.625,
+      "step": 9901
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.009309262973821e-05,
+      "loss": 0.7248,
+      "step": 9902
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.0090521010132184e-05,
+      "loss": 0.554,
+      "step": 9903
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.0087949390526154e-05,
+      "loss": 0.6991,
+      "step": 9904
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.0085377770920127e-05,
+      "loss": 0.8725,
+      "step": 9905
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.0082806151314097e-05,
+      "loss": 0.8184,
+      "step": 9906
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.0080234531708073e-05,
+      "loss": 0.738,
+      "step": 9907
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.0077662912102046e-05,
+      "loss": 0.7862,
+      "step": 9908
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.0075091292496016e-05,
+      "loss": 0.8449,
+      "step": 9909
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.0072519672889986e-05,
+      "loss": 0.8554,
+      "step": 9910
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.006994805328396e-05,
+      "loss": 0.4887,
+      "step": 9911
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.006737643367793e-05,
+      "loss": 0.6438,
+      "step": 9912
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.0064804814071905e-05,
+      "loss": 0.8802,
+      "step": 9913
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.0062233194465878e-05,
+      "loss": 0.6005,
+      "step": 9914
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.0059661574859848e-05,
+      "loss": 0.8659,
+      "step": 9915
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.005708995525382e-05,
+      "loss": 0.8982,
+      "step": 9916
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.005451833564779e-05,
+      "loss": 0.6733,
+      "step": 9917
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.0051946716041764e-05,
+      "loss": 0.7408,
+      "step": 9918
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.004937509643574e-05,
+      "loss": 0.6188,
+      "step": 9919
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.004680347682971e-05,
+      "loss": 0.7491,
+      "step": 9920
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.0044231857223683e-05,
+      "loss": 0.761,
+      "step": 9921
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.0041660237617653e-05,
+      "loss": 0.8126,
+      "step": 9922
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.0039088618011622e-05,
+      "loss": 0.6899,
+      "step": 9923
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.0036516998405595e-05,
+      "loss": 0.7733,
+      "step": 9924
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.0033945378799572e-05,
+      "loss": 0.4719,
+      "step": 9925
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.003137375919354e-05,
+      "loss": 0.8411,
+      "step": 9926
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.0028802139587515e-05,
+      "loss": 0.9087,
+      "step": 9927
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.0026230519981484e-05,
+      "loss": 0.4437,
+      "step": 9928
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.0023658900375457e-05,
+      "loss": 0.6041,
+      "step": 9929
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.0021087280769427e-05,
+      "loss": 0.88,
+      "step": 9930
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.0018515661163403e-05,
+      "loss": 0.876,
+      "step": 9931
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.0015944041557377e-05,
+      "loss": 0.9767,
+      "step": 9932
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.0013372421951346e-05,
+      "loss": 0.8803,
+      "step": 9933
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.001080080234532e-05,
+      "loss": 0.8356,
+      "step": 9934
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.000822918273929e-05,
+      "loss": 0.6715,
+      "step": 9935
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.0005657563133262e-05,
+      "loss": 0.9585,
+      "step": 9936
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.000308594352724e-05,
+      "loss": 0.7436,
+      "step": 9937
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.0000514323921208e-05,
+      "loss": 0.7952,
+      "step": 9938
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 2.9997942704315178e-05,
+      "loss": 0.9275,
+      "step": 9939
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 2.999537108470915e-05,
+      "loss": 0.6194,
+      "step": 9940
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 2.999279946510312e-05,
+      "loss": 0.9441,
+      "step": 9941
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 2.9990227845497094e-05,
+      "loss": 0.8479,
+      "step": 9942
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 2.998765622589107e-05,
+      "loss": 0.6912,
+      "step": 9943
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 2.998508460628504e-05,
+      "loss": 0.573,
+      "step": 9944
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 2.9982512986679013e-05,
+      "loss": 1.3324,
+      "step": 9945
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 2.9979941367072983e-05,
+      "loss": 0.6752,
+      "step": 9946
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 2.9977369747466956e-05,
+      "loss": 1.0068,
+      "step": 9947
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 2.9974798127860925e-05,
+      "loss": 0.8062,
+      "step": 9948
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 2.9972226508254902e-05,
+      "loss": 0.9304,
+      "step": 9949
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 2.9969654888648875e-05,
+      "loss": 1.1819,
+      "step": 9950
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 2.9967083269042845e-05,
+      "loss": 0.4628,
+      "step": 9951
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 2.9964511649436818e-05,
+      "loss": 0.7988,
+      "step": 9952
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 2.9961940029830787e-05,
+      "loss": 1.0786,
+      "step": 9953
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 2.9959368410224757e-05,
+      "loss": 0.7536,
+      "step": 9954
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 2.9956796790618737e-05,
+      "loss": 1.1587,
+      "step": 9955
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 2.9954225171012707e-05,
+      "loss": 0.7899,
+      "step": 9956
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 2.9951653551406676e-05,
+      "loss": 0.9461,
+      "step": 9957
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 2.994908193180065e-05,
+      "loss": 0.7692,
+      "step": 9958
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 2.994651031219462e-05,
+      "loss": 0.8321,
+      "step": 9959
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 2.9943938692588592e-05,
+      "loss": 1.1688,
+      "step": 9960
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 2.994136707298257e-05,
+      "loss": 0.4871,
+      "step": 9961
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 2.9938795453376538e-05,
+      "loss": 0.641,
+      "step": 9962
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 2.993622383377051e-05,
+      "loss": 0.7712,
+      "step": 9963
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 2.993365221416448e-05,
+      "loss": 0.6291,
+      "step": 9964
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 2.9931080594558454e-05,
+      "loss": 0.8073,
+      "step": 9965
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.9928508974952424e-05,
+      "loss": 0.6443,
+      "step": 9966
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.99259373553464e-05,
+      "loss": 0.7398,
+      "step": 9967
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.9923365735740373e-05,
+      "loss": 0.8057,
+      "step": 9968
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.9920794116134343e-05,
+      "loss": 0.5192,
+      "step": 9969
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.9918222496528313e-05,
+      "loss": 0.6768,
+      "step": 9970
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.9915650876922286e-05,
+      "loss": 0.942,
+      "step": 9971
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.9913079257316255e-05,
+      "loss": 0.8417,
+      "step": 9972
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.9910507637710232e-05,
+      "loss": 0.9539,
+      "step": 9973
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.9907936018104205e-05,
+      "loss": 1.2391,
+      "step": 9974
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.9905364398498175e-05,
+      "loss": 0.6314,
+      "step": 9975
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.9902792778892148e-05,
+      "loss": 0.9153,
+      "step": 9976
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.9900221159286117e-05,
+      "loss": 0.851,
+      "step": 9977
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.989764953968009e-05,
+      "loss": 0.8319,
+      "step": 9978
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.9895077920074067e-05,
+      "loss": 1.1141,
+      "step": 9979
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.9892506300468037e-05,
+      "loss": 0.7737,
+      "step": 9980
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.988993468086201e-05,
+      "loss": 0.8631,
+      "step": 9981
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.988736306125598e-05,
+      "loss": 0.9605,
+      "step": 9982
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.9884791441649952e-05,
+      "loss": 0.669,
+      "step": 9983
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.9882219822043922e-05,
+      "loss": 0.7542,
+      "step": 9984
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.98796482024379e-05,
+      "loss": 0.6435,
+      "step": 9985
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.9877076582831868e-05,
+      "loss": 0.7761,
+      "step": 9986
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.987450496322584e-05,
+      "loss": 0.9545,
+      "step": 9987
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.987193334361981e-05,
+      "loss": 0.7775,
+      "step": 9988
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.9869361724013784e-05,
+      "loss": 0.573,
+      "step": 9989
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.9866790104407754e-05,
+      "loss": 0.7508,
+      "step": 9990
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.986421848480173e-05,
+      "loss": 0.9265,
+      "step": 9991
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.9861646865195703e-05,
+      "loss": 0.6079,
+      "step": 9992
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.9859075245589673e-05,
+      "loss": 0.6887,
+      "step": 9993
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.9856503625983646e-05,
+      "loss": 0.7398,
+      "step": 9994
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.9853932006377616e-05,
+      "loss": 0.8979,
+      "step": 9995
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.985136038677159e-05,
+      "loss": 0.806,
+      "step": 9996
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.9848788767165565e-05,
+      "loss": 0.901,
+      "step": 9997
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.9846217147559535e-05,
+      "loss": 0.6407,
+      "step": 9998
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.9843645527953508e-05,
+      "loss": 0.8835,
+      "step": 9999
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.9841073908347478e-05,
+      "loss": 0.7186,
+      "step": 10000
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.9838502288741447e-05,
+      "loss": 0.5235,
+      "step": 10001
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.9835930669135427e-05,
+      "loss": 0.6087,
+      "step": 10002
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.9833359049529397e-05,
+      "loss": 0.6631,
+      "step": 10003
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.9830787429923367e-05,
+      "loss": 0.9841,
+      "step": 10004
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.982821581031734e-05,
+      "loss": 0.5846,
+      "step": 10005
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.982564419071131e-05,
+      "loss": 0.6564,
+      "step": 10006
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.9823072571105282e-05,
+      "loss": 0.9772,
+      "step": 10007
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.982050095149926e-05,
+      "loss": 0.6957,
+      "step": 10008
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.981792933189323e-05,
+      "loss": 0.9122,
+      "step": 10009
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.9815357712287202e-05,
+      "loss": 0.3944,
+      "step": 10010
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.981278609268117e-05,
+      "loss": 0.8666,
+      "step": 10011
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.9810214473075144e-05,
+      "loss": 1.0403,
+      "step": 10012
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.9807642853469114e-05,
+      "loss": 0.7989,
+      "step": 10013
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.980507123386309e-05,
+      "loss": 0.7145,
+      "step": 10014
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.9802499614257064e-05,
+      "loss": 1.0523,
+      "step": 10015
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.9799927994651033e-05,
+      "loss": 0.9191,
+      "step": 10016
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.9797356375045003e-05,
+      "loss": 0.6345,
+      "step": 10017
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.9794784755438976e-05,
+      "loss": 0.6147,
+      "step": 10018
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.9792213135832946e-05,
+      "loss": 0.8409,
+      "step": 10019
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.9789641516226922e-05,
+      "loss": 0.7278,
+      "step": 10020
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.9787069896620895e-05,
+      "loss": 1.0281,
+      "step": 10021
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.9784498277014865e-05,
+      "loss": 0.6618,
+      "step": 10022
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.9781926657408838e-05,
+      "loss": 0.5898,
+      "step": 10023
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.9779355037802808e-05,
+      "loss": 0.6018,
+      "step": 10024
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.977678341819678e-05,
+      "loss": 0.7529,
+      "step": 10025
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.9774211798590757e-05,
+      "loss": 0.7273,
+      "step": 10026
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.9771640178984727e-05,
+      "loss": 0.712,
+      "step": 10027
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.97690685593787e-05,
+      "loss": 1.1045,
+      "step": 10028
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.976649693977267e-05,
+      "loss": 1.0274,
+      "step": 10029
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.9763925320166643e-05,
+      "loss": 0.5897,
+      "step": 10030
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.9761353700560613e-05,
+      "loss": 0.6888,
+      "step": 10031
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.975878208095459e-05,
+      "loss": 0.8543,
+      "step": 10032
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.9756210461348562e-05,
+      "loss": 0.917,
+      "step": 10033
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.9753638841742532e-05,
+      "loss": 0.9745,
+      "step": 10034
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.97510672221365e-05,
+      "loss": 0.7692,
+      "step": 10035
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.9748495602530475e-05,
+      "loss": 0.826,
+      "step": 10036
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.9745923982924444e-05,
+      "loss": 1.0757,
+      "step": 10037
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.974335236331842e-05,
+      "loss": 0.8002,
+      "step": 10038
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.9740780743712394e-05,
+      "loss": 0.7914,
+      "step": 10039
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.9738209124106363e-05,
+      "loss": 0.8346,
+      "step": 10040
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.9735637504500336e-05,
+      "loss": 1.0206,
+      "step": 10041
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.9733065884894306e-05,
+      "loss": 0.7851,
+      "step": 10042
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.973049426528828e-05,
+      "loss": 0.844,
+      "step": 10043
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.9727922645682256e-05,
+      "loss": 0.6611,
+      "step": 10044
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.9725351026076225e-05,
+      "loss": 0.7719,
+      "step": 10045
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.97227794064702e-05,
+      "loss": 0.7828,
+      "step": 10046
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.9720207786864168e-05,
+      "loss": 1.0913,
+      "step": 10047
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.9717636167258138e-05,
+      "loss": 0.7182,
+      "step": 10048
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.971506454765211e-05,
+      "loss": 0.7366,
+      "step": 10049
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.9712492928046087e-05,
+      "loss": 0.7909,
+      "step": 10050
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.9709921308440057e-05,
+      "loss": 0.986,
+      "step": 10051
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.970734968883403e-05,
+      "loss": 0.5029,
+      "step": 10052
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.9704778069228e-05,
+      "loss": 0.7846,
+      "step": 10053
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.9702206449621973e-05,
+      "loss": 0.6459,
+      "step": 10054
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.9699634830015943e-05,
+      "loss": 0.9707,
+      "step": 10055
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.969706321040992e-05,
+      "loss": 0.5978,
+      "step": 10056
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.9694491590803892e-05,
+      "loss": 0.7166,
+      "step": 10057
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.9691919971197862e-05,
+      "loss": 0.4095,
+      "step": 10058
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.9689348351591835e-05,
+      "loss": 0.7752,
+      "step": 10059
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.9686776731985805e-05,
+      "loss": 0.9627,
+      "step": 10060
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.9684205112379778e-05,
+      "loss": 0.6233,
+      "step": 10061
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.9681633492773754e-05,
+      "loss": 0.914,
+      "step": 10062
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.9679061873167724e-05,
+      "loss": 1.0002,
+      "step": 10063
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.9676490253561693e-05,
+      "loss": 0.6786,
+      "step": 10064
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.9673918633955667e-05,
+      "loss": 0.7562,
+      "step": 10065
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.9671347014349636e-05,
+      "loss": 0.9227,
+      "step": 10066
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.966877539474361e-05,
+      "loss": 0.9778,
+      "step": 10067
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.9666203775137586e-05,
+      "loss": 0.7065,
+      "step": 10068
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.9663632155531555e-05,
+      "loss": 1.0885,
+      "step": 10069
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.966106053592553e-05,
+      "loss": 0.9004,
+      "step": 10070
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.9658488916319498e-05,
+      "loss": 0.6637,
+      "step": 10071
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.965591729671347e-05,
+      "loss": 0.6603,
+      "step": 10072
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.965334567710744e-05,
+      "loss": 0.8274,
+      "step": 10073
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.9650774057501417e-05,
+      "loss": 0.5614,
+      "step": 10074
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.964820243789539e-05,
+      "loss": 0.702,
+      "step": 10075
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.964563081828936e-05,
+      "loss": 0.8657,
+      "step": 10076
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.9643059198683333e-05,
+      "loss": 0.727,
+      "step": 10077
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.9640487579077303e-05,
+      "loss": 0.8202,
+      "step": 10078
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.9637915959471273e-05,
+      "loss": 0.7424,
+      "step": 10079
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.9635344339865252e-05,
+      "loss": 0.4959,
+      "step": 10080
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.9632772720259222e-05,
+      "loss": 0.794,
+      "step": 10081
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.9630201100653192e-05,
+      "loss": 0.9763,
+      "step": 10082
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.9627629481047165e-05,
+      "loss": 0.9945,
+      "step": 10083
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.9625057861441135e-05,
+      "loss": 0.5566,
+      "step": 10084
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.9622486241835108e-05,
+      "loss": 1.184,
+      "step": 10085
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.9619914622229084e-05,
+      "loss": 0.8331,
+      "step": 10086
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.9617343002623054e-05,
+      "loss": 0.6181,
+      "step": 10087
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.9614771383017027e-05,
+      "loss": 0.7449,
+      "step": 10088
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.9612199763410997e-05,
+      "loss": 0.6432,
+      "step": 10089
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.960962814380497e-05,
+      "loss": 0.9908,
+      "step": 10090
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.960705652419894e-05,
+      "loss": 0.7461,
+      "step": 10091
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.9604484904592916e-05,
+      "loss": 0.8452,
+      "step": 10092
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.960191328498689e-05,
+      "loss": 1.0323,
+      "step": 10093
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.959934166538086e-05,
+      "loss": 0.9689,
+      "step": 10094
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.9596770045774828e-05,
+      "loss": 0.4777,
+      "step": 10095
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.95941984261688e-05,
+      "loss": 0.8316,
+      "step": 10096
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.959162680656277e-05,
+      "loss": 0.7244,
+      "step": 10097
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.9589055186956747e-05,
+      "loss": 0.8439,
+      "step": 10098
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.958648356735072e-05,
+      "loss": 1.2616,
+      "step": 10099
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.958391194774469e-05,
+      "loss": 0.9132,
+      "step": 10100
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.9581340328138663e-05,
+      "loss": 0.8925,
+      "step": 10101
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.9578768708532633e-05,
+      "loss": 1.0067,
+      "step": 10102
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.9576197088926606e-05,
+      "loss": 0.8102,
+      "step": 10103
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.9573625469320583e-05,
+      "loss": 0.8718,
+      "step": 10104
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.9571053849714552e-05,
+      "loss": 0.6926,
+      "step": 10105
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.9568482230108525e-05,
+      "loss": 1.1372,
+      "step": 10106
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.9565910610502495e-05,
+      "loss": 1.0074,
+      "step": 10107
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.9563338990896468e-05,
+      "loss": 0.6925,
+      "step": 10108
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.9560767371290438e-05,
+      "loss": 0.8636,
+      "step": 10109
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.9558195751684414e-05,
+      "loss": 0.6733,
+      "step": 10110
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.9555624132078384e-05,
+      "loss": 0.8162,
+      "step": 10111
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.9553052512472357e-05,
+      "loss": 0.9576,
+      "step": 10112
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.9550480892866327e-05,
+      "loss": 1.0248,
+      "step": 10113
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.95479092732603e-05,
+      "loss": 0.6348,
+      "step": 10114
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.954533765365427e-05,
+      "loss": 0.7842,
+      "step": 10115
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.9542766034048246e-05,
+      "loss": 0.6827,
+      "step": 10116
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.954019441444222e-05,
+      "loss": 0.5757,
+      "step": 10117
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.953762279483619e-05,
+      "loss": 0.7233,
+      "step": 10118
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.953505117523016e-05,
+      "loss": 0.7432,
+      "step": 10119
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.953247955562413e-05,
+      "loss": 0.6057,
+      "step": 10120
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.9529907936018104e-05,
+      "loss": 0.8507,
+      "step": 10121
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.952733631641208e-05,
+      "loss": 0.6696,
+      "step": 10122
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.952476469680605e-05,
+      "loss": 0.9399,
+      "step": 10123
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.9522193077200024e-05,
+      "loss": 0.8279,
+      "step": 10124
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.9519621457593993e-05,
+      "loss": 0.8036,
+      "step": 10125
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.9517049837987963e-05,
+      "loss": 0.966,
+      "step": 10126
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.9514478218381936e-05,
+      "loss": 0.7443,
+      "step": 10127
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.9511906598775913e-05,
+      "loss": 0.7045,
+      "step": 10128
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.9509334979169882e-05,
+      "loss": 0.9195,
+      "step": 10129
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.9506763359563855e-05,
+      "loss": 0.8723,
+      "step": 10130
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.9504191739957825e-05,
+      "loss": 0.6105,
+      "step": 10131
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.9501620120351798e-05,
+      "loss": 0.7722,
+      "step": 10132
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.9499048500745768e-05,
+      "loss": 0.7373,
+      "step": 10133
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.9496476881139744e-05,
+      "loss": 0.6912,
+      "step": 10134
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.9493905261533717e-05,
+      "loss": 1.0901,
+      "step": 10135
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.9491333641927687e-05,
+      "loss": 0.9738,
+      "step": 10136
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.948876202232166e-05,
+      "loss": 0.9728,
+      "step": 10137
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.948619040271563e-05,
+      "loss": 0.7439,
+      "step": 10138
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.9483618783109603e-05,
+      "loss": 0.8164,
+      "step": 10139
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.948104716350358e-05,
+      "loss": 0.5694,
+      "step": 10140
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.947847554389755e-05,
+      "loss": 0.7153,
+      "step": 10141
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.947590392429152e-05,
+      "loss": 0.9052,
+      "step": 10142
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.9473332304685492e-05,
+      "loss": 0.9119,
+      "step": 10143
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.947076068507946e-05,
+      "loss": 0.6983,
+      "step": 10144
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.9468189065473434e-05,
+      "loss": 0.5769,
+      "step": 10145
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.946561744586741e-05,
+      "loss": 0.9171,
+      "step": 10146
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.946304582626138e-05,
+      "loss": 0.81,
+      "step": 10147
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.9460474206655354e-05,
+      "loss": 1.2139,
+      "step": 10148
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.9457902587049323e-05,
+      "loss": 0.9391,
+      "step": 10149
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.9455330967443296e-05,
+      "loss": 0.834,
+      "step": 10150
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.9452759347837273e-05,
+      "loss": 0.7142,
+      "step": 10151
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.9450187728231243e-05,
+      "loss": 1.0926,
+      "step": 10152
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.9447616108625216e-05,
+      "loss": 1.0509,
+      "step": 10153
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.9445044489019185e-05,
+      "loss": 0.9315,
+      "step": 10154
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.944247286941316e-05,
+      "loss": 0.6054,
+      "step": 10155
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.9439901249807128e-05,
+      "loss": 0.8036,
+      "step": 10156
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.9437329630201105e-05,
+      "loss": 0.6582,
+      "step": 10157
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.9434758010595078e-05,
+      "loss": 0.8915,
+      "step": 10158
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.9432186390989047e-05,
+      "loss": 0.8188,
+      "step": 10159
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.9429614771383017e-05,
+      "loss": 0.7124,
+      "step": 10160
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.942704315177699e-05,
+      "loss": 0.7178,
+      "step": 10161
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.942447153217096e-05,
+      "loss": 0.9612,
+      "step": 10162
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.9421899912564936e-05,
+      "loss": 0.6475,
+      "step": 10163
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.941932829295891e-05,
+      "loss": 0.709,
+      "step": 10164
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.941675667335288e-05,
+      "loss": 1.0637,
+      "step": 10165
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.9414185053746852e-05,
+      "loss": 0.8409,
+      "step": 10166
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.9411613434140822e-05,
+      "loss": 1.1007,
+      "step": 10167
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.9409041814534795e-05,
+      "loss": 0.6838,
+      "step": 10168
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.940647019492877e-05,
+      "loss": 0.5602,
+      "step": 10169
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.940389857532274e-05,
+      "loss": 0.655,
+      "step": 10170
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.9401326955716714e-05,
+      "loss": 0.8457,
+      "step": 10171
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.9398755336110684e-05,
+      "loss": 0.779,
+      "step": 10172
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.9396183716504653e-05,
+      "loss": 0.5797,
+      "step": 10173
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.9393612096898626e-05,
+      "loss": 0.7949,
+      "step": 10174
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.9391040477292603e-05,
+      "loss": 0.8697,
+      "step": 10175
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.9388468857686573e-05,
+      "loss": 0.7141,
+      "step": 10176
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.9385897238080546e-05,
+      "loss": 0.9517,
+      "step": 10177
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.9383325618474515e-05,
+      "loss": 0.634,
+      "step": 10178
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.938075399886849e-05,
+      "loss": 0.7463,
+      "step": 10179
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.9378182379262458e-05,
+      "loss": 0.8328,
+      "step": 10180
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.9375610759656435e-05,
+      "loss": 0.875,
+      "step": 10181
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 2.9373039140050408e-05,
+      "loss": 1.0969,
+      "step": 10182
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 2.9370467520444377e-05,
+      "loss": 0.782,
+      "step": 10183
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 2.936789590083835e-05,
+      "loss": 0.9189,
+      "step": 10184
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 2.936532428123232e-05,
+      "loss": 0.808,
+      "step": 10185
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 2.9362752661626293e-05,
+      "loss": 0.429,
+      "step": 10186
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 2.936018104202027e-05,
+      "loss": 0.7805,
+      "step": 10187
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 2.935760942241424e-05,
+      "loss": 0.7019,
+      "step": 10188
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 2.935503780280821e-05,
+      "loss": 0.7712,
+      "step": 10189
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 2.9352466183202182e-05,
+      "loss": 0.8404,
+      "step": 10190
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 2.9349894563596152e-05,
+      "loss": 0.8548,
+      "step": 10191
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 2.9347322943990125e-05,
+      "loss": 0.5602,
+      "step": 10192
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 2.93447513243841e-05,
+      "loss": 0.9285,
+      "step": 10193
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 2.934217970477807e-05,
+      "loss": 0.6209,
+      "step": 10194
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 2.9339608085172044e-05,
+      "loss": 0.5466,
+      "step": 10195
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 2.9337036465566014e-05,
+      "loss": 0.552,
+      "step": 10196
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 2.9334464845959987e-05,
+      "loss": 0.4618,
+      "step": 10197
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 2.9331893226353957e-05,
+      "loss": 0.7927,
+      "step": 10198
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 2.9329321606747933e-05,
+      "loss": 0.8571,
+      "step": 10199
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 2.9326749987141906e-05,
+      "loss": 0.8266,
+      "step": 10200
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 2.9324178367535876e-05,
+      "loss": 0.8537,
+      "step": 10201
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 2.932160674792985e-05,
+      "loss": 0.4982,
+      "step": 10202
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 2.931903512832382e-05,
+      "loss": 0.975,
+      "step": 10203
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 2.9316463508717788e-05,
+      "loss": 0.4821,
+      "step": 10204
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 2.9313891889111768e-05,
+      "loss": 0.6506,
+      "step": 10205
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 2.9311320269505738e-05,
+      "loss": 0.934,
+      "step": 10206
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 2.9308748649899707e-05,
+      "loss": 0.935,
+      "step": 10207
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 2.930617703029368e-05,
+      "loss": 0.8274,
+      "step": 10208
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 2.930360541068765e-05,
+      "loss": 0.8582,
+      "step": 10209
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 2.9301033791081623e-05,
+      "loss": 0.7807,
+      "step": 10210
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 2.92984621714756e-05,
+      "loss": 0.7214,
+      "step": 10211
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 2.929589055186957e-05,
+      "loss": 0.7186,
+      "step": 10212
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 2.9293318932263542e-05,
+      "loss": 0.7128,
+      "step": 10213
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 2.9290747312657512e-05,
+      "loss": 0.9664,
+      "step": 10214
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 2.9288175693051485e-05,
+      "loss": 1.0629,
+      "step": 10215
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 2.9285604073445455e-05,
+      "loss": 1.3564,
+      "step": 10216
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 2.928303245383943e-05,
+      "loss": 0.6707,
+      "step": 10217
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 2.9280460834233404e-05,
+      "loss": 0.9408,
+      "step": 10218
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 2.9277889214627374e-05,
+      "loss": 0.7182,
+      "step": 10219
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 2.9275317595021344e-05,
+      "loss": 0.8563,
+      "step": 10220
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 2.9272745975415317e-05,
+      "loss": 0.8908,
+      "step": 10221
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 2.9270174355809287e-05,
+      "loss": 0.744,
+      "step": 10222
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 2.9267602736203263e-05,
+      "loss": 0.5786,
+      "step": 10223
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 2.9265031116597236e-05,
+      "loss": 0.6383,
+      "step": 10224
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 2.9262459496991206e-05,
+      "loss": 0.6726,
+      "step": 10225
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 2.925988787738518e-05,
+      "loss": 0.7886,
+      "step": 10226
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 2.925731625777915e-05,
+      "loss": 0.4316,
+      "step": 10227
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 2.925474463817312e-05,
+      "loss": 0.8597,
+      "step": 10228
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 2.9252173018567098e-05,
+      "loss": 0.5537,
+      "step": 10229
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 2.9249601398961068e-05,
+      "loss": 0.9088,
+      "step": 10230
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 2.924702977935504e-05,
+      "loss": 0.7648,
+      "step": 10231
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 2.924445815974901e-05,
+      "loss": 0.6508,
+      "step": 10232
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 2.9241886540142984e-05,
+      "loss": 0.5424,
+      "step": 10233
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 2.9239314920536953e-05,
+      "loss": 0.7461,
+      "step": 10234
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 2.923674330093093e-05,
+      "loss": 0.7371,
+      "step": 10235
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 2.92341716813249e-05,
+      "loss": 0.7004,
+      "step": 10236
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 2.9231600061718873e-05,
+      "loss": 0.7507,
+      "step": 10237
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 2.9229028442112842e-05,
+      "loss": 0.7493,
+      "step": 10238
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 2.9226456822506815e-05,
+      "loss": 0.6472,
+      "step": 10239
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 2.9223885202900785e-05,
+      "loss": 0.9892,
+      "step": 10240
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 2.922131358329476e-05,
+      "loss": 0.7508,
+      "step": 10241
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 2.9218741963688735e-05,
+      "loss": 0.8183,
+      "step": 10242
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 2.9216170344082704e-05,
+      "loss": 0.8024,
+      "step": 10243
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 2.9213598724476677e-05,
+      "loss": 0.6871,
+      "step": 10244
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 2.9211027104870647e-05,
+      "loss": 0.8153,
+      "step": 10245
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 2.920845548526462e-05,
+      "loss": 0.6007,
+      "step": 10246
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 2.9205883865658596e-05,
+      "loss": 0.7208,
+      "step": 10247
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 2.9203312246052566e-05,
+      "loss": 0.8432,
+      "step": 10248
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 2.920074062644654e-05,
+      "loss": 0.6881,
+      "step": 10249
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 2.919816900684051e-05,
+      "loss": 0.869,
+      "step": 10250
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 2.919559738723448e-05,
+      "loss": 1.2389,
+      "step": 10251
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 2.919302576762845e-05,
+      "loss": 0.6096,
+      "step": 10252
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 2.9190454148022428e-05,
+      "loss": 0.5942,
+      "step": 10253
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 2.9187882528416398e-05,
+      "loss": 0.6873,
+      "step": 10254
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 2.918531090881037e-05,
+      "loss": 0.5425,
+      "step": 10255
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 2.918273928920434e-05,
+      "loss": 0.3638,
+      "step": 10256
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 2.9180167669598314e-05,
+      "loss": 0.9004,
+      "step": 10257
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 2.9177596049992283e-05,
+      "loss": 0.6649,
+      "step": 10258
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 2.917502443038626e-05,
+      "loss": 0.9437,
+      "step": 10259
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 2.9172452810780233e-05,
+      "loss": 0.635,
+      "step": 10260
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 2.9169881191174203e-05,
+      "loss": 1.2571,
+      "step": 10261
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 2.9167309571568176e-05,
+      "loss": 0.6239,
+      "step": 10262
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 2.9164737951962145e-05,
+      "loss": 0.7401,
+      "step": 10263
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 2.916216633235612e-05,
+      "loss": 0.7267,
+      "step": 10264
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 2.9159594712750095e-05,
+      "loss": 1.0399,
+      "step": 10265
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 2.9157023093144065e-05,
+      "loss": 1.1939,
+      "step": 10266
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 2.9154451473538034e-05,
+      "loss": 0.9453,
+      "step": 10267
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 2.9151879853932007e-05,
+      "loss": 0.8706,
+      "step": 10268
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 2.9149308234325977e-05,
+      "loss": 0.8191,
+      "step": 10269
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 2.914673661471995e-05,
+      "loss": 0.8972,
+      "step": 10270
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 2.9144164995113927e-05,
+      "loss": 1.1512,
+      "step": 10271
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 2.9141593375507896e-05,
+      "loss": 0.7249,
+      "step": 10272
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 2.913902175590187e-05,
+      "loss": 0.8213,
+      "step": 10273
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 2.913645013629584e-05,
+      "loss": 0.7568,
+      "step": 10274
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 2.9133878516689812e-05,
+      "loss": 0.7323,
+      "step": 10275
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 2.9131306897083782e-05,
+      "loss": 0.5948,
+      "step": 10276
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 2.9128735277477758e-05,
+      "loss": 0.8045,
+      "step": 10277
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 2.912616365787173e-05,
+      "loss": 0.8954,
+      "step": 10278
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 2.91235920382657e-05,
+      "loss": 0.7548,
+      "step": 10279
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 2.9121020418659674e-05,
+      "loss": 0.7226,
+      "step": 10280
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 2.9118448799053644e-05,
+      "loss": 1.031,
+      "step": 10281
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 2.9115877179447613e-05,
+      "loss": 0.6523,
+      "step": 10282
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 2.9113305559841593e-05,
+      "loss": 0.7117,
+      "step": 10283
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 2.9110733940235563e-05,
+      "loss": 0.4898,
+      "step": 10284
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 2.9108162320629533e-05,
+      "loss": 0.7654,
+      "step": 10285
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 2.9105590701023506e-05,
+      "loss": 0.8476,
+      "step": 10286
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 2.9103019081417475e-05,
+      "loss": 0.7337,
+      "step": 10287
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 2.910044746181145e-05,
+      "loss": 0.8476,
+      "step": 10288
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 2.9097875842205425e-05,
+      "loss": 0.8516,
+      "step": 10289
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 2.9095304222599395e-05,
+      "loss": 0.6775,
+      "step": 10290
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 2.9092732602993368e-05,
+      "loss": 0.7931,
+      "step": 10291
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 2.9090160983387337e-05,
+      "loss": 1.1155,
+      "step": 10292
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 2.908758936378131e-05,
+      "loss": 0.8007,
+      "step": 10293
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 2.908501774417528e-05,
+      "loss": 0.9857,
+      "step": 10294
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 2.9082446124569257e-05,
+      "loss": 0.8361,
+      "step": 10295
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 2.907987450496323e-05,
+      "loss": 1.0254,
+      "step": 10296
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 2.90773028853572e-05,
+      "loss": 0.6548,
+      "step": 10297
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 2.907473126575117e-05,
+      "loss": 0.8303,
+      "step": 10298
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 2.9072159646145142e-05,
+      "loss": 0.7673,
+      "step": 10299
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 2.906958802653912e-05,
+      "loss": 0.9171,
+      "step": 10300
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 2.9067016406933088e-05,
+      "loss": 0.9482,
+      "step": 10301
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 2.906444478732706e-05,
+      "loss": 0.831,
+      "step": 10302
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 2.906187316772103e-05,
+      "loss": 1.049,
+      "step": 10303
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 2.9059301548115004e-05,
+      "loss": 0.8992,
+      "step": 10304
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 2.9056729928508974e-05,
+      "loss": 0.8839,
+      "step": 10305
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 2.905415830890295e-05,
+      "loss": 0.8644,
+      "step": 10306
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 2.9051586689296923e-05,
+      "loss": 1.0023,
+      "step": 10307
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 2.9049015069690893e-05,
+      "loss": 0.9197,
+      "step": 10308
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 2.9046443450084866e-05,
+      "loss": 0.455,
+      "step": 10309
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 2.9043871830478836e-05,
+      "loss": 0.7082,
+      "step": 10310
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 2.904130021087281e-05,
+      "loss": 0.7396,
+      "step": 10311
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 2.9038728591266785e-05,
+      "loss": 0.7462,
+      "step": 10312
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 2.9036156971660755e-05,
+      "loss": 0.7257,
+      "step": 10313
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 2.9033585352054725e-05,
+      "loss": 1.0149,
+      "step": 10314
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 2.9031013732448698e-05,
+      "loss": 1.0868,
+      "step": 10315
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 2.9028442112842667e-05,
+      "loss": 0.896,
+      "step": 10316
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 2.902587049323664e-05,
+      "loss": 1.0052,
+      "step": 10317
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 2.9023298873630617e-05,
+      "loss": 0.861,
+      "step": 10318
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 2.9020727254024587e-05,
+      "loss": 0.7493,
+      "step": 10319
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 2.901815563441856e-05,
+      "loss": 0.8575,
+      "step": 10320
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 2.901558401481253e-05,
+      "loss": 0.7621,
+      "step": 10321
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 2.9013012395206502e-05,
+      "loss": 0.7656,
+      "step": 10322
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 2.9010440775600472e-05,
+      "loss": 0.5898,
+      "step": 10323
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 2.900786915599445e-05,
+      "loss": 0.8207,
+      "step": 10324
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 2.900529753638842e-05,
+      "loss": 0.9813,
+      "step": 10325
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 2.900272591678239e-05,
+      "loss": 0.8217,
+      "step": 10326
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 2.9000154297176364e-05,
+      "loss": 0.8429,
+      "step": 10327
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 2.8997582677570334e-05,
+      "loss": 0.6865,
+      "step": 10328
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 2.8995011057964304e-05,
+      "loss": 0.5062,
+      "step": 10329
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 2.8992439438358284e-05,
+      "loss": 0.83,
+      "step": 10330
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 2.8989867818752253e-05,
+      "loss": 0.5797,
+      "step": 10331
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 2.8987296199146223e-05,
+      "loss": 0.9641,
+      "step": 10332
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 2.8984724579540196e-05,
+      "loss": 0.8367,
+      "step": 10333
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 2.8982152959934166e-05,
+      "loss": 0.8756,
+      "step": 10334
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 2.897958134032814e-05,
+      "loss": 1.3461,
+      "step": 10335
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 2.8977009720722115e-05,
+      "loss": 0.557,
+      "step": 10336
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 2.8974438101116085e-05,
+      "loss": 0.7502,
+      "step": 10337
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 2.8971866481510058e-05,
+      "loss": 0.9119,
+      "step": 10338
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 2.8969294861904028e-05,
+      "loss": 0.9058,
+      "step": 10339
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 2.8966723242298e-05,
+      "loss": 0.6467,
+      "step": 10340
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 2.896415162269197e-05,
+      "loss": 0.9813,
+      "step": 10341
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 2.8961580003085947e-05,
+      "loss": 0.7924,
+      "step": 10342
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 2.895900838347992e-05,
+      "loss": 0.8865,
+      "step": 10343
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 2.895643676387389e-05,
+      "loss": 0.6649,
+      "step": 10344
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 2.895386514426786e-05,
+      "loss": 0.9315,
+      "step": 10345
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 2.8951293524661832e-05,
+      "loss": 0.7861,
+      "step": 10346
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 2.8948721905055802e-05,
+      "loss": 0.8377,
+      "step": 10347
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 2.894615028544978e-05,
+      "loss": 0.853,
+      "step": 10348
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 2.8943578665843752e-05,
+      "loss": 1.0885,
+      "step": 10349
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 2.894100704623772e-05,
+      "loss": 1.0096,
+      "step": 10350
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 2.8938435426631694e-05,
+      "loss": 0.7696,
+      "step": 10351
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 2.8935863807025664e-05,
+      "loss": 0.8375,
+      "step": 10352
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 2.8933292187419637e-05,
+      "loss": 0.9262,
+      "step": 10353
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 2.8930720567813614e-05,
+      "loss": 0.5794,
+      "step": 10354
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 2.8928148948207583e-05,
+      "loss": 0.5485,
+      "step": 10355
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 2.8925577328601556e-05,
+      "loss": 0.5322,
+      "step": 10356
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 2.8923005708995526e-05,
+      "loss": 0.637,
+      "step": 10357
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 2.89204340893895e-05,
+      "loss": 1.076,
+      "step": 10358
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 2.891786246978347e-05,
+      "loss": 0.5315,
+      "step": 10359
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 2.8915290850177445e-05,
+      "loss": 0.5754,
+      "step": 10360
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 2.891271923057142e-05,
+      "loss": 1.0501,
+      "step": 10361
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 2.8910147610965388e-05,
+      "loss": 0.8428,
+      "step": 10362
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 2.8907575991359358e-05,
+      "loss": 1.0788,
+      "step": 10363
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 2.890500437175333e-05,
+      "loss": 0.904,
+      "step": 10364
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 2.89024327521473e-05,
+      "loss": 0.5504,
+      "step": 10365
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 2.8899861132541277e-05,
+      "loss": 0.8499,
+      "step": 10366
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 2.889728951293525e-05,
+      "loss": 0.6772,
+      "step": 10367
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 2.889471789332922e-05,
+      "loss": 0.8975,
+      "step": 10368
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 2.8892146273723193e-05,
+      "loss": 0.7877,
+      "step": 10369
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 2.8889574654117163e-05,
+      "loss": 0.6861,
+      "step": 10370
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 2.8887003034511136e-05,
+      "loss": 0.8347,
+      "step": 10371
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 2.8884431414905112e-05,
+      "loss": 0.6942,
+      "step": 10372
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 2.8881859795299082e-05,
+      "loss": 0.6959,
+      "step": 10373
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 2.8879288175693055e-05,
+      "loss": 0.9102,
+      "step": 10374
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 2.8876716556087025e-05,
+      "loss": 0.6772,
+      "step": 10375
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 2.8874144936480994e-05,
+      "loss": 0.7915,
+      "step": 10376
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 2.8871573316874967e-05,
+      "loss": 0.8087,
+      "step": 10377
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 2.8869001697268944e-05,
+      "loss": 0.5164,
+      "step": 10378
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 2.8866430077662913e-05,
+      "loss": 0.7536,
+      "step": 10379
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 2.8863858458056886e-05,
+      "loss": 0.5616,
+      "step": 10380
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 2.8861286838450856e-05,
+      "loss": 0.8615,
+      "step": 10381
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 2.885871521884483e-05,
+      "loss": 0.8019,
+      "step": 10382
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 2.88561435992388e-05,
+      "loss": 0.7576,
+      "step": 10383
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 2.8853571979632775e-05,
+      "loss": 0.9199,
+      "step": 10384
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 2.885100036002675e-05,
+      "loss": 1.1991,
+      "step": 10385
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 2.8848428740420718e-05,
+      "loss": 0.753,
+      "step": 10386
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 2.884585712081469e-05,
+      "loss": 0.8569,
+      "step": 10387
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 2.884328550120866e-05,
+      "loss": 0.8765,
+      "step": 10388
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 2.8840713881602634e-05,
+      "loss": 0.6765,
+      "step": 10389
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 2.883814226199661e-05,
+      "loss": 0.5939,
+      "step": 10390
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 2.883557064239058e-05,
+      "loss": 0.6272,
+      "step": 10391
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 2.883299902278455e-05,
+      "loss": 0.8737,
+      "step": 10392
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 2.8830427403178523e-05,
+      "loss": 0.811,
+      "step": 10393
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 2.8827855783572493e-05,
+      "loss": 0.6528,
+      "step": 10394
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 2.8825284163966466e-05,
+      "loss": 0.6526,
+      "step": 10395
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 2.8822712544360442e-05,
+      "loss": 0.7505,
+      "step": 10396
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 2.8820140924754412e-05,
+      "loss": 0.7163,
+      "step": 10397
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 2.8817569305148385e-05,
+      "loss": 0.6713,
+      "step": 10398
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 2.8814997685542355e-05,
+      "loss": 0.8438,
+      "step": 10399
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 2.8812426065936328e-05,
+      "loss": 0.7577,
+      "step": 10400
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 2.8809854446330297e-05,
+      "loss": 0.6784,
+      "step": 10401
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 2.8807282826724274e-05,
+      "loss": 0.4333,
+      "step": 10402
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 2.8804711207118247e-05,
+      "loss": 0.5812,
+      "step": 10403
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 2.8802139587512217e-05,
+      "loss": 0.4673,
+      "step": 10404
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 2.879956796790619e-05,
+      "loss": 0.5444,
+      "step": 10405
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 2.879699634830016e-05,
+      "loss": 0.6192,
+      "step": 10406
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 2.879442472869413e-05,
+      "loss": 0.8057,
+      "step": 10407
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 2.879185310908811e-05,
+      "loss": 1.0505,
+      "step": 10408
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 2.878928148948208e-05,
+      "loss": 0.8094,
+      "step": 10409
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 2.8786709869876048e-05,
+      "loss": 0.7984,
+      "step": 10410
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 2.878413825027002e-05,
+      "loss": 0.6723,
+      "step": 10411
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 2.878156663066399e-05,
+      "loss": 1.0229,
+      "step": 10412
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 2.8778995011057964e-05,
+      "loss": 0.6333,
+      "step": 10413
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 2.877642339145194e-05,
+      "loss": 0.8619,
+      "step": 10414
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 2.877385177184591e-05,
+      "loss": 0.9863,
+      "step": 10415
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 2.8771280152239883e-05,
+      "loss": 0.4322,
+      "step": 10416
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 2.8768708532633853e-05,
+      "loss": 0.9186,
+      "step": 10417
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 2.8766136913027826e-05,
+      "loss": 0.8619,
+      "step": 10418
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 2.8763565293421796e-05,
+      "loss": 0.6613,
+      "step": 10419
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 2.8760993673815772e-05,
+      "loss": 0.6188,
+      "step": 10420
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 2.8758422054209745e-05,
+      "loss": 0.858,
+      "step": 10421
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 2.8755850434603715e-05,
+      "loss": 0.7821,
+      "step": 10422
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 2.8753278814997685e-05,
+      "loss": 0.7479,
+      "step": 10423
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 2.8750707195391658e-05,
+      "loss": 0.7015,
+      "step": 10424
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 2.8748135575785627e-05,
+      "loss": 0.6976,
+      "step": 10425
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 2.8745563956179604e-05,
+      "loss": 0.6889,
+      "step": 10426
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 2.8742992336573577e-05,
+      "loss": 0.7789,
+      "step": 10427
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 2.8740420716967547e-05,
+      "loss": 0.8481,
+      "step": 10428
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 2.873784909736152e-05,
+      "loss": 0.6257,
+      "step": 10429
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 2.873527747775549e-05,
+      "loss": 0.8142,
+      "step": 10430
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 2.8732705858149462e-05,
+      "loss": 0.5795,
+      "step": 10431
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 2.873013423854344e-05,
+      "loss": 0.7627,
+      "step": 10432
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 2.872756261893741e-05,
+      "loss": 0.8195,
+      "step": 10433
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 2.872499099933138e-05,
+      "loss": 1.2402,
+      "step": 10434
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 2.872241937972535e-05,
+      "loss": 0.6325,
+      "step": 10435
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 2.8719847760119324e-05,
+      "loss": 0.8159,
+      "step": 10436
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 2.8717276140513294e-05,
+      "loss": 0.3676,
+      "step": 10437
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 2.871470452090727e-05,
+      "loss": 0.7382,
+      "step": 10438
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 2.871213290130124e-05,
+      "loss": 0.7759,
+      "step": 10439
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 2.8709561281695213e-05,
+      "loss": 0.7515,
+      "step": 10440
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 2.8706989662089183e-05,
+      "loss": 0.686,
+      "step": 10441
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 2.8704418042483156e-05,
+      "loss": 0.7592,
+      "step": 10442
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 2.8701846422877126e-05,
+      "loss": 0.5764,
+      "step": 10443
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 2.8699274803271102e-05,
+      "loss": 0.7453,
+      "step": 10444
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 2.8696703183665075e-05,
+      "loss": 1.2262,
+      "step": 10445
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 2.8694131564059045e-05,
+      "loss": 0.8455,
+      "step": 10446
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 2.8691559944453018e-05,
+      "loss": 0.8551,
+      "step": 10447
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 2.8688988324846988e-05,
+      "loss": 0.5396,
+      "step": 10448
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 2.868641670524096e-05,
+      "loss": 0.6002,
+      "step": 10449
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 2.8683845085634937e-05,
+      "loss": 0.8885,
+      "step": 10450
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 2.8681273466028907e-05,
+      "loss": 0.6389,
+      "step": 10451
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 2.867870184642288e-05,
+      "loss": 0.5354,
+      "step": 10452
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 2.867613022681685e-05,
+      "loss": 0.8372,
+      "step": 10453
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 2.867355860721082e-05,
+      "loss": 0.9006,
+      "step": 10454
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 2.86709869876048e-05,
+      "loss": 1.3055,
+      "step": 10455
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 2.866841536799877e-05,
+      "loss": 0.7541,
+      "step": 10456
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 2.866584374839274e-05,
+      "loss": 0.8437,
+      "step": 10457
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 2.866327212878671e-05,
+      "loss": 0.721,
+      "step": 10458
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 2.866070050918068e-05,
+      "loss": 1.0202,
+      "step": 10459
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 2.8658128889574654e-05,
+      "loss": 1.0489,
+      "step": 10460
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 2.865555726996863e-05,
+      "loss": 0.7109,
+      "step": 10461
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 2.86529856503626e-05,
+      "loss": 0.7226,
+      "step": 10462
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 2.8650414030756574e-05,
+      "loss": 0.4703,
+      "step": 10463
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 2.8647842411150543e-05,
+      "loss": 0.6418,
+      "step": 10464
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 2.8645270791544516e-05,
+      "loss": 0.8732,
+      "step": 10465
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 2.8642699171938486e-05,
+      "loss": 0.8459,
+      "step": 10466
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 2.8640127552332463e-05,
+      "loss": 0.684,
+      "step": 10467
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 2.8637555932726436e-05,
+      "loss": 1.1534,
+      "step": 10468
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 2.8634984313120405e-05,
+      "loss": 0.6017,
+      "step": 10469
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 2.8632412693514375e-05,
+      "loss": 0.9685,
+      "step": 10470
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 2.8629841073908348e-05,
+      "loss": 0.7989,
+      "step": 10471
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 2.8627269454302318e-05,
+      "loss": 0.8187,
+      "step": 10472
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 2.8624697834696294e-05,
+      "loss": 0.7982,
+      "step": 10473
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 2.8622126215090267e-05,
+      "loss": 0.6642,
+      "step": 10474
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 2.8619554595484237e-05,
+      "loss": 0.6092,
+      "step": 10475
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 2.861698297587821e-05,
+      "loss": 0.5584,
+      "step": 10476
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 2.861441135627218e-05,
+      "loss": 0.7455,
+      "step": 10477
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 2.8611839736666153e-05,
+      "loss": 0.7775,
+      "step": 10478
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 2.860926811706013e-05,
+      "loss": 0.837,
+      "step": 10479
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 2.86066964974541e-05,
+      "loss": 1.0086,
+      "step": 10480
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 2.8604124877848072e-05,
+      "loss": 0.8529,
+      "step": 10481
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 2.8601553258242042e-05,
+      "loss": 0.8905,
+      "step": 10482
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 2.8598981638636015e-05,
+      "loss": 0.8617,
+      "step": 10483
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 2.8596410019029984e-05,
+      "loss": 0.4417,
+      "step": 10484
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 2.859383839942396e-05,
+      "loss": 0.6636,
+      "step": 10485
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 2.8591266779817934e-05,
+      "loss": 0.7493,
+      "step": 10486
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 2.8588695160211904e-05,
+      "loss": 1.0344,
+      "step": 10487
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 2.8586123540605873e-05,
+      "loss": 0.7067,
+      "step": 10488
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 2.8583551920999846e-05,
+      "loss": 0.7785,
+      "step": 10489
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 2.8580980301393816e-05,
+      "loss": 0.6127,
+      "step": 10490
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 2.8578408681787793e-05,
+      "loss": 0.4587,
+      "step": 10491
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 2.8575837062181766e-05,
+      "loss": 0.9608,
+      "step": 10492
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 2.8573265442575735e-05,
+      "loss": 0.8816,
+      "step": 10493
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 2.857069382296971e-05,
+      "loss": 0.7952,
+      "step": 10494
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 2.8568122203363678e-05,
+      "loss": 0.8046,
+      "step": 10495
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 2.856555058375765e-05,
+      "loss": 0.8921,
+      "step": 10496
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 2.8562978964151628e-05,
+      "loss": 1.0959,
+      "step": 10497
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 2.8560407344545597e-05,
+      "loss": 1.0523,
+      "step": 10498
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 2.855783572493957e-05,
+      "loss": 0.7894,
+      "step": 10499
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 2.855526410533354e-05,
+      "loss": 0.6945,
+      "step": 10500
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 2.855269248572751e-05,
+      "loss": 0.4223,
+      "step": 10501
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 2.8550120866121483e-05,
+      "loss": 0.8332,
+      "step": 10502
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 2.854754924651546e-05,
+      "loss": 0.6515,
+      "step": 10503
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 2.854497762690943e-05,
+      "loss": 0.5855,
+      "step": 10504
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 2.8542406007303402e-05,
+      "loss": 0.8741,
+      "step": 10505
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 2.8539834387697372e-05,
+      "loss": 0.8347,
+      "step": 10506
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 2.8537262768091345e-05,
+      "loss": 0.8678,
+      "step": 10507
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 2.8534691148485315e-05,
+      "loss": 0.7771,
+      "step": 10508
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 2.853211952887929e-05,
+      "loss": 1.053,
+      "step": 10509
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 2.8529547909273264e-05,
+      "loss": 0.6986,
+      "step": 10510
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 2.8526976289667234e-05,
+      "loss": 0.97,
+      "step": 10511
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 2.8524404670061207e-05,
+      "loss": 0.6931,
+      "step": 10512
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 2.8521833050455177e-05,
+      "loss": 0.893,
+      "step": 10513
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 2.851926143084915e-05,
+      "loss": 0.5152,
+      "step": 10514
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 2.8516689811243126e-05,
+      "loss": 0.619,
+      "step": 10515
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 2.8514118191637096e-05,
+      "loss": 0.8868,
+      "step": 10516
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 2.8511546572031065e-05,
+      "loss": 0.7546,
+      "step": 10517
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 2.850897495242504e-05,
+      "loss": 0.8943,
+      "step": 10518
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 2.8506403332819008e-05,
+      "loss": 0.7195,
+      "step": 10519
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 2.850383171321298e-05,
+      "loss": 0.8948,
+      "step": 10520
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 2.8501260093606958e-05,
+      "loss": 0.9482,
+      "step": 10521
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 2.8498688474000927e-05,
+      "loss": 0.9337,
+      "step": 10522
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 2.84961168543949e-05,
+      "loss": 0.8752,
+      "step": 10523
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 2.849354523478887e-05,
+      "loss": 1.0832,
+      "step": 10524
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 2.8490973615182843e-05,
+      "loss": 0.8814,
+      "step": 10525
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 2.8488401995576813e-05,
+      "loss": 0.5885,
+      "step": 10526
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 2.848583037597079e-05,
+      "loss": 0.888,
+      "step": 10527
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 2.8483258756364762e-05,
+      "loss": 0.908,
+      "step": 10528
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 2.8480687136758732e-05,
+      "loss": 0.7406,
+      "step": 10529
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 2.8478115517152705e-05,
+      "loss": 0.7819,
+      "step": 10530
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 2.8475543897546675e-05,
+      "loss": 0.7041,
+      "step": 10531
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 2.8472972277940645e-05,
+      "loss": 0.8805,
+      "step": 10532
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 2.8470400658334624e-05,
+      "loss": 0.8267,
+      "step": 10533
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 2.8467829038728594e-05,
+      "loss": 0.4705,
+      "step": 10534
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 2.8465257419122564e-05,
+      "loss": 1.2649,
+      "step": 10535
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 2.8462685799516537e-05,
+      "loss": 0.7373,
+      "step": 10536
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 2.8460114179910507e-05,
+      "loss": 0.812,
+      "step": 10537
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 2.845754256030448e-05,
+      "loss": 0.9404,
+      "step": 10538
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 2.8454970940698456e-05,
+      "loss": 1.2581,
+      "step": 10539
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 2.8452399321092426e-05,
+      "loss": 0.8093,
+      "step": 10540
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 2.84498277014864e-05,
+      "loss": 0.8441,
+      "step": 10541
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 2.844725608188037e-05,
+      "loss": 0.6156,
+      "step": 10542
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 2.844468446227434e-05,
+      "loss": 0.6014,
+      "step": 10543
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 2.844211284266831e-05,
+      "loss": 0.7377,
+      "step": 10544
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 2.8439541223062288e-05,
+      "loss": 0.8767,
+      "step": 10545
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 2.843696960345626e-05,
+      "loss": 0.8147,
+      "step": 10546
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 2.843439798385023e-05,
+      "loss": 0.8629,
+      "step": 10547
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 2.84318263642442e-05,
+      "loss": 0.7415,
+      "step": 10548
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 2.8429254744638173e-05,
+      "loss": 0.9802,
+      "step": 10549
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 2.8426683125032143e-05,
+      "loss": 0.7677,
+      "step": 10550
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 2.842411150542612e-05,
+      "loss": 0.6227,
+      "step": 10551
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 2.8421539885820092e-05,
+      "loss": 0.7788,
+      "step": 10552
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 2.8418968266214062e-05,
+      "loss": 0.4369,
+      "step": 10553
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 2.8416396646608035e-05,
+      "loss": 0.9425,
+      "step": 10554
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 2.8413825027002005e-05,
+      "loss": 1.0699,
+      "step": 10555
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 2.8411253407395978e-05,
+      "loss": 1.2425,
+      "step": 10556
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 2.8408681787789954e-05,
+      "loss": 0.5027,
+      "step": 10557
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 2.8406110168183924e-05,
+      "loss": 0.7737,
+      "step": 10558
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 2.8403538548577897e-05,
+      "loss": 0.531,
+      "step": 10559
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.8400966928971867e-05,
+      "loss": 0.6373,
+      "step": 10560
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.839839530936584e-05,
+      "loss": 0.7586,
+      "step": 10561
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.839582368975981e-05,
+      "loss": 1.1198,
+      "step": 10562
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.8393252070153786e-05,
+      "loss": 0.972,
+      "step": 10563
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.8390680450547756e-05,
+      "loss": 0.8469,
+      "step": 10564
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.838810883094173e-05,
+      "loss": 0.7733,
+      "step": 10565
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.83855372113357e-05,
+      "loss": 0.7143,
+      "step": 10566
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.838296559172967e-05,
+      "loss": 0.7632,
+      "step": 10567
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.838039397212364e-05,
+      "loss": 0.783,
+      "step": 10568
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.8377822352517618e-05,
+      "loss": 0.8596,
+      "step": 10569
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.837525073291159e-05,
+      "loss": 0.809,
+      "step": 10570
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.837267911330556e-05,
+      "loss": 0.6135,
+      "step": 10571
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.8370107493699534e-05,
+      "loss": 0.5253,
+      "step": 10572
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.8367535874093503e-05,
+      "loss": 1.1703,
+      "step": 10573
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.8364964254487476e-05,
+      "loss": 0.6334,
+      "step": 10574
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.8362392634881453e-05,
+      "loss": 0.8691,
+      "step": 10575
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.8359821015275423e-05,
+      "loss": 0.9118,
+      "step": 10576
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.8357249395669396e-05,
+      "loss": 0.841,
+      "step": 10577
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.8354677776063365e-05,
+      "loss": 0.9213,
+      "step": 10578
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.8352106156457335e-05,
+      "loss": 1.0419,
+      "step": 10579
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.8349534536851308e-05,
+      "loss": 0.7876,
+      "step": 10580
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.8346962917245285e-05,
+      "loss": 0.6078,
+      "step": 10581
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.8344391297639254e-05,
+      "loss": 1.1109,
+      "step": 10582
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.8341819678033227e-05,
+      "loss": 0.6193,
+      "step": 10583
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.8339248058427197e-05,
+      "loss": 0.7815,
+      "step": 10584
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.833667643882117e-05,
+      "loss": 0.6776,
+      "step": 10585
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.833410481921514e-05,
+      "loss": 0.4706,
+      "step": 10586
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.8331533199609116e-05,
+      "loss": 0.9086,
+      "step": 10587
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.832896158000309e-05,
+      "loss": 0.7644,
+      "step": 10588
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.832638996039706e-05,
+      "loss": 0.7479,
+      "step": 10589
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.8323818340791032e-05,
+      "loss": 0.5114,
+      "step": 10590
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.8321246721185002e-05,
+      "loss": 0.5555,
+      "step": 10591
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.831867510157897e-05,
+      "loss": 0.7334,
+      "step": 10592
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.831610348197295e-05,
+      "loss": 0.7681,
+      "step": 10593
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.831353186236692e-05,
+      "loss": 0.8406,
+      "step": 10594
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.831096024276089e-05,
+      "loss": 0.5707,
+      "step": 10595
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.8308388623154864e-05,
+      "loss": 1.0509,
+      "step": 10596
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.8305817003548833e-05,
+      "loss": 1.0922,
+      "step": 10597
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.8303245383942806e-05,
+      "loss": 0.6265,
+      "step": 10598
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.8300673764336783e-05,
+      "loss": 0.7628,
+      "step": 10599
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.8298102144730753e-05,
+      "loss": 0.7878,
+      "step": 10600
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.8295530525124726e-05,
+      "loss": 1.1349,
+      "step": 10601
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.8292958905518695e-05,
+      "loss": 0.5346,
+      "step": 10602
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.829038728591267e-05,
+      "loss": 1.1294,
+      "step": 10603
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.8287815666306645e-05,
+      "loss": 0.6673,
+      "step": 10604
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.8285244046700615e-05,
+      "loss": 0.5846,
+      "step": 10605
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.8282672427094588e-05,
+      "loss": 0.6959,
+      "step": 10606
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.8280100807488557e-05,
+      "loss": 0.6159,
+      "step": 10607
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.827752918788253e-05,
+      "loss": 1.2218,
+      "step": 10608
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.82749575682765e-05,
+      "loss": 0.525,
+      "step": 10609
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.8272385948670477e-05,
+      "loss": 0.7621,
+      "step": 10610
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.826981432906445e-05,
+      "loss": 0.6343,
+      "step": 10611
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.826724270945842e-05,
+      "loss": 0.7911,
+      "step": 10612
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.826467108985239e-05,
+      "loss": 0.8473,
+      "step": 10613
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 2.8262099470246362e-05,
+      "loss": 0.6805,
+      "step": 10614
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 2.8259527850640332e-05,
+      "loss": 0.8845,
+      "step": 10615
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 2.8256956231034308e-05,
+      "loss": 1.1597,
+      "step": 10616
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 2.825438461142828e-05,
+      "loss": 0.6689,
+      "step": 10617
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 2.825181299182225e-05,
+      "loss": 0.7182,
+      "step": 10618
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 2.8249241372216224e-05,
+      "loss": 0.8,
+      "step": 10619
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 2.8246669752610194e-05,
+      "loss": 0.6936,
+      "step": 10620
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 2.8244098133004167e-05,
+      "loss": 0.8121,
+      "step": 10621
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 2.8241526513398143e-05,
+      "loss": 0.7133,
+      "step": 10622
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 2.8238954893792113e-05,
+      "loss": 0.7464,
+      "step": 10623
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 2.8236383274186086e-05,
+      "loss": 0.8691,
+      "step": 10624
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 2.8233811654580056e-05,
+      "loss": 0.8934,
+      "step": 10625
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 2.8231240034974025e-05,
+      "loss": 1.2596,
+      "step": 10626
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 2.8228668415368e-05,
+      "loss": 0.3824,
+      "step": 10627
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 2.8226096795761975e-05,
+      "loss": 0.8664,
+      "step": 10628
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 2.8223525176155945e-05,
+      "loss": 0.8492,
+      "step": 10629
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 2.8220953556549918e-05,
+      "loss": 1.1261,
+      "step": 10630
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 2.8218381936943887e-05,
+      "loss": 0.7914,
+      "step": 10631
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 2.821581031733786e-05,
+      "loss": 0.7695,
+      "step": 10632
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 2.821323869773183e-05,
+      "loss": 0.8413,
+      "step": 10633
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 2.8210667078125807e-05,
+      "loss": 0.7448,
+      "step": 10634
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 2.820809545851978e-05,
+      "loss": 0.6288,
+      "step": 10635
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 2.820552383891375e-05,
+      "loss": 0.6755,
+      "step": 10636
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 2.8202952219307722e-05,
+      "loss": 0.949,
+      "step": 10637
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 2.8200380599701692e-05,
+      "loss": 0.8835,
+      "step": 10638
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 2.8197808980095665e-05,
+      "loss": 0.9744,
+      "step": 10639
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 2.819523736048964e-05,
+      "loss": 0.6658,
+      "step": 10640
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 2.819266574088361e-05,
+      "loss": 0.7731,
+      "step": 10641
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 2.819009412127758e-05,
+      "loss": 0.8423,
+      "step": 10642
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 2.8187522501671554e-05,
+      "loss": 0.8743,
+      "step": 10643
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 2.8184950882065524e-05,
+      "loss": 0.6099,
+      "step": 10644
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 2.8182379262459497e-05,
+      "loss": 1.1329,
+      "step": 10645
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 2.8179807642853473e-05,
+      "loss": 0.7079,
+      "step": 10646
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 2.8177236023247443e-05,
+      "loss": 0.7761,
+      "step": 10647
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 2.8174664403641416e-05,
+      "loss": 0.5892,
+      "step": 10648
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 2.8172092784035386e-05,
+      "loss": 0.6697,
+      "step": 10649
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 2.816952116442936e-05,
+      "loss": 0.983,
+      "step": 10650
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 2.816694954482333e-05,
+      "loss": 0.7034,
+      "step": 10651
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 2.8164377925217305e-05,
+      "loss": 0.9495,
+      "step": 10652
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 2.8161806305611278e-05,
+      "loss": 0.5834,
+      "step": 10653
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 2.8159234686005248e-05,
+      "loss": 0.9894,
+      "step": 10654
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 2.815666306639922e-05,
+      "loss": 0.9483,
+      "step": 10655
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 2.815409144679319e-05,
+      "loss": 1.0011,
+      "step": 10656
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 2.815151982718716e-05,
+      "loss": 0.9294,
+      "step": 10657
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 2.814894820758114e-05,
+      "loss": 0.638,
+      "step": 10658
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 2.814637658797511e-05,
+      "loss": 1.0078,
+      "step": 10659
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 2.814380496836908e-05,
+      "loss": 0.7848,
+      "step": 10660
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 2.8141233348763052e-05,
+      "loss": 0.9901,
+      "step": 10661
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 2.8138661729157022e-05,
+      "loss": 0.7386,
+      "step": 10662
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 2.8136090109550995e-05,
+      "loss": 1.1298,
+      "step": 10663
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 2.813351848994497e-05,
+      "loss": 0.8704,
+      "step": 10664
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 2.813094687033894e-05,
+      "loss": 0.6875,
+      "step": 10665
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 2.8128375250732914e-05,
+      "loss": 0.5836,
+      "step": 10666
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 2.8125803631126884e-05,
+      "loss": 0.8487,
+      "step": 10667
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 2.8123232011520857e-05,
+      "loss": 1.1507,
+      "step": 10668
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 2.8120660391914827e-05,
+      "loss": 0.7267,
+      "step": 10669
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 2.8118088772308803e-05,
+      "loss": 0.7113,
+      "step": 10670
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 2.8115517152702776e-05,
+      "loss": 1.3084,
+      "step": 10671
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 2.8112945533096746e-05,
+      "loss": 0.9555,
+      "step": 10672
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 2.8110373913490716e-05,
+      "loss": 0.6104,
+      "step": 10673
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 2.810780229388469e-05,
+      "loss": 0.3966,
+      "step": 10674
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 2.810523067427866e-05,
+      "loss": 0.9497,
+      "step": 10675
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 2.8102659054672635e-05,
+      "loss": 1.0578,
+      "step": 10676
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 2.8100087435066608e-05,
+      "loss": 0.9237,
+      "step": 10677
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 2.8097515815460578e-05,
+      "loss": 0.7673,
+      "step": 10678
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 2.809494419585455e-05,
+      "loss": 0.8266,
+      "step": 10679
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 2.809237257624852e-05,
+      "loss": 0.9873,
+      "step": 10680
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 2.8089800956642494e-05,
+      "loss": 0.6681,
+      "step": 10681
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 2.808722933703647e-05,
+      "loss": 0.8178,
+      "step": 10682
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 2.808465771743044e-05,
+      "loss": 0.8732,
+      "step": 10683
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 2.8082086097824413e-05,
+      "loss": 0.79,
+      "step": 10684
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 2.8079514478218383e-05,
+      "loss": 0.7871,
+      "step": 10685
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 2.8076942858612356e-05,
+      "loss": 0.7483,
+      "step": 10686
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 2.8074371239006325e-05,
+      "loss": 0.6647,
+      "step": 10687
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 2.8071799619400302e-05,
+      "loss": 0.5758,
+      "step": 10688
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 2.806922799979427e-05,
+      "loss": 0.954,
+      "step": 10689
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 2.8066656380188244e-05,
+      "loss": 0.8354,
+      "step": 10690
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 2.8064084760582214e-05,
+      "loss": 0.8941,
+      "step": 10691
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 2.8061513140976187e-05,
+      "loss": 0.879,
+      "step": 10692
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 2.8058941521370157e-05,
+      "loss": 0.6597,
+      "step": 10693
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 2.8056369901764133e-05,
+      "loss": 0.7448,
+      "step": 10694
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 2.8053798282158106e-05,
+      "loss": 0.8003,
+      "step": 10695
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 2.8051226662552076e-05,
+      "loss": 0.8503,
+      "step": 10696
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 2.804865504294605e-05,
+      "loss": 1.0408,
+      "step": 10697
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 2.804608342334002e-05,
+      "loss": 0.7881,
+      "step": 10698
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 2.8043511803733992e-05,
+      "loss": 0.7578,
+      "step": 10699
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 2.804094018412797e-05,
+      "loss": 0.7906,
+      "step": 10700
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 2.8038368564521938e-05,
+      "loss": 0.3682,
+      "step": 10701
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 2.803579694491591e-05,
+      "loss": 0.4975,
+      "step": 10702
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 2.803322532530988e-05,
+      "loss": 0.9278,
+      "step": 10703
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 2.803065370570385e-05,
+      "loss": 0.9195,
+      "step": 10704
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 2.8028082086097824e-05,
+      "loss": 0.648,
+      "step": 10705
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 2.80255104664918e-05,
+      "loss": 0.8339,
+      "step": 10706
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 2.802293884688577e-05,
+      "loss": 0.8018,
+      "step": 10707
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 2.8020367227279743e-05,
+      "loss": 0.8661,
+      "step": 10708
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 2.8017795607673713e-05,
+      "loss": 0.8219,
+      "step": 10709
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 2.8015223988067686e-05,
+      "loss": 0.8117,
+      "step": 10710
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 2.8012652368461655e-05,
+      "loss": 0.8653,
+      "step": 10711
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 2.8010080748855632e-05,
+      "loss": 0.4956,
+      "step": 10712
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 2.8007509129249605e-05,
+      "loss": 0.8525,
+      "step": 10713
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 2.8004937509643575e-05,
+      "loss": 0.7441,
+      "step": 10714
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 2.8002365890037548e-05,
+      "loss": 0.8173,
+      "step": 10715
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 2.7999794270431517e-05,
+      "loss": 0.8328,
+      "step": 10716
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 2.799722265082549e-05,
+      "loss": 0.7644,
+      "step": 10717
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 2.7994651031219467e-05,
+      "loss": 0.7376,
+      "step": 10718
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 2.7992079411613437e-05,
+      "loss": 0.8178,
+      "step": 10719
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 2.7989507792007406e-05,
+      "loss": 0.7683,
+      "step": 10720
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 2.798693617240138e-05,
+      "loss": 1.0727,
+      "step": 10721
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 2.798436455279535e-05,
+      "loss": 0.5735,
+      "step": 10722
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 2.7981792933189322e-05,
+      "loss": 0.7104,
+      "step": 10723
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 2.79792213135833e-05,
+      "loss": 0.8632,
+      "step": 10724
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 2.7976649693977268e-05,
+      "loss": 0.8106,
+      "step": 10725
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 2.797407807437124e-05,
+      "loss": 0.7738,
+      "step": 10726
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 2.797150645476521e-05,
+      "loss": 0.8323,
+      "step": 10727
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 2.7968934835159184e-05,
+      "loss": 0.7459,
+      "step": 10728
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 2.7966363215553154e-05,
+      "loss": 0.8063,
+      "step": 10729
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 2.796379159594713e-05,
+      "loss": 0.6545,
+      "step": 10730
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 2.7961219976341103e-05,
+      "loss": 0.6448,
+      "step": 10731
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 2.7958648356735073e-05,
+      "loss": 0.6969,
+      "step": 10732
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 2.7956076737129046e-05,
+      "loss": 0.9046,
+      "step": 10733
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 2.7953505117523016e-05,
+      "loss": 0.8367,
+      "step": 10734
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 2.7950933497916985e-05,
+      "loss": 0.7655,
+      "step": 10735
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 2.7948361878310965e-05,
+      "loss": 0.4076,
+      "step": 10736
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 2.7945790258704935e-05,
+      "loss": 1.0742,
+      "step": 10737
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 2.7943218639098905e-05,
+      "loss": 0.744,
+      "step": 10738
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 2.7940647019492878e-05,
+      "loss": 0.8925,
+      "step": 10739
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 2.7938075399886847e-05,
+      "loss": 0.7502,
+      "step": 10740
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 2.793550378028082e-05,
+      "loss": 0.7699,
+      "step": 10741
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 2.7932932160674797e-05,
+      "loss": 0.514,
+      "step": 10742
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 2.7930360541068767e-05,
+      "loss": 0.8442,
+      "step": 10743
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 2.792778892146274e-05,
+      "loss": 0.796,
+      "step": 10744
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 2.792521730185671e-05,
+      "loss": 0.7149,
+      "step": 10745
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 2.7922645682250682e-05,
+      "loss": 0.5407,
+      "step": 10746
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 2.7920074062644652e-05,
+      "loss": 0.4953,
+      "step": 10747
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 2.791750244303863e-05,
+      "loss": 0.9454,
+      "step": 10748
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 2.79149308234326e-05,
+      "loss": 0.6667,
+      "step": 10749
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 2.791235920382657e-05,
+      "loss": 0.7489,
+      "step": 10750
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 2.790978758422054e-05,
+      "loss": 0.7424,
+      "step": 10751
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 2.7907215964614514e-05,
+      "loss": 0.7726,
+      "step": 10752
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 2.790464434500849e-05,
+      "loss": 0.5816,
+      "step": 10753
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 2.790207272540246e-05,
+      "loss": 0.9778,
+      "step": 10754
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 2.7899501105796433e-05,
+      "loss": 0.7095,
+      "step": 10755
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 2.7896929486190403e-05,
+      "loss": 0.8353,
+      "step": 10756
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 2.7894357866584376e-05,
+      "loss": 0.7836,
+      "step": 10757
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 2.7891786246978346e-05,
+      "loss": 0.8278,
+      "step": 10758
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 2.7889214627372322e-05,
+      "loss": 0.7825,
+      "step": 10759
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 2.7886643007766295e-05,
+      "loss": 0.6088,
+      "step": 10760
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 2.7884071388160265e-05,
+      "loss": 0.9393,
+      "step": 10761
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 2.7881499768554238e-05,
+      "loss": 0.7065,
+      "step": 10762
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 2.7878928148948208e-05,
+      "loss": 0.8368,
+      "step": 10763
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 2.787635652934218e-05,
+      "loss": 1.2215,
+      "step": 10764
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 2.7873784909736157e-05,
+      "loss": 1.0376,
+      "step": 10765
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 2.7871213290130127e-05,
+      "loss": 0.859,
+      "step": 10766
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 2.7868641670524097e-05,
+      "loss": 0.6408,
+      "step": 10767
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 2.786607005091807e-05,
+      "loss": 0.5942,
+      "step": 10768
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 2.786349843131204e-05,
+      "loss": 0.8686,
+      "step": 10769
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 2.7860926811706012e-05,
+      "loss": 0.6699,
+      "step": 10770
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 2.785835519209999e-05,
+      "loss": 0.7183,
+      "step": 10771
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 2.785578357249396e-05,
+      "loss": 0.9663,
+      "step": 10772
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 2.785321195288793e-05,
+      "loss": 0.4946,
+      "step": 10773
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 2.78506403332819e-05,
+      "loss": 0.8017,
+      "step": 10774
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 2.7848068713675874e-05,
+      "loss": 0.5988,
+      "step": 10775
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 2.7845497094069844e-05,
+      "loss": 0.6125,
+      "step": 10776
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 2.784292547446382e-05,
+      "loss": 0.6314,
+      "step": 10777
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 2.7840353854857794e-05,
+      "loss": 0.809,
+      "step": 10778
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 2.7837782235251763e-05,
+      "loss": 0.7119,
+      "step": 10779
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 2.7835210615645736e-05,
+      "loss": 0.6515,
+      "step": 10780
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 2.7832638996039706e-05,
+      "loss": 0.6796,
+      "step": 10781
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 2.7830067376433676e-05,
+      "loss": 0.7794,
+      "step": 10782
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 2.7827495756827656e-05,
+      "loss": 0.5262,
+      "step": 10783
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 2.7824924137221625e-05,
+      "loss": 1.0091,
+      "step": 10784
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 2.7822352517615595e-05,
+      "loss": 0.861,
+      "step": 10785
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 2.7819780898009568e-05,
+      "loss": 0.6077,
+      "step": 10786
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 2.7817209278403538e-05,
+      "loss": 0.6098,
+      "step": 10787
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 2.781463765879751e-05,
+      "loss": 0.9084,
+      "step": 10788
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 2.7812066039191487e-05,
+      "loss": 0.682,
+      "step": 10789
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 2.7809494419585457e-05,
+      "loss": 0.745,
+      "step": 10790
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 2.780692279997943e-05,
+      "loss": 0.8515,
+      "step": 10791
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 2.78043511803734e-05,
+      "loss": 0.653,
+      "step": 10792
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 2.7801779560767373e-05,
+      "loss": 0.657,
+      "step": 10793
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 2.7799207941161342e-05,
+      "loss": 0.8583,
+      "step": 10794
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 2.779663632155532e-05,
+      "loss": 1.046,
+      "step": 10795
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 2.7794064701949292e-05,
+      "loss": 0.6274,
+      "step": 10796
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 2.7791493082343262e-05,
+      "loss": 0.6924,
+      "step": 10797
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 2.778892146273723e-05,
+      "loss": 0.7979,
+      "step": 10798
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 2.7786349843131204e-05,
+      "loss": 0.9388,
+      "step": 10799
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 2.7783778223525174e-05,
+      "loss": 0.7274,
+      "step": 10800
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 2.778120660391915e-05,
+      "loss": 0.7406,
+      "step": 10801
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 2.7778634984313124e-05,
+      "loss": 0.6503,
+      "step": 10802
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 2.7776063364707093e-05,
+      "loss": 0.6523,
+      "step": 10803
+    },
+    {
+      "epoch": 2.0,
+      "eval_accuracy": 0.7156023814069741,
+      "eval_loss": 0.828673779964447,
+      "eval_runtime": 40.5272,
+      "eval_samples_per_second": 1421.589,
+      "eval_steps_per_second": 44.439,
+      "step": 10803
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 2.7773491745101066e-05,
+      "loss": 0.6888,
+      "step": 10804
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 2.7770920125495036e-05,
+      "loss": 0.5712,
+      "step": 10805
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 2.776834850588901e-05,
+      "loss": 0.7416,
+      "step": 10806
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 2.7765776886282986e-05,
+      "loss": 0.7222,
+      "step": 10807
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 2.7763205266676955e-05,
+      "loss": 0.5802,
+      "step": 10808
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 2.776063364707093e-05,
+      "loss": 0.7688,
+      "step": 10809
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 2.7758062027464898e-05,
+      "loss": 0.6734,
+      "step": 10810
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 2.775549040785887e-05,
+      "loss": 0.6593,
+      "step": 10811
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 2.775291878825284e-05,
+      "loss": 0.6901,
+      "step": 10812
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 2.7750347168646817e-05,
+      "loss": 0.928,
+      "step": 10813
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 2.7747775549040787e-05,
+      "loss": 0.796,
+      "step": 10814
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 2.774520392943476e-05,
+      "loss": 0.6771,
+      "step": 10815
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 2.774263230982873e-05,
+      "loss": 0.8418,
+      "step": 10816
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 2.7740060690222703e-05,
+      "loss": 0.7852,
+      "step": 10817
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 2.7737489070616673e-05,
+      "loss": 0.6737,
+      "step": 10818
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 2.773491745101065e-05,
+      "loss": 0.6787,
+      "step": 10819
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 2.7732345831404622e-05,
+      "loss": 0.4554,
+      "step": 10820
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 2.7729774211798592e-05,
+      "loss": 0.7565,
+      "step": 10821
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 2.7727202592192565e-05,
+      "loss": 0.7103,
+      "step": 10822
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 2.7724630972586535e-05,
+      "loss": 0.822,
+      "step": 10823
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 2.7722059352980508e-05,
+      "loss": 0.8612,
+      "step": 10824
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 2.7719487733374484e-05,
+      "loss": 0.523,
+      "step": 10825
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 2.7716916113768454e-05,
+      "loss": 0.4696,
+      "step": 10826
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 2.7714344494162427e-05,
+      "loss": 0.785,
+      "step": 10827
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 2.7711772874556396e-05,
+      "loss": 0.6291,
+      "step": 10828
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 2.7709201254950366e-05,
+      "loss": 0.4915,
+      "step": 10829
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 2.770662963534434e-05,
+      "loss": 0.6522,
+      "step": 10830
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 2.7704058015738316e-05,
+      "loss": 0.8934,
+      "step": 10831
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 2.7701486396132285e-05,
+      "loss": 0.5288,
+      "step": 10832
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 2.769891477652626e-05,
+      "loss": 0.561,
+      "step": 10833
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 2.7696343156920228e-05,
+      "loss": 0.662,
+      "step": 10834
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 2.76937715373142e-05,
+      "loss": 0.6435,
+      "step": 10835
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 2.769119991770817e-05,
+      "loss": 0.6442,
+      "step": 10836
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 2.7688628298102147e-05,
+      "loss": 0.4979,
+      "step": 10837
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 2.768605667849612e-05,
+      "loss": 0.8872,
+      "step": 10838
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 2.768348505889009e-05,
+      "loss": 1.1935,
+      "step": 10839
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 2.7680913439284063e-05,
+      "loss": 0.5636,
+      "step": 10840
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 2.7678341819678033e-05,
+      "loss": 1.113,
+      "step": 10841
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 2.7675770200072006e-05,
+      "loss": 0.6125,
+      "step": 10842
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 2.7673198580465982e-05,
+      "loss": 0.6759,
+      "step": 10843
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 2.7670626960859952e-05,
+      "loss": 0.7915,
+      "step": 10844
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 2.7668055341253922e-05,
+      "loss": 0.5371,
+      "step": 10845
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 2.7665483721647895e-05,
+      "loss": 0.7335,
+      "step": 10846
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 2.7662912102041865e-05,
+      "loss": 0.4114,
+      "step": 10847
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 2.7660340482435838e-05,
+      "loss": 0.8672,
+      "step": 10848
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 2.7657768862829814e-05,
+      "loss": 1.048,
+      "step": 10849
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 2.7655197243223784e-05,
+      "loss": 0.6761,
+      "step": 10850
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 2.7652625623617757e-05,
+      "loss": 0.608,
+      "step": 10851
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 2.7650054004011727e-05,
+      "loss": 0.5499,
+      "step": 10852
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 2.76474823844057e-05,
+      "loss": 0.9285,
+      "step": 10853
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 2.764491076479967e-05,
+      "loss": 0.8498,
+      "step": 10854
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 2.7642339145193646e-05,
+      "loss": 0.503,
+      "step": 10855
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 2.763976752558762e-05,
+      "loss": 0.494,
+      "step": 10856
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 2.763719590598159e-05,
+      "loss": 0.8422,
+      "step": 10857
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 2.763462428637556e-05,
+      "loss": 0.7023,
+      "step": 10858
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 2.763205266676953e-05,
+      "loss": 0.5736,
+      "step": 10859
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 2.76294810471635e-05,
+      "loss": 0.6992,
+      "step": 10860
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 2.762690942755748e-05,
+      "loss": 0.5682,
+      "step": 10861
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 2.762433780795145e-05,
+      "loss": 0.9554,
+      "step": 10862
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 2.762176618834542e-05,
+      "loss": 0.9656,
+      "step": 10863
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 2.7619194568739393e-05,
+      "loss": 0.9177,
+      "step": 10864
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 2.7616622949133363e-05,
+      "loss": 0.6682,
+      "step": 10865
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 2.7614051329527336e-05,
+      "loss": 0.8417,
+      "step": 10866
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 2.7611479709921312e-05,
+      "loss": 0.985,
+      "step": 10867
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 2.7608908090315282e-05,
+      "loss": 0.6289,
+      "step": 10868
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 2.7606336470709255e-05,
+      "loss": 0.7654,
+      "step": 10869
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 2.7603764851103225e-05,
+      "loss": 0.4352,
+      "step": 10870
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 2.7601193231497198e-05,
+      "loss": 0.6907,
+      "step": 10871
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 2.7598621611891168e-05,
+      "loss": 0.6854,
+      "step": 10872
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 2.7596049992285144e-05,
+      "loss": 0.7669,
+      "step": 10873
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 2.7593478372679117e-05,
+      "loss": 0.8098,
+      "step": 10874
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 2.7590906753073087e-05,
+      "loss": 0.4497,
+      "step": 10875
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 2.7588335133467057e-05,
+      "loss": 0.4966,
+      "step": 10876
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 2.758576351386103e-05,
+      "loss": 0.6339,
+      "step": 10877
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 2.7583191894255e-05,
+      "loss": 1.1313,
+      "step": 10878
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 2.7580620274648976e-05,
+      "loss": 0.7845,
+      "step": 10879
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 2.757804865504295e-05,
+      "loss": 0.8156,
+      "step": 10880
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 2.757547703543692e-05,
+      "loss": 0.8286,
+      "step": 10881
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 2.757290541583089e-05,
+      "loss": 0.8254,
+      "step": 10882
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 2.757033379622486e-05,
+      "loss": 0.6534,
+      "step": 10883
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 2.7567762176618834e-05,
+      "loss": 0.763,
+      "step": 10884
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 2.756519055701281e-05,
+      "loss": 0.5795,
+      "step": 10885
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 2.756261893740678e-05,
+      "loss": 0.7743,
+      "step": 10886
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 2.7560047317800754e-05,
+      "loss": 0.5779,
+      "step": 10887
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 2.7557475698194723e-05,
+      "loss": 0.7785,
+      "step": 10888
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 2.7554904078588696e-05,
+      "loss": 0.9285,
+      "step": 10889
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 2.7552332458982666e-05,
+      "loss": 0.6413,
+      "step": 10890
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 2.7549760839376643e-05,
+      "loss": 0.6636,
+      "step": 10891
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 2.7547189219770612e-05,
+      "loss": 0.8276,
+      "step": 10892
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 2.7544617600164585e-05,
+      "loss": 0.5108,
+      "step": 10893
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 2.7542045980558555e-05,
+      "loss": 0.5737,
+      "step": 10894
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 2.7539474360952528e-05,
+      "loss": 0.9253,
+      "step": 10895
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 2.7536902741346498e-05,
+      "loss": 0.5446,
+      "step": 10896
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 2.7534331121740474e-05,
+      "loss": 0.7119,
+      "step": 10897
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 2.7531759502134447e-05,
+      "loss": 1.0739,
+      "step": 10898
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 2.7529187882528417e-05,
+      "loss": 0.5922,
+      "step": 10899
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 2.752661626292239e-05,
+      "loss": 0.7401,
+      "step": 10900
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 2.752404464331636e-05,
+      "loss": 0.8563,
+      "step": 10901
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 2.7521473023710333e-05,
+      "loss": 0.932,
+      "step": 10902
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 2.751890140410431e-05,
+      "loss": 0.8753,
+      "step": 10903
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 2.751632978449828e-05,
+      "loss": 1.1146,
+      "step": 10904
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 2.7513758164892252e-05,
+      "loss": 0.9985,
+      "step": 10905
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 2.751118654528622e-05,
+      "loss": 1.0425,
+      "step": 10906
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 2.750861492568019e-05,
+      "loss": 0.5275,
+      "step": 10907
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 2.750604330607417e-05,
+      "loss": 0.6562,
+      "step": 10908
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 2.750347168646814e-05,
+      "loss": 0.7886,
+      "step": 10909
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 2.750090006686211e-05,
+      "loss": 1.0369,
+      "step": 10910
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 2.7498328447256084e-05,
+      "loss": 0.7396,
+      "step": 10911
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 2.7495756827650053e-05,
+      "loss": 0.6865,
+      "step": 10912
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 2.7493185208044026e-05,
+      "loss": 0.8445,
+      "step": 10913
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 2.7490613588438003e-05,
+      "loss": 0.7058,
+      "step": 10914
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 2.7488041968831973e-05,
+      "loss": 0.7524,
+      "step": 10915
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 2.7485470349225946e-05,
+      "loss": 0.8814,
+      "step": 10916
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 2.7482898729619915e-05,
+      "loss": 0.7346,
+      "step": 10917
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 2.748032711001389e-05,
+      "loss": 0.5109,
+      "step": 10918
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 2.7477755490407858e-05,
+      "loss": 1.1664,
+      "step": 10919
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 2.7475183870801835e-05,
+      "loss": 0.689,
+      "step": 10920
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 2.7472612251195808e-05,
+      "loss": 0.6777,
+      "step": 10921
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 2.7470040631589777e-05,
+      "loss": 0.5998,
+      "step": 10922
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 2.7467469011983747e-05,
+      "loss": 0.6383,
+      "step": 10923
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 2.746489739237772e-05,
+      "loss": 0.6073,
+      "step": 10924
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 2.746232577277169e-05,
+      "loss": 0.5509,
+      "step": 10925
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 2.7459754153165666e-05,
+      "loss": 0.6002,
+      "step": 10926
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 2.745718253355964e-05,
+      "loss": 0.5194,
+      "step": 10927
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 2.745461091395361e-05,
+      "loss": 0.8159,
+      "step": 10928
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 2.7452039294347582e-05,
+      "loss": 0.7554,
+      "step": 10929
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 2.7449467674741552e-05,
+      "loss": 0.6945,
+      "step": 10930
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 2.7446896055135525e-05,
+      "loss": 0.751,
+      "step": 10931
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 2.74443244355295e-05,
+      "loss": 0.8,
+      "step": 10932
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 2.744175281592347e-05,
+      "loss": 0.9538,
+      "step": 10933
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 2.7439181196317444e-05,
+      "loss": 0.5311,
+      "step": 10934
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 2.7436609576711414e-05,
+      "loss": 0.7382,
+      "step": 10935
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 2.7434037957105387e-05,
+      "loss": 0.6346,
+      "step": 10936
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 2.7431466337499356e-05,
+      "loss": 0.5777,
+      "step": 10937
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 2.7428894717893333e-05,
+      "loss": 0.6556,
+      "step": 10938
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 2.7426323098287303e-05,
+      "loss": 0.8313,
+      "step": 10939
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 2.7423751478681276e-05,
+      "loss": 0.6246,
+      "step": 10940
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 2.7421179859075245e-05,
+      "loss": 0.5597,
+      "step": 10941
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 2.741860823946922e-05,
+      "loss": 0.4766,
+      "step": 10942
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 2.7416036619863188e-05,
+      "loss": 0.6805,
+      "step": 10943
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 2.7413465000257165e-05,
+      "loss": 0.7242,
+      "step": 10944
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 2.7410893380651138e-05,
+      "loss": 0.6812,
+      "step": 10945
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 2.7408321761045107e-05,
+      "loss": 0.6912,
+      "step": 10946
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 2.740575014143908e-05,
+      "loss": 0.7464,
+      "step": 10947
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 2.740317852183305e-05,
+      "loss": 0.4066,
+      "step": 10948
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 2.7400606902227023e-05,
+      "loss": 0.7726,
+      "step": 10949
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 2.7398035282621e-05,
+      "loss": 0.8975,
+      "step": 10950
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 2.739546366301497e-05,
+      "loss": 0.8148,
+      "step": 10951
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 2.7392892043408942e-05,
+      "loss": 0.5228,
+      "step": 10952
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 2.7390320423802912e-05,
+      "loss": 0.4663,
+      "step": 10953
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 2.7387748804196882e-05,
+      "loss": 0.7114,
+      "step": 10954
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 2.7385177184590855e-05,
+      "loss": 1.0691,
+      "step": 10955
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 2.738260556498483e-05,
+      "loss": 0.7327,
+      "step": 10956
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 2.73800339453788e-05,
+      "loss": 0.4577,
+      "step": 10957
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 2.7377462325772774e-05,
+      "loss": 0.3819,
+      "step": 10958
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 2.7374890706166744e-05,
+      "loss": 0.667,
+      "step": 10959
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 2.7372319086560717e-05,
+      "loss": 0.8464,
+      "step": 10960
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 2.7369747466954686e-05,
+      "loss": 0.8327,
+      "step": 10961
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 2.7367175847348663e-05,
+      "loss": 0.922,
+      "step": 10962
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 2.7364604227742636e-05,
+      "loss": 1.0404,
+      "step": 10963
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 2.7362032608136606e-05,
+      "loss": 0.9584,
+      "step": 10964
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 2.735946098853058e-05,
+      "loss": 0.5973,
+      "step": 10965
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 2.735688936892455e-05,
+      "loss": 0.4412,
+      "step": 10966
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 2.735431774931852e-05,
+      "loss": 0.6068,
+      "step": 10967
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 2.7351746129712498e-05,
+      "loss": 0.5369,
+      "step": 10968
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 2.7349174510106468e-05,
+      "loss": 0.4179,
+      "step": 10969
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 2.7346602890500437e-05,
+      "loss": 0.7149,
+      "step": 10970
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 2.734403127089441e-05,
+      "loss": 0.7184,
+      "step": 10971
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 2.734145965128838e-05,
+      "loss": 0.8005,
+      "step": 10972
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 2.7338888031682353e-05,
+      "loss": 0.5696,
+      "step": 10973
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 2.733631641207633e-05,
+      "loss": 0.6085,
+      "step": 10974
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 2.73337447924703e-05,
+      "loss": 0.8059,
+      "step": 10975
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 2.7331173172864272e-05,
+      "loss": 0.7701,
+      "step": 10976
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 2.7328601553258242e-05,
+      "loss": 0.5815,
+      "step": 10977
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 2.7326029933652215e-05,
+      "loss": 0.7871,
+      "step": 10978
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 2.7323458314046185e-05,
+      "loss": 0.5457,
+      "step": 10979
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 2.732088669444016e-05,
+      "loss": 0.7725,
+      "step": 10980
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 2.7318315074834134e-05,
+      "loss": 0.4665,
+      "step": 10981
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 2.7315743455228104e-05,
+      "loss": 0.672,
+      "step": 10982
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 2.7313171835622077e-05,
+      "loss": 1.0744,
+      "step": 10983
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 2.7310600216016047e-05,
+      "loss": 0.4927,
+      "step": 10984
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 2.7308028596410017e-05,
+      "loss": 0.6199,
+      "step": 10985
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 2.7305456976803996e-05,
+      "loss": 0.8381,
+      "step": 10986
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 2.7302885357197966e-05,
+      "loss": 0.9273,
+      "step": 10987
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 2.7300313737591936e-05,
+      "loss": 0.918,
+      "step": 10988
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 2.729774211798591e-05,
+      "loss": 0.5085,
+      "step": 10989
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 2.729517049837988e-05,
+      "loss": 0.5501,
+      "step": 10990
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 2.729259887877385e-05,
+      "loss": 0.5605,
+      "step": 10991
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 2.7290027259167828e-05,
+      "loss": 0.615,
+      "step": 10992
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 2.7287455639561798e-05,
+      "loss": 0.8267,
+      "step": 10993
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 2.728488401995577e-05,
+      "loss": 0.6797,
+      "step": 10994
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 2.728231240034974e-05,
+      "loss": 0.5205,
+      "step": 10995
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 2.7279740780743714e-05,
+      "loss": 0.8903,
+      "step": 10996
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 2.7277169161137683e-05,
+      "loss": 0.5043,
+      "step": 10997
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 2.727459754153166e-05,
+      "loss": 0.5841,
+      "step": 10998
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 2.7272025921925633e-05,
+      "loss": 0.4754,
+      "step": 10999
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 2.7269454302319602e-05,
+      "loss": 0.6294,
+      "step": 11000
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 2.7266882682713572e-05,
+      "loss": 0.567,
+      "step": 11001
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 2.7264311063107545e-05,
+      "loss": 0.8646,
+      "step": 11002
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 2.7261739443501515e-05,
+      "loss": 0.6742,
+      "step": 11003
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 2.725916782389549e-05,
+      "loss": 0.7822,
+      "step": 11004
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 2.7256596204289464e-05,
+      "loss": 0.4887,
+      "step": 11005
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 2.7254024584683434e-05,
+      "loss": 0.765,
+      "step": 11006
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 2.7251452965077407e-05,
+      "loss": 0.5203,
+      "step": 11007
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 2.7248881345471377e-05,
+      "loss": 0.8716,
+      "step": 11008
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 2.724630972586535e-05,
+      "loss": 0.6326,
+      "step": 11009
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 2.7243738106259326e-05,
+      "loss": 0.483,
+      "step": 11010
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 2.7241166486653296e-05,
+      "loss": 0.7229,
+      "step": 11011
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 2.723859486704727e-05,
+      "loss": 0.6598,
+      "step": 11012
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 2.723602324744124e-05,
+      "loss": 1.0771,
+      "step": 11013
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 2.7233451627835212e-05,
+      "loss": 1.035,
+      "step": 11014
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 2.723088000822918e-05,
+      "loss": 0.5596,
+      "step": 11015
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 2.7228308388623158e-05,
+      "loss": 0.716,
+      "step": 11016
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 2.7225736769017128e-05,
+      "loss": 0.7594,
+      "step": 11017
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 2.72231651494111e-05,
+      "loss": 0.5887,
+      "step": 11018
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 2.722059352980507e-05,
+      "loss": 0.9081,
+      "step": 11019
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 2.7218021910199044e-05,
+      "loss": 0.7016,
+      "step": 11020
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 2.7215450290593013e-05,
+      "loss": 0.8406,
+      "step": 11021
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 2.721287867098699e-05,
+      "loss": 0.574,
+      "step": 11022
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 2.7210307051380963e-05,
+      "loss": 0.7759,
+      "step": 11023
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 2.7207735431774933e-05,
+      "loss": 0.5995,
+      "step": 11024
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 2.7205163812168906e-05,
+      "loss": 0.6068,
+      "step": 11025
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 2.7202592192562875e-05,
+      "loss": 0.8689,
+      "step": 11026
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 2.720002057295685e-05,
+      "loss": 0.7604,
+      "step": 11027
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 2.7197448953350825e-05,
+      "loss": 0.7396,
+      "step": 11028
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 2.7194877333744795e-05,
+      "loss": 0.7309,
+      "step": 11029
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 2.7192305714138768e-05,
+      "loss": 0.7955,
+      "step": 11030
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 2.7189734094532737e-05,
+      "loss": 0.5276,
+      "step": 11031
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 2.7187162474926707e-05,
+      "loss": 1.0328,
+      "step": 11032
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 2.718459085532068e-05,
+      "loss": 0.5793,
+      "step": 11033
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 2.7182019235714656e-05,
+      "loss": 0.5664,
+      "step": 11034
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 2.7179447616108626e-05,
+      "loss": 0.6126,
+      "step": 11035
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 2.71768759965026e-05,
+      "loss": 0.6759,
+      "step": 11036
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 2.717430437689657e-05,
+      "loss": 1.0049,
+      "step": 11037
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 2.7171732757290542e-05,
+      "loss": 0.9996,
+      "step": 11038
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 2.716916113768451e-05,
+      "loss": 1.1171,
+      "step": 11039
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 2.7166589518078488e-05,
+      "loss": 0.7345,
+      "step": 11040
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 2.716401789847246e-05,
+      "loss": 1.0447,
+      "step": 11041
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 2.716144627886643e-05,
+      "loss": 0.9225,
+      "step": 11042
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 2.7158874659260404e-05,
+      "loss": 0.5705,
+      "step": 11043
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 2.7156303039654374e-05,
+      "loss": 0.459,
+      "step": 11044
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 2.7153731420048343e-05,
+      "loss": 0.5097,
+      "step": 11045
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 2.7151159800442323e-05,
+      "loss": 0.9359,
+      "step": 11046
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 2.7148588180836293e-05,
+      "loss": 0.6977,
+      "step": 11047
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 2.7146016561230263e-05,
+      "loss": 0.8488,
+      "step": 11048
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 2.7143444941624236e-05,
+      "loss": 0.7146,
+      "step": 11049
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 2.7140873322018205e-05,
+      "loss": 0.8134,
+      "step": 11050
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 2.713830170241218e-05,
+      "loss": 0.5225,
+      "step": 11051
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 2.7135730082806155e-05,
+      "loss": 0.779,
+      "step": 11052
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 2.7133158463200125e-05,
+      "loss": 0.6226,
+      "step": 11053
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 2.7130586843594098e-05,
+      "loss": 1.0889,
+      "step": 11054
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 2.7128015223988067e-05,
+      "loss": 0.7692,
+      "step": 11055
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 2.712544360438204e-05,
+      "loss": 0.8298,
+      "step": 11056
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 2.7122871984776017e-05,
+      "loss": 0.7641,
+      "step": 11057
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 2.7120300365169987e-05,
+      "loss": 0.7054,
+      "step": 11058
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 2.711772874556396e-05,
+      "loss": 0.7472,
+      "step": 11059
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 2.711515712595793e-05,
+      "loss": 0.3228,
+      "step": 11060
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 2.7112585506351902e-05,
+      "loss": 0.5613,
+      "step": 11061
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 2.7110013886745872e-05,
+      "loss": 0.773,
+      "step": 11062
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 2.710744226713985e-05,
+      "loss": 0.6315,
+      "step": 11063
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 2.710487064753382e-05,
+      "loss": 0.55,
+      "step": 11064
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 2.710229902792779e-05,
+      "loss": 0.4736,
+      "step": 11065
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 2.709972740832176e-05,
+      "loss": 0.6871,
+      "step": 11066
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 2.7097155788715734e-05,
+      "loss": 0.6541,
+      "step": 11067
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 2.7094584169109704e-05,
+      "loss": 0.6487,
+      "step": 11068
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 2.709201254950368e-05,
+      "loss": 0.7307,
+      "step": 11069
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 2.7089440929897653e-05,
+      "loss": 0.8114,
+      "step": 11070
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 2.7086869310291623e-05,
+      "loss": 0.5331,
+      "step": 11071
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 2.7084297690685596e-05,
+      "loss": 0.7123,
+      "step": 11072
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 2.7081726071079566e-05,
+      "loss": 0.8809,
+      "step": 11073
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 2.707915445147354e-05,
+      "loss": 0.6267,
+      "step": 11074
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 2.7076582831867515e-05,
+      "loss": 0.7842,
+      "step": 11075
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 2.7074011212261485e-05,
+      "loss": 0.686,
+      "step": 11076
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 2.7071439592655458e-05,
+      "loss": 1.0557,
+      "step": 11077
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 2.7068867973049428e-05,
+      "loss": 1.0258,
+      "step": 11078
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 2.7066296353443397e-05,
+      "loss": 0.4591,
+      "step": 11079
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 2.706372473383737e-05,
+      "loss": 0.4146,
+      "step": 11080
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 2.7061153114231347e-05,
+      "loss": 0.6588,
+      "step": 11081
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 2.7058581494625317e-05,
+      "loss": 0.6166,
+      "step": 11082
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 2.705600987501929e-05,
+      "loss": 0.7686,
+      "step": 11083
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 2.705343825541326e-05,
+      "loss": 0.6095,
+      "step": 11084
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 2.7050866635807232e-05,
+      "loss": 0.7129,
+      "step": 11085
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 2.7048295016201202e-05,
+      "loss": 1.1255,
+      "step": 11086
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 2.704572339659518e-05,
+      "loss": 0.8399,
+      "step": 11087
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 2.704315177698915e-05,
+      "loss": 0.6091,
+      "step": 11088
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 2.704058015738312e-05,
+      "loss": 0.573,
+      "step": 11089
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 2.7038008537777094e-05,
+      "loss": 0.7466,
+      "step": 11090
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 2.7035436918171064e-05,
+      "loss": 0.6474,
+      "step": 11091
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 2.7032865298565037e-05,
+      "loss": 0.8141,
+      "step": 11092
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 2.7030293678959014e-05,
+      "loss": 0.6284,
+      "step": 11093
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 2.7027722059352983e-05,
+      "loss": 0.4088,
+      "step": 11094
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 2.7025150439746953e-05,
+      "loss": 0.4278,
+      "step": 11095
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 2.7022578820140926e-05,
+      "loss": 0.5357,
+      "step": 11096
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 2.7020007200534896e-05,
+      "loss": 0.8245,
+      "step": 11097
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 2.701743558092887e-05,
+      "loss": 0.5271,
+      "step": 11098
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 2.7014863961322845e-05,
+      "loss": 0.3849,
+      "step": 11099
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 2.7012292341716815e-05,
+      "loss": 0.7246,
+      "step": 11100
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 2.7009720722110788e-05,
+      "loss": 0.4686,
+      "step": 11101
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 2.7007149102504758e-05,
+      "loss": 0.4982,
+      "step": 11102
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 2.700457748289873e-05,
+      "loss": 0.5223,
+      "step": 11103
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 2.70020058632927e-05,
+      "loss": 0.6404,
+      "step": 11104
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 2.6999434243686677e-05,
+      "loss": 0.8405,
+      "step": 11105
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 2.699686262408065e-05,
+      "loss": 0.696,
+      "step": 11106
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 2.699429100447462e-05,
+      "loss": 0.7924,
+      "step": 11107
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 2.6991719384868593e-05,
+      "loss": 0.4861,
+      "step": 11108
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 2.6989147765262562e-05,
+      "loss": 0.6055,
+      "step": 11109
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 2.6986576145656532e-05,
+      "loss": 0.6963,
+      "step": 11110
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 2.6984004526050512e-05,
+      "loss": 0.7928,
+      "step": 11111
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 2.698143290644448e-05,
+      "loss": 0.6022,
+      "step": 11112
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 2.697886128683845e-05,
+      "loss": 1.0664,
+      "step": 11113
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 2.6976289667232424e-05,
+      "loss": 0.6785,
+      "step": 11114
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 2.6973718047626394e-05,
+      "loss": 0.8135,
+      "step": 11115
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 2.6971146428020367e-05,
+      "loss": 0.6799,
+      "step": 11116
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 2.6968574808414344e-05,
+      "loss": 0.9863,
+      "step": 11117
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 2.6966003188808313e-05,
+      "loss": 0.7584,
+      "step": 11118
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 2.6963431569202286e-05,
+      "loss": 0.6006,
+      "step": 11119
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 2.6960859949596256e-05,
+      "loss": 0.9261,
+      "step": 11120
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 2.695828832999023e-05,
+      "loss": 0.5483,
+      "step": 11121
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 2.69557167103842e-05,
+      "loss": 0.6416,
+      "step": 11122
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 2.6953145090778175e-05,
+      "loss": 0.6446,
+      "step": 11123
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 2.695057347117215e-05,
+      "loss": 0.6237,
+      "step": 11124
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 2.6948001851566118e-05,
+      "loss": 0.5589,
+      "step": 11125
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 2.6945430231960088e-05,
+      "loss": 0.5537,
+      "step": 11126
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 2.694285861235406e-05,
+      "loss": 1.2079,
+      "step": 11127
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 2.694028699274803e-05,
+      "loss": 0.4183,
+      "step": 11128
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 2.6937715373142007e-05,
+      "loss": 0.7303,
+      "step": 11129
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 2.693514375353598e-05,
+      "loss": 0.4945,
+      "step": 11130
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 2.693257213392995e-05,
+      "loss": 0.8297,
+      "step": 11131
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 2.6930000514323923e-05,
+      "loss": 0.7042,
+      "step": 11132
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 2.6927428894717892e-05,
+      "loss": 0.9363,
+      "step": 11133
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 2.6924857275111866e-05,
+      "loss": 0.5659,
+      "step": 11134
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 2.6922285655505842e-05,
+      "loss": 0.5943,
+      "step": 11135
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 2.6919714035899812e-05,
+      "loss": 0.7994,
+      "step": 11136
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 2.6917142416293785e-05,
+      "loss": 0.8862,
+      "step": 11137
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 2.6914570796687754e-05,
+      "loss": 0.7111,
+      "step": 11138
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 2.6911999177081728e-05,
+      "loss": 0.6737,
+      "step": 11139
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 2.6909427557475697e-05,
+      "loss": 0.4691,
+      "step": 11140
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 2.6906855937869674e-05,
+      "loss": 1.1552,
+      "step": 11141
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 2.6904284318263643e-05,
+      "loss": 0.5153,
+      "step": 11142
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 2.6901712698657616e-05,
+      "loss": 0.4384,
+      "step": 11143
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 2.6899141079051586e-05,
+      "loss": 0.551,
+      "step": 11144
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 2.689656945944556e-05,
+      "loss": 0.8546,
+      "step": 11145
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 2.689399783983953e-05,
+      "loss": 0.7682,
+      "step": 11146
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 2.6891426220233505e-05,
+      "loss": 1.3743,
+      "step": 11147
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 2.688885460062748e-05,
+      "loss": 0.5293,
+      "step": 11148
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 2.6886282981021448e-05,
+      "loss": 0.5205,
+      "step": 11149
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 2.688371136141542e-05,
+      "loss": 0.6008,
+      "step": 11150
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 2.688113974180939e-05,
+      "loss": 0.779,
+      "step": 11151
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 2.6878568122203364e-05,
+      "loss": 0.5515,
+      "step": 11152
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 2.687599650259734e-05,
+      "loss": 0.5366,
+      "step": 11153
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 2.687342488299131e-05,
+      "loss": 0.4017,
+      "step": 11154
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 2.6870853263385283e-05,
+      "loss": 0.6554,
+      "step": 11155
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 2.6868281643779253e-05,
+      "loss": 0.9736,
+      "step": 11156
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 2.6865710024173223e-05,
+      "loss": 0.9607,
+      "step": 11157
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 2.6863138404567196e-05,
+      "loss": 0.6032,
+      "step": 11158
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 2.6860566784961172e-05,
+      "loss": 0.553,
+      "step": 11159
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 2.6857995165355142e-05,
+      "loss": 0.8245,
+      "step": 11160
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 2.6855423545749115e-05,
+      "loss": 0.6677,
+      "step": 11161
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 2.6852851926143085e-05,
+      "loss": 0.329,
+      "step": 11162
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 2.6850280306537058e-05,
+      "loss": 0.6788,
+      "step": 11163
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 2.6847708686931027e-05,
+      "loss": 0.5883,
+      "step": 11164
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 2.6845137067325004e-05,
+      "loss": 0.8402,
+      "step": 11165
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 2.6842565447718977e-05,
+      "loss": 0.9511,
+      "step": 11166
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 2.6839993828112946e-05,
+      "loss": 0.7858,
+      "step": 11167
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 2.683742220850692e-05,
+      "loss": 0.7923,
+      "step": 11168
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 2.683485058890089e-05,
+      "loss": 0.9361,
+      "step": 11169
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 2.683227896929486e-05,
+      "loss": 0.8942,
+      "step": 11170
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 2.682970734968884e-05,
+      "loss": 0.6374,
+      "step": 11171
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 2.682713573008281e-05,
+      "loss": 0.5802,
+      "step": 11172
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 2.6824564110476778e-05,
+      "loss": 0.6003,
+      "step": 11173
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 2.682199249087075e-05,
+      "loss": 0.8049,
+      "step": 11174
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 2.681942087126472e-05,
+      "loss": 0.6851,
+      "step": 11175
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 2.6816849251658694e-05,
+      "loss": 0.7116,
+      "step": 11176
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 2.681427763205267e-05,
+      "loss": 0.6653,
+      "step": 11177
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 2.681170601244664e-05,
+      "loss": 0.703,
+      "step": 11178
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 2.6809134392840613e-05,
+      "loss": 0.2516,
+      "step": 11179
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 2.6806562773234583e-05,
+      "loss": 0.7265,
+      "step": 11180
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 2.6803991153628556e-05,
+      "loss": 0.5059,
+      "step": 11181
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 2.6801419534022526e-05,
+      "loss": 0.8446,
+      "step": 11182
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 2.6798847914416502e-05,
+      "loss": 0.6371,
+      "step": 11183
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 2.6796276294810475e-05,
+      "loss": 0.5305,
+      "step": 11184
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 2.6793704675204445e-05,
+      "loss": 0.7476,
+      "step": 11185
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 2.6791133055598418e-05,
+      "loss": 0.685,
+      "step": 11186
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 2.6788561435992388e-05,
+      "loss": 0.9175,
+      "step": 11187
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 2.6785989816386357e-05,
+      "loss": 0.6373,
+      "step": 11188
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 2.6783418196780337e-05,
+      "loss": 0.9762,
+      "step": 11189
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 2.6780846577174307e-05,
+      "loss": 0.6325,
+      "step": 11190
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 2.6778274957568277e-05,
+      "loss": 0.7814,
+      "step": 11191
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 2.677570333796225e-05,
+      "loss": 1.027,
+      "step": 11192
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 2.677313171835622e-05,
+      "loss": 0.6726,
+      "step": 11193
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 2.6770560098750192e-05,
+      "loss": 1.1246,
+      "step": 11194
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 2.676798847914417e-05,
+      "loss": 0.745,
+      "step": 11195
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 2.676541685953814e-05,
+      "loss": 0.5272,
+      "step": 11196
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 2.676284523993211e-05,
+      "loss": 0.4807,
+      "step": 11197
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 2.676027362032608e-05,
+      "loss": 0.7536,
+      "step": 11198
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 2.6757702000720054e-05,
+      "loss": 0.9932,
+      "step": 11199
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 2.6755130381114024e-05,
+      "loss": 0.5851,
+      "step": 11200
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 2.6752558761508e-05,
+      "loss": 0.564,
+      "step": 11201
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 2.6749987141901974e-05,
+      "loss": 0.475,
+      "step": 11202
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 2.6747415522295943e-05,
+      "loss": 0.5101,
+      "step": 11203
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 2.6744843902689913e-05,
+      "loss": 0.686,
+      "step": 11204
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 2.6742272283083886e-05,
+      "loss": 0.5315,
+      "step": 11205
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 2.6739700663477862e-05,
+      "loss": 0.8988,
+      "step": 11206
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 2.6737129043871832e-05,
+      "loss": 0.5761,
+      "step": 11207
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 2.6734557424265805e-05,
+      "loss": 0.5382,
+      "step": 11208
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 2.6731985804659775e-05,
+      "loss": 0.6768,
+      "step": 11209
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 2.6729414185053748e-05,
+      "loss": 0.8468,
+      "step": 11210
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 2.6726842565447718e-05,
+      "loss": 0.6138,
+      "step": 11211
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 2.6724270945841694e-05,
+      "loss": 0.5017,
+      "step": 11212
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 2.6721699326235667e-05,
+      "loss": 0.5309,
+      "step": 11213
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 2.6719127706629637e-05,
+      "loss": 0.6114,
+      "step": 11214
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 2.671655608702361e-05,
+      "loss": 0.6114,
+      "step": 11215
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 2.671398446741758e-05,
+      "loss": 0.4365,
+      "step": 11216
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 2.6711412847811553e-05,
+      "loss": 0.523,
+      "step": 11217
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 2.670884122820553e-05,
+      "loss": 0.6361,
+      "step": 11218
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 2.67062696085995e-05,
+      "loss": 0.6428,
+      "step": 11219
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 2.670369798899347e-05,
+      "loss": 0.6439,
+      "step": 11220
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 2.670112636938744e-05,
+      "loss": 0.6025,
+      "step": 11221
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 2.669855474978141e-05,
+      "loss": 0.6739,
+      "step": 11222
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 2.6695983130175384e-05,
+      "loss": 1.016,
+      "step": 11223
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 2.669341151056936e-05,
+      "loss": 0.8007,
+      "step": 11224
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 2.669083989096333e-05,
+      "loss": 0.5984,
+      "step": 11225
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 2.6688268271357304e-05,
+      "loss": 0.6137,
+      "step": 11226
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 2.6685696651751273e-05,
+      "loss": 0.9012,
+      "step": 11227
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 2.6683125032145246e-05,
+      "loss": 0.4608,
+      "step": 11228
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 2.6680553412539216e-05,
+      "loss": 0.5449,
+      "step": 11229
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 2.6677981792933193e-05,
+      "loss": 1.1404,
+      "step": 11230
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 2.6675410173327166e-05,
+      "loss": 0.5944,
+      "step": 11231
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 2.6672838553721135e-05,
+      "loss": 0.5495,
+      "step": 11232
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 2.667026693411511e-05,
+      "loss": 0.7082,
+      "step": 11233
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 2.6667695314509078e-05,
+      "loss": 0.6753,
+      "step": 11234
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 2.6665123694903048e-05,
+      "loss": 0.5486,
+      "step": 11235
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 2.6662552075297028e-05,
+      "loss": 0.7609,
+      "step": 11236
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 2.6659980455690997e-05,
+      "loss": 0.7503,
+      "step": 11237
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 2.6657408836084967e-05,
+      "loss": 0.5031,
+      "step": 11238
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 2.665483721647894e-05,
+      "loss": 0.7137,
+      "step": 11239
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 2.665226559687291e-05,
+      "loss": 0.4783,
+      "step": 11240
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 2.6649693977266883e-05,
+      "loss": 1.0404,
+      "step": 11241
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 2.664712235766086e-05,
+      "loss": 0.6568,
+      "step": 11242
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 2.664455073805483e-05,
+      "loss": 0.7145,
+      "step": 11243
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 2.6641979118448802e-05,
+      "loss": 0.8265,
+      "step": 11244
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 2.663940749884277e-05,
+      "loss": 0.6065,
+      "step": 11245
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 2.6636835879236745e-05,
+      "loss": 0.7671,
+      "step": 11246
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 2.6634264259630714e-05,
+      "loss": 0.6649,
+      "step": 11247
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 2.663169264002469e-05,
+      "loss": 0.4817,
+      "step": 11248
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 2.6629121020418664e-05,
+      "loss": 0.7452,
+      "step": 11249
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 2.6626549400812634e-05,
+      "loss": 0.3449,
+      "step": 11250
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 2.6623977781206603e-05,
+      "loss": 0.4035,
+      "step": 11251
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 2.6621406161600576e-05,
+      "loss": 0.4848,
+      "step": 11252
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 2.6618834541994546e-05,
+      "loss": 0.4764,
+      "step": 11253
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 2.6616262922388523e-05,
+      "loss": 0.7932,
+      "step": 11254
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 2.6613691302782496e-05,
+      "loss": 0.7052,
+      "step": 11255
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 2.6611119683176465e-05,
+      "loss": 0.4775,
+      "step": 11256
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 2.660854806357044e-05,
+      "loss": 0.6015,
+      "step": 11257
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 2.6605976443964408e-05,
+      "loss": 0.7577,
+      "step": 11258
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 2.660340482435838e-05,
+      "loss": 0.7667,
+      "step": 11259
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 2.6600833204752358e-05,
+      "loss": 0.9052,
+      "step": 11260
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 2.6598261585146327e-05,
+      "loss": 0.9224,
+      "step": 11261
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 2.65956899655403e-05,
+      "loss": 0.5622,
+      "step": 11262
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 2.659311834593427e-05,
+      "loss": 0.786,
+      "step": 11263
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 2.6590546726328243e-05,
+      "loss": 0.6168,
+      "step": 11264
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 2.6587975106722213e-05,
+      "loss": 0.5519,
+      "step": 11265
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 2.658540348711619e-05,
+      "loss": 0.8872,
+      "step": 11266
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 2.658283186751016e-05,
+      "loss": 0.7299,
+      "step": 11267
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 2.6580260247904132e-05,
+      "loss": 0.5712,
+      "step": 11268
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 2.6577688628298102e-05,
+      "loss": 0.6576,
+      "step": 11269
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 2.6575117008692075e-05,
+      "loss": 0.6918,
+      "step": 11270
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 2.6572545389086044e-05,
+      "loss": 0.9212,
+      "step": 11271
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 2.656997376948002e-05,
+      "loss": 0.7154,
+      "step": 11272
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 2.6567402149873994e-05,
+      "loss": 0.6281,
+      "step": 11273
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 2.6564830530267964e-05,
+      "loss": 0.6194,
+      "step": 11274
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 2.6562258910661937e-05,
+      "loss": 0.4456,
+      "step": 11275
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 2.6559687291055906e-05,
+      "loss": 0.8008,
+      "step": 11276
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 2.655711567144988e-05,
+      "loss": 0.6254,
+      "step": 11277
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 2.6554544051843856e-05,
+      "loss": 0.8058,
+      "step": 11278
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 2.6551972432237826e-05,
+      "loss": 0.7432,
+      "step": 11279
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 2.65494008126318e-05,
+      "loss": 0.5887,
+      "step": 11280
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 2.654682919302577e-05,
+      "loss": 0.8407,
+      "step": 11281
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 2.6544257573419738e-05,
+      "loss": 0.6248,
+      "step": 11282
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 2.654168595381371e-05,
+      "loss": 0.9219,
+      "step": 11283
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 2.6539114334207688e-05,
+      "loss": 0.7035,
+      "step": 11284
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 2.6536542714601657e-05,
+      "loss": 0.5895,
+      "step": 11285
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 2.653397109499563e-05,
+      "loss": 0.8631,
+      "step": 11286
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 2.65313994753896e-05,
+      "loss": 0.729,
+      "step": 11287
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 2.6528827855783573e-05,
+      "loss": 0.6338,
+      "step": 11288
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 2.6526256236177543e-05,
+      "loss": 0.7149,
+      "step": 11289
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 2.652368461657152e-05,
+      "loss": 0.8975,
+      "step": 11290
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 2.6521112996965492e-05,
+      "loss": 0.9642,
+      "step": 11291
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 2.6518541377359462e-05,
+      "loss": 0.7167,
+      "step": 11292
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 2.6515969757753435e-05,
+      "loss": 0.7082,
+      "step": 11293
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 2.6513398138147405e-05,
+      "loss": 0.8748,
+      "step": 11294
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 2.6510826518541375e-05,
+      "loss": 0.5414,
+      "step": 11295
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 2.6508254898935354e-05,
+      "loss": 0.6209,
+      "step": 11296
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 2.6505683279329324e-05,
+      "loss": 0.6163,
+      "step": 11297
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 2.6503111659723294e-05,
+      "loss": 0.99,
+      "step": 11298
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 2.6500540040117267e-05,
+      "loss": 0.6172,
+      "step": 11299
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 2.6497968420511237e-05,
+      "loss": 0.8226,
+      "step": 11300
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 2.649539680090521e-05,
+      "loss": 0.4584,
+      "step": 11301
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 2.6492825181299186e-05,
+      "loss": 0.8446,
+      "step": 11302
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 2.6490253561693156e-05,
+      "loss": 0.8607,
+      "step": 11303
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 2.648768194208713e-05,
+      "loss": 0.7101,
+      "step": 11304
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 2.64851103224811e-05,
+      "loss": 0.6562,
+      "step": 11305
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 2.648253870287507e-05,
+      "loss": 0.5356,
+      "step": 11306
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 2.647996708326904e-05,
+      "loss": 0.5872,
+      "step": 11307
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 2.6477395463663018e-05,
+      "loss": 0.5886,
+      "step": 11308
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 2.647482384405699e-05,
+      "loss": 0.6054,
+      "step": 11309
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 2.647225222445096e-05,
+      "loss": 0.6248,
+      "step": 11310
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 2.6469680604844934e-05,
+      "loss": 0.8465,
+      "step": 11311
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 2.6467108985238903e-05,
+      "loss": 0.838,
+      "step": 11312
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 2.6464537365632873e-05,
+      "loss": 0.5731,
+      "step": 11313
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 2.6461965746026853e-05,
+      "loss": 0.5779,
+      "step": 11314
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 2.6459394126420822e-05,
+      "loss": 0.857,
+      "step": 11315
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 2.6456822506814792e-05,
+      "loss": 0.7433,
+      "step": 11316
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 2.6454250887208765e-05,
+      "loss": 0.6852,
+      "step": 11317
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 2.6451679267602735e-05,
+      "loss": 0.8646,
+      "step": 11318
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 2.6449107647996708e-05,
+      "loss": 0.6986,
+      "step": 11319
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 2.6446536028390684e-05,
+      "loss": 0.9475,
+      "step": 11320
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 2.6443964408784654e-05,
+      "loss": 0.5799,
+      "step": 11321
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 2.6441392789178627e-05,
+      "loss": 0.6216,
+      "step": 11322
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 2.6438821169572597e-05,
+      "loss": 0.9199,
+      "step": 11323
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 2.643624954996657e-05,
+      "loss": 0.6875,
+      "step": 11324
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 2.643367793036054e-05,
+      "loss": 0.6273,
+      "step": 11325
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 2.6431106310754516e-05,
+      "loss": 0.7473,
+      "step": 11326
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 2.642853469114849e-05,
+      "loss": 0.6502,
+      "step": 11327
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 2.642596307154246e-05,
+      "loss": 0.4102,
+      "step": 11328
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 2.642339145193643e-05,
+      "loss": 0.8311,
+      "step": 11329
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 2.64208198323304e-05,
+      "loss": 1.0143,
+      "step": 11330
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 2.641824821272437e-05,
+      "loss": 0.5761,
+      "step": 11331
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 2.6415676593118348e-05,
+      "loss": 0.5335,
+      "step": 11332
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 2.641310497351232e-05,
+      "loss": 0.5978,
+      "step": 11333
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 2.641053335390629e-05,
+      "loss": 0.6754,
+      "step": 11334
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 2.6407961734300264e-05,
+      "loss": 0.3593,
+      "step": 11335
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 2.6405390114694233e-05,
+      "loss": 0.6681,
+      "step": 11336
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 2.6402818495088206e-05,
+      "loss": 0.9574,
+      "step": 11337
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 2.6400246875482183e-05,
+      "loss": 0.5045,
+      "step": 11338
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 2.6397675255876152e-05,
+      "loss": 0.5042,
+      "step": 11339
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 2.6395103636270126e-05,
+      "loss": 0.742,
+      "step": 11340
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 2.6392532016664095e-05,
+      "loss": 0.3693,
+      "step": 11341
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 2.638996039705807e-05,
+      "loss": 0.6408,
+      "step": 11342
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 2.6387388777452038e-05,
+      "loss": 0.5549,
+      "step": 11343
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 2.6384817157846014e-05,
+      "loss": 0.5455,
+      "step": 11344
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 2.6382245538239984e-05,
+      "loss": 0.4044,
+      "step": 11345
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 2.6379673918633957e-05,
+      "loss": 0.417,
+      "step": 11346
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 2.6377102299027927e-05,
+      "loss": 0.6741,
+      "step": 11347
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 2.63745306794219e-05,
+      "loss": 0.7035,
+      "step": 11348
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 2.637195905981587e-05,
+      "loss": 0.4507,
+      "step": 11349
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 2.6369387440209846e-05,
+      "loss": 0.7829,
+      "step": 11350
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 2.636681582060382e-05,
+      "loss": 0.535,
+      "step": 11351
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 2.636424420099779e-05,
+      "loss": 0.4876,
+      "step": 11352
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 2.6361672581391762e-05,
+      "loss": 0.8969,
+      "step": 11353
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 2.635910096178573e-05,
+      "loss": 0.5251,
+      "step": 11354
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 2.6356529342179708e-05,
+      "loss": 0.6105,
+      "step": 11355
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 2.635395772257368e-05,
+      "loss": 0.5823,
+      "step": 11356
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 2.635138610296765e-05,
+      "loss": 0.368,
+      "step": 11357
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 2.6348814483361624e-05,
+      "loss": 0.6428,
+      "step": 11358
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 2.6346242863755594e-05,
+      "loss": 0.6091,
+      "step": 11359
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 2.6343671244149563e-05,
+      "loss": 1.0102,
+      "step": 11360
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 2.6341099624543543e-05,
+      "loss": 1.1367,
+      "step": 11361
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 2.6338528004937513e-05,
+      "loss": 0.9282,
+      "step": 11362
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 2.6335956385331483e-05,
+      "loss": 0.7806,
+      "step": 11363
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 2.6333384765725456e-05,
+      "loss": 0.6035,
+      "step": 11364
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 2.6330813146119425e-05,
+      "loss": 0.641,
+      "step": 11365
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 2.63282415265134e-05,
+      "loss": 0.6445,
+      "step": 11366
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 2.6325669906907375e-05,
+      "loss": 0.5788,
+      "step": 11367
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 2.6323098287301345e-05,
+      "loss": 0.6308,
+      "step": 11368
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 2.6320526667695318e-05,
+      "loss": 0.7332,
+      "step": 11369
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 2.6317955048089287e-05,
+      "loss": 0.6645,
+      "step": 11370
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 2.631538342848326e-05,
+      "loss": 0.6684,
+      "step": 11371
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 2.631281180887723e-05,
+      "loss": 0.6062,
+      "step": 11372
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 2.6310240189271206e-05,
+      "loss": 0.6244,
+      "step": 11373
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 2.630766856966518e-05,
+      "loss": 0.6941,
+      "step": 11374
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 2.630509695005915e-05,
+      "loss": 0.8841,
+      "step": 11375
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 2.630252533045312e-05,
+      "loss": 0.6372,
+      "step": 11376
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 2.6299953710847092e-05,
+      "loss": 0.9081,
+      "step": 11377
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 2.6297382091241062e-05,
+      "loss": 0.6302,
+      "step": 11378
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 2.6294810471635038e-05,
+      "loss": 0.5528,
+      "step": 11379
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 2.629223885202901e-05,
+      "loss": 0.7456,
+      "step": 11380
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 2.628966723242298e-05,
+      "loss": 0.6977,
+      "step": 11381
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 2.6287095612816954e-05,
+      "loss": 0.5774,
+      "step": 11382
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 2.6284523993210924e-05,
+      "loss": 0.9048,
+      "step": 11383
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 2.6281952373604897e-05,
+      "loss": 0.795,
+      "step": 11384
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 2.6279380753998873e-05,
+      "loss": 0.7893,
+      "step": 11385
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 2.6276809134392843e-05,
+      "loss": 0.7654,
+      "step": 11386
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 2.6274237514786816e-05,
+      "loss": 0.6345,
+      "step": 11387
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 2.6271665895180786e-05,
+      "loss": 0.7848,
+      "step": 11388
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 2.626909427557476e-05,
+      "loss": 0.7683,
+      "step": 11389
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 2.626652265596873e-05,
+      "loss": 0.7796,
+      "step": 11390
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 2.6263951036362705e-05,
+      "loss": 0.7595,
+      "step": 11391
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 2.6261379416756675e-05,
+      "loss": 0.6009,
+      "step": 11392
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 2.6258807797150648e-05,
+      "loss": 0.7747,
+      "step": 11393
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 2.6256236177544617e-05,
+      "loss": 0.6949,
+      "step": 11394
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 2.625366455793859e-05,
+      "loss": 0.8465,
+      "step": 11395
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 2.625109293833256e-05,
+      "loss": 0.3675,
+      "step": 11396
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 2.6248521318726537e-05,
+      "loss": 0.7022,
+      "step": 11397
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 2.624594969912051e-05,
+      "loss": 0.8411,
+      "step": 11398
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 2.624337807951448e-05,
+      "loss": 0.5128,
+      "step": 11399
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 2.6240806459908452e-05,
+      "loss": 0.5438,
+      "step": 11400
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 2.6238234840302422e-05,
+      "loss": 0.6536,
+      "step": 11401
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 2.6235663220696395e-05,
+      "loss": 0.5862,
+      "step": 11402
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 2.623309160109037e-05,
+      "loss": 0.7905,
+      "step": 11403
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 2.623051998148434e-05,
+      "loss": 0.9563,
+      "step": 11404
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 2.6227948361878314e-05,
+      "loss": 0.8517,
+      "step": 11405
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 2.6225376742272284e-05,
+      "loss": 0.7508,
+      "step": 11406
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 2.6222805122666254e-05,
+      "loss": 1.052,
+      "step": 11407
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 2.6220233503060227e-05,
+      "loss": 0.4516,
+      "step": 11408
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 2.6217661883454203e-05,
+      "loss": 0.6068,
+      "step": 11409
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 2.6215090263848173e-05,
+      "loss": 0.5777,
+      "step": 11410
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 2.6212518644242146e-05,
+      "loss": 0.8722,
+      "step": 11411
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 2.6209947024636116e-05,
+      "loss": 0.5077,
+      "step": 11412
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 2.620737540503009e-05,
+      "loss": 0.7169,
+      "step": 11413
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 2.620480378542406e-05,
+      "loss": 0.6994,
+      "step": 11414
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 2.6202232165818035e-05,
+      "loss": 0.6462,
+      "step": 11415
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 2.6199660546212008e-05,
+      "loss": 0.4096,
+      "step": 11416
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 2.6197088926605978e-05,
+      "loss": 0.7208,
+      "step": 11417
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 2.619451730699995e-05,
+      "loss": 1.2339,
+      "step": 11418
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 2.619194568739392e-05,
+      "loss": 0.5285,
+      "step": 11419
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 2.6189374067787894e-05,
+      "loss": 0.668,
+      "step": 11420
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 2.618680244818187e-05,
+      "loss": 0.8948,
+      "step": 11421
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 2.618423082857584e-05,
+      "loss": 0.6497,
+      "step": 11422
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 2.618165920896981e-05,
+      "loss": 0.7896,
+      "step": 11423
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 2.6179087589363782e-05,
+      "loss": 0.5822,
+      "step": 11424
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 2.6176515969757752e-05,
+      "loss": 0.6909,
+      "step": 11425
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 2.6173944350151725e-05,
+      "loss": 0.4693,
+      "step": 11426
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 2.61713727305457e-05,
+      "loss": 0.7052,
+      "step": 11427
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 2.616880111093967e-05,
+      "loss": 0.5403,
+      "step": 11428
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 2.6166229491333644e-05,
+      "loss": 0.9506,
+      "step": 11429
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 2.6163657871727614e-05,
+      "loss": 0.5851,
+      "step": 11430
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 2.6161086252121587e-05,
+      "loss": 0.4742,
+      "step": 11431
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 2.6158514632515557e-05,
+      "loss": 0.5288,
+      "step": 11432
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 2.6155943012909533e-05,
+      "loss": 0.6565,
+      "step": 11433
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 2.6153371393303506e-05,
+      "loss": 0.6517,
+      "step": 11434
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 2.6150799773697476e-05,
+      "loss": 0.5026,
+      "step": 11435
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 2.614822815409145e-05,
+      "loss": 0.864,
+      "step": 11436
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 2.614565653448542e-05,
+      "loss": 0.5642,
+      "step": 11437
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 2.614308491487939e-05,
+      "loss": 0.8748,
+      "step": 11438
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 2.614051329527337e-05,
+      "loss": 0.9422,
+      "step": 11439
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 2.6137941675667338e-05,
+      "loss": 0.8474,
+      "step": 11440
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 2.6135370056061308e-05,
+      "loss": 0.5642,
+      "step": 11441
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 2.613279843645528e-05,
+      "loss": 0.9088,
+      "step": 11442
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 2.613022681684925e-05,
+      "loss": 0.7521,
+      "step": 11443
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 2.6127655197243224e-05,
+      "loss": 0.5636,
+      "step": 11444
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 2.61250835776372e-05,
+      "loss": 0.7493,
+      "step": 11445
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 2.612251195803117e-05,
+      "loss": 1.0533,
+      "step": 11446
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 2.6119940338425143e-05,
+      "loss": 0.659,
+      "step": 11447
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 2.6117368718819112e-05,
+      "loss": 0.7556,
+      "step": 11448
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 2.6114797099213086e-05,
+      "loss": 0.6186,
+      "step": 11449
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 2.6112225479607055e-05,
+      "loss": 0.5353,
+      "step": 11450
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 2.610965386000103e-05,
+      "loss": 0.4809,
+      "step": 11451
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 2.6107082240395005e-05,
+      "loss": 0.9296,
+      "step": 11452
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 2.6104510620788974e-05,
+      "loss": 0.6815,
+      "step": 11453
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 2.6101939001182944e-05,
+      "loss": 0.7787,
+      "step": 11454
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 2.6099367381576917e-05,
+      "loss": 0.7834,
+      "step": 11455
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 2.6096795761970887e-05,
+      "loss": 0.6479,
+      "step": 11456
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 2.6094224142364863e-05,
+      "loss": 0.7245,
+      "step": 11457
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 2.6091652522758836e-05,
+      "loss": 0.6965,
+      "step": 11458
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 2.6089080903152806e-05,
+      "loss": 0.5711,
+      "step": 11459
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 2.608650928354678e-05,
+      "loss": 0.4166,
+      "step": 11460
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 2.608393766394075e-05,
+      "loss": 0.6244,
+      "step": 11461
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 2.6081366044334722e-05,
+      "loss": 0.3863,
+      "step": 11462
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 2.60787944247287e-05,
+      "loss": 0.785,
+      "step": 11463
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 2.6076222805122668e-05,
+      "loss": 0.7202,
+      "step": 11464
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 2.607365118551664e-05,
+      "loss": 0.5262,
+      "step": 11465
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 2.607107956591061e-05,
+      "loss": 0.8113,
+      "step": 11466
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 2.6068507946304584e-05,
+      "loss": 0.8237,
+      "step": 11467
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 2.6065936326698554e-05,
+      "loss": 0.4636,
+      "step": 11468
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 2.606336470709253e-05,
+      "loss": 0.7411,
+      "step": 11469
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 2.60607930874865e-05,
+      "loss": 0.8051,
+      "step": 11470
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 2.6058221467880473e-05,
+      "loss": 0.6455,
+      "step": 11471
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 2.6055649848274443e-05,
+      "loss": 0.8531,
+      "step": 11472
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 2.6053078228668416e-05,
+      "loss": 0.55,
+      "step": 11473
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 2.6050506609062385e-05,
+      "loss": 0.6027,
+      "step": 11474
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 2.6047934989456362e-05,
+      "loss": 0.5949,
+      "step": 11475
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 2.6045363369850335e-05,
+      "loss": 0.5749,
+      "step": 11476
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 2.6042791750244304e-05,
+      "loss": 0.5393,
+      "step": 11477
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 2.6040220130638278e-05,
+      "loss": 0.5587,
+      "step": 11478
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 2.6037648511032247e-05,
+      "loss": 0.5504,
+      "step": 11479
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 2.603507689142622e-05,
+      "loss": 0.6609,
+      "step": 11480
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 2.6032505271820197e-05,
+      "loss": 0.6152,
+      "step": 11481
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 2.6029933652214166e-05,
+      "loss": 0.7081,
+      "step": 11482
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 2.602736203260814e-05,
+      "loss": 0.5762,
+      "step": 11483
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 2.602479041300211e-05,
+      "loss": 1.1399,
+      "step": 11484
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 2.602221879339608e-05,
+      "loss": 0.8473,
+      "step": 11485
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 2.6019647173790052e-05,
+      "loss": 0.6351,
+      "step": 11486
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 2.601707555418403e-05,
+      "loss": 0.5192,
+      "step": 11487
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 2.6014503934577998e-05,
+      "loss": 0.6298,
+      "step": 11488
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 2.601193231497197e-05,
+      "loss": 0.6783,
+      "step": 11489
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 2.600936069536594e-05,
+      "loss": 0.6699,
+      "step": 11490
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 2.6006789075759914e-05,
+      "loss": 0.7647,
+      "step": 11491
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 2.6004217456153884e-05,
+      "loss": 0.5972,
+      "step": 11492
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 2.600164583654786e-05,
+      "loss": 0.6559,
+      "step": 11493
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 2.5999074216941833e-05,
+      "loss": 0.6436,
+      "step": 11494
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 2.5996502597335803e-05,
+      "loss": 0.6952,
+      "step": 11495
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 2.5993930977729776e-05,
+      "loss": 0.8187,
+      "step": 11496
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 2.5991359358123746e-05,
+      "loss": 0.6884,
+      "step": 11497
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 2.5988787738517715e-05,
+      "loss": 0.773,
+      "step": 11498
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 2.5986216118911695e-05,
+      "loss": 0.6511,
+      "step": 11499
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 2.5983644499305665e-05,
+      "loss": 0.8332,
+      "step": 11500
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 2.5981072879699635e-05,
+      "loss": 0.4992,
+      "step": 11501
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 2.5978501260093608e-05,
+      "loss": 0.8612,
+      "step": 11502
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 2.5975929640487577e-05,
+      "loss": 0.6269,
+      "step": 11503
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 2.597335802088155e-05,
+      "loss": 0.505,
+      "step": 11504
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 2.5970786401275527e-05,
+      "loss": 0.8105,
+      "step": 11505
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 2.5968214781669497e-05,
+      "loss": 0.8909,
+      "step": 11506
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 2.596564316206347e-05,
+      "loss": 0.6679,
+      "step": 11507
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 2.596307154245744e-05,
+      "loss": 0.4526,
+      "step": 11508
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 2.5960499922851412e-05,
+      "loss": 0.7803,
+      "step": 11509
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 2.595792830324539e-05,
+      "loss": 0.529,
+      "step": 11510
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 2.595535668363936e-05,
+      "loss": 0.6161,
+      "step": 11511
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 2.595278506403333e-05,
+      "loss": 0.7468,
+      "step": 11512
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 2.59502134444273e-05,
+      "loss": 0.7447,
+      "step": 11513
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 2.5947641824821274e-05,
+      "loss": 0.9995,
+      "step": 11514
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 2.5945070205215244e-05,
+      "loss": 0.7466,
+      "step": 11515
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 2.594249858560922e-05,
+      "loss": 1.0034,
+      "step": 11516
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 2.593992696600319e-05,
+      "loss": 0.6049,
+      "step": 11517
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 2.5937355346397163e-05,
+      "loss": 0.6854,
+      "step": 11518
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 2.5934783726791133e-05,
+      "loss": 0.7145,
+      "step": 11519
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 2.5932212107185106e-05,
+      "loss": 0.4032,
+      "step": 11520
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 2.5929640487579076e-05,
+      "loss": 0.809,
+      "step": 11521
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 2.5927068867973052e-05,
+      "loss": 0.9912,
+      "step": 11522
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 2.5924497248367025e-05,
+      "loss": 0.896,
+      "step": 11523
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 2.5921925628760995e-05,
+      "loss": 0.6307,
+      "step": 11524
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 2.5919354009154968e-05,
+      "loss": 0.6748,
+      "step": 11525
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 2.5916782389548938e-05,
+      "loss": 0.66,
+      "step": 11526
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 2.591421076994291e-05,
+      "loss": 0.4941,
+      "step": 11527
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 2.5911639150336887e-05,
+      "loss": 0.6269,
+      "step": 11528
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 2.5909067530730857e-05,
+      "loss": 0.5909,
+      "step": 11529
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 2.590649591112483e-05,
+      "loss": 0.6965,
+      "step": 11530
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 2.59039242915188e-05,
+      "loss": 0.8397,
+      "step": 11531
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 2.590135267191277e-05,
+      "loss": 0.4837,
+      "step": 11532
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 2.5898781052306742e-05,
+      "loss": 0.4309,
+      "step": 11533
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 2.589620943270072e-05,
+      "loss": 0.4529,
+      "step": 11534
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 2.589363781309469e-05,
+      "loss": 0.6124,
+      "step": 11535
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 2.589106619348866e-05,
+      "loss": 0.6394,
+      "step": 11536
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 2.588849457388263e-05,
+      "loss": 0.6549,
+      "step": 11537
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 2.5885922954276604e-05,
+      "loss": 0.655,
+      "step": 11538
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 2.5883351334670574e-05,
+      "loss": 0.5864,
+      "step": 11539
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 2.588077971506455e-05,
+      "loss": 0.5912,
+      "step": 11540
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 2.5878208095458524e-05,
+      "loss": 0.7681,
+      "step": 11541
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 2.5875636475852493e-05,
+      "loss": 0.7525,
+      "step": 11542
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 2.5873064856246466e-05,
+      "loss": 0.5838,
+      "step": 11543
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 2.5870493236640436e-05,
+      "loss": 0.5852,
+      "step": 11544
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 2.586792161703441e-05,
+      "loss": 1.0469,
+      "step": 11545
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 2.5865349997428386e-05,
+      "loss": 0.7444,
+      "step": 11546
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 2.5862778377822355e-05,
+      "loss": 0.6624,
+      "step": 11547
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 2.5860206758216325e-05,
+      "loss": 0.4199,
+      "step": 11548
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 2.5857635138610298e-05,
+      "loss": 0.4281,
+      "step": 11549
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 2.5855063519004268e-05,
+      "loss": 0.7381,
+      "step": 11550
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 2.585249189939824e-05,
+      "loss": 0.68,
+      "step": 11551
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 2.5849920279792217e-05,
+      "loss": 0.8163,
+      "step": 11552
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 2.5847348660186187e-05,
+      "loss": 0.6125,
+      "step": 11553
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 2.584477704058016e-05,
+      "loss": 0.8785,
+      "step": 11554
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 2.584220542097413e-05,
+      "loss": 0.8887,
+      "step": 11555
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 2.5839633801368103e-05,
+      "loss": 0.472,
+      "step": 11556
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 2.5837062181762072e-05,
+      "loss": 0.7304,
+      "step": 11557
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 2.583449056215605e-05,
+      "loss": 0.6984,
+      "step": 11558
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 2.5831918942550022e-05,
+      "loss": 0.5034,
+      "step": 11559
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 2.582934732294399e-05,
+      "loss": 0.5922,
+      "step": 11560
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 2.5826775703337965e-05,
+      "loss": 0.7751,
+      "step": 11561
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 2.5824204083731934e-05,
+      "loss": 0.7819,
+      "step": 11562
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 2.5821632464125904e-05,
+      "loss": 0.693,
+      "step": 11563
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 2.5819060844519884e-05,
+      "loss": 0.5417,
+      "step": 11564
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 2.5816489224913854e-05,
+      "loss": 0.8353,
+      "step": 11565
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 2.5813917605307823e-05,
+      "loss": 0.6,
+      "step": 11566
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 2.5811345985701796e-05,
+      "loss": 0.5286,
+      "step": 11567
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 2.5808774366095766e-05,
+      "loss": 0.7967,
+      "step": 11568
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 2.580620274648974e-05,
+      "loss": 0.7734,
+      "step": 11569
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 2.5803631126883716e-05,
+      "loss": 0.5731,
+      "step": 11570
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 2.5801059507277685e-05,
+      "loss": 0.4888,
+      "step": 11571
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 2.579848788767166e-05,
+      "loss": 0.3969,
+      "step": 11572
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 2.5795916268065628e-05,
+      "loss": 1.0727,
+      "step": 11573
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 2.57933446484596e-05,
+      "loss": 0.9169,
+      "step": 11574
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 2.579077302885357e-05,
+      "loss": 0.8954,
+      "step": 11575
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 2.5788201409247547e-05,
+      "loss": 0.6479,
+      "step": 11576
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 2.578562978964152e-05,
+      "loss": 0.504,
+      "step": 11577
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 2.578305817003549e-05,
+      "loss": 0.8538,
+      "step": 11578
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 2.578048655042946e-05,
+      "loss": 0.5787,
+      "step": 11579
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 2.5777914930823433e-05,
+      "loss": 0.6455,
+      "step": 11580
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 2.5775343311217402e-05,
+      "loss": 0.9614,
+      "step": 11581
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 2.577277169161138e-05,
+      "loss": 0.6497,
+      "step": 11582
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 2.5770200072005352e-05,
+      "loss": 0.7631,
+      "step": 11583
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 2.5767628452399322e-05,
+      "loss": 0.6394,
+      "step": 11584
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 2.5765056832793295e-05,
+      "loss": 0.8671,
+      "step": 11585
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 2.5762485213187264e-05,
+      "loss": 0.5357,
+      "step": 11586
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 2.5759913593581238e-05,
+      "loss": 0.6821,
+      "step": 11587
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 2.5757341973975214e-05,
+      "loss": 0.7354,
+      "step": 11588
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 2.5754770354369184e-05,
+      "loss": 0.6072,
+      "step": 11589
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 2.5752198734763157e-05,
+      "loss": 0.8386,
+      "step": 11590
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 2.5749627115157126e-05,
+      "loss": 0.5121,
+      "step": 11591
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 2.57470554955511e-05,
+      "loss": 0.5398,
+      "step": 11592
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 2.574448387594507e-05,
+      "loss": 0.6636,
+      "step": 11593
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 2.5741912256339046e-05,
+      "loss": 0.4752,
+      "step": 11594
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 2.5739340636733015e-05,
+      "loss": 0.8086,
+      "step": 11595
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 2.573676901712699e-05,
+      "loss": 0.6891,
+      "step": 11596
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 2.5734197397520958e-05,
+      "loss": 0.6105,
+      "step": 11597
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 2.573162577791493e-05,
+      "loss": 0.6872,
+      "step": 11598
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 2.57290541583089e-05,
+      "loss": 0.4108,
+      "step": 11599
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 2.5726482538702877e-05,
+      "loss": 0.482,
+      "step": 11600
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 2.572391091909685e-05,
+      "loss": 0.6499,
+      "step": 11601
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 2.572133929949082e-05,
+      "loss": 0.8327,
+      "step": 11602
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 2.5718767679884793e-05,
+      "loss": 0.8171,
+      "step": 11603
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 2.5716196060278763e-05,
+      "loss": 0.7837,
+      "step": 11604
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 2.5713624440672736e-05,
+      "loss": 0.7303,
+      "step": 11605
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 2.5711052821066712e-05,
+      "loss": 0.6807,
+      "step": 11606
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 2.5708481201460682e-05,
+      "loss": 0.6477,
+      "step": 11607
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 2.5705909581854655e-05,
+      "loss": 0.5098,
+      "step": 11608
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 2.5703337962248625e-05,
+      "loss": 0.6451,
+      "step": 11609
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 2.5700766342642594e-05,
+      "loss": 0.7117,
+      "step": 11610
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 2.5698194723036568e-05,
+      "loss": 0.788,
+      "step": 11611
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 2.5695623103430544e-05,
+      "loss": 0.5772,
+      "step": 11612
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 2.5693051483824514e-05,
+      "loss": 0.7456,
+      "step": 11613
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 2.5690479864218487e-05,
+      "loss": 0.6412,
+      "step": 11614
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 2.5687908244612456e-05,
+      "loss": 0.8215,
+      "step": 11615
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 2.568533662500643e-05,
+      "loss": 0.5203,
+      "step": 11616
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 2.56827650054004e-05,
+      "loss": 0.6464,
+      "step": 11617
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 2.5680193385794376e-05,
+      "loss": 0.8141,
+      "step": 11618
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 2.567762176618835e-05,
+      "loss": 0.4363,
+      "step": 11619
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 2.567505014658232e-05,
+      "loss": 0.5946,
+      "step": 11620
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 2.567247852697629e-05,
+      "loss": 1.1271,
+      "step": 11621
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 2.566990690737026e-05,
+      "loss": 0.6726,
+      "step": 11622
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 2.566733528776423e-05,
+      "loss": 0.6048,
+      "step": 11623
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 2.566476366815821e-05,
+      "loss": 0.6443,
+      "step": 11624
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 2.566219204855218e-05,
+      "loss": 1.0095,
+      "step": 11625
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 2.565962042894615e-05,
+      "loss": 0.7871,
+      "step": 11626
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 2.5657048809340123e-05,
+      "loss": 0.5892,
+      "step": 11627
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 2.5654477189734093e-05,
+      "loss": 0.6528,
+      "step": 11628
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 2.5651905570128066e-05,
+      "loss": 0.5421,
+      "step": 11629
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 2.5649333950522042e-05,
+      "loss": 0.4955,
+      "step": 11630
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 2.5646762330916012e-05,
+      "loss": 0.4493,
+      "step": 11631
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 2.5644190711309985e-05,
+      "loss": 1.0509,
+      "step": 11632
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 2.5641619091703955e-05,
+      "loss": 0.8935,
+      "step": 11633
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 2.5639047472097928e-05,
+      "loss": 1.0095,
+      "step": 11634
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 2.5636475852491898e-05,
+      "loss": 0.6098,
+      "step": 11635
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 2.5633904232885874e-05,
+      "loss": 0.6224,
+      "step": 11636
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 2.5631332613279847e-05,
+      "loss": 0.6666,
+      "step": 11637
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 2.5628760993673817e-05,
+      "loss": 0.7703,
+      "step": 11638
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 2.562618937406779e-05,
+      "loss": 0.77,
+      "step": 11639
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 2.562361775446176e-05,
+      "loss": 0.7236,
+      "step": 11640
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 2.562104613485573e-05,
+      "loss": 0.6984,
+      "step": 11641
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 2.5618474515249706e-05,
+      "loss": 0.9429,
+      "step": 11642
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 2.561590289564368e-05,
+      "loss": 0.6297,
+      "step": 11643
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 2.561333127603765e-05,
+      "loss": 0.6417,
+      "step": 11644
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 2.561075965643162e-05,
+      "loss": 0.7279,
+      "step": 11645
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 2.560818803682559e-05,
+      "loss": 0.8148,
+      "step": 11646
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 2.5605616417219564e-05,
+      "loss": 0.5982,
+      "step": 11647
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 2.560304479761354e-05,
+      "loss": 0.7495,
+      "step": 11648
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 2.560047317800751e-05,
+      "loss": 0.7733,
+      "step": 11649
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 2.5597901558401484e-05,
+      "loss": 0.8511,
+      "step": 11650
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 2.5595329938795453e-05,
+      "loss": 0.8779,
+      "step": 11651
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 2.5592758319189426e-05,
+      "loss": 0.7291,
+      "step": 11652
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 2.5590186699583396e-05,
+      "loss": 0.6426,
+      "step": 11653
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 2.5587615079977372e-05,
+      "loss": 0.6044,
+      "step": 11654
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 2.5585043460371346e-05,
+      "loss": 0.7564,
+      "step": 11655
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 2.5582471840765315e-05,
+      "loss": 0.6544,
+      "step": 11656
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 2.5579900221159285e-05,
+      "loss": 0.5073,
+      "step": 11657
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 2.5577328601553258e-05,
+      "loss": 0.584,
+      "step": 11658
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 2.5574756981947234e-05,
+      "loss": 0.6458,
+      "step": 11659
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 2.5572185362341204e-05,
+      "loss": 0.5817,
+      "step": 11660
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 2.5569613742735177e-05,
+      "loss": 0.6967,
+      "step": 11661
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 2.5567042123129147e-05,
+      "loss": 0.6168,
+      "step": 11662
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 2.556447050352312e-05,
+      "loss": 0.6452,
+      "step": 11663
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 2.556189888391709e-05,
+      "loss": 1.0817,
+      "step": 11664
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 2.5559327264311066e-05,
+      "loss": 0.9965,
+      "step": 11665
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 2.555675564470504e-05,
+      "loss": 0.5378,
+      "step": 11666
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 2.555418402509901e-05,
+      "loss": 0.8335,
+      "step": 11667
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 2.5551612405492982e-05,
+      "loss": 0.8337,
+      "step": 11668
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 2.554904078588695e-05,
+      "loss": 0.8045,
+      "step": 11669
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 2.5546469166280925e-05,
+      "loss": 0.5458,
+      "step": 11670
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 2.55438975466749e-05,
+      "loss": 1.0549,
+      "step": 11671
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 2.554132592706887e-05,
+      "loss": 0.6802,
+      "step": 11672
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 2.553875430746284e-05,
+      "loss": 0.7416,
+      "step": 11673
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 2.5536182687856814e-05,
+      "loss": 0.6516,
+      "step": 11674
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 2.5533611068250783e-05,
+      "loss": 1.0539,
+      "step": 11675
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 2.5531039448644756e-05,
+      "loss": 0.9613,
+      "step": 11676
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 2.5528467829038733e-05,
+      "loss": 0.8911,
+      "step": 11677
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 2.5525896209432703e-05,
+      "loss": 0.476,
+      "step": 11678
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 2.5523324589826676e-05,
+      "loss": 0.6798,
+      "step": 11679
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 2.5520752970220645e-05,
+      "loss": 0.6182,
+      "step": 11680
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 2.551818135061462e-05,
+      "loss": 0.7097,
+      "step": 11681
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 2.5515609731008588e-05,
+      "loss": 0.8575,
+      "step": 11682
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 2.5513038111402564e-05,
+      "loss": 0.664,
+      "step": 11683
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 2.5510466491796538e-05,
+      "loss": 1.0075,
+      "step": 11684
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 2.5507894872190507e-05,
+      "loss": 0.8956,
+      "step": 11685
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 2.550532325258448e-05,
+      "loss": 0.8823,
+      "step": 11686
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 2.550275163297845e-05,
+      "loss": 0.6761,
+      "step": 11687
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 2.550018001337242e-05,
+      "loss": 0.7532,
+      "step": 11688
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 2.54976083937664e-05,
+      "loss": 0.6692,
+      "step": 11689
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 2.549503677416037e-05,
+      "loss": 0.6225,
+      "step": 11690
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 2.549246515455434e-05,
+      "loss": 0.7017,
+      "step": 11691
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 2.5489893534948312e-05,
+      "loss": 0.6235,
+      "step": 11692
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 2.548732191534228e-05,
+      "loss": 0.5715,
+      "step": 11693
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 2.5484750295736255e-05,
+      "loss": 0.7064,
+      "step": 11694
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 2.548217867613023e-05,
+      "loss": 0.8224,
+      "step": 11695
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 2.54796070565242e-05,
+      "loss": 0.9694,
+      "step": 11696
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 2.5477035436918174e-05,
+      "loss": 0.8271,
+      "step": 11697
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 2.5474463817312144e-05,
+      "loss": 0.6317,
+      "step": 11698
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 2.5471892197706117e-05,
+      "loss": 0.7342,
+      "step": 11699
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 2.5469320578100086e-05,
+      "loss": 0.6589,
+      "step": 11700
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 2.5466748958494063e-05,
+      "loss": 0.9061,
+      "step": 11701
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 2.5464177338888036e-05,
+      "loss": 0.6355,
+      "step": 11702
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 2.5461605719282006e-05,
+      "loss": 0.5685,
+      "step": 11703
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 2.5459034099675975e-05,
+      "loss": 0.895,
+      "step": 11704
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 2.545646248006995e-05,
+      "loss": 0.4707,
+      "step": 11705
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 2.5453890860463918e-05,
+      "loss": 0.6618,
+      "step": 11706
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 2.5451319240857895e-05,
+      "loss": 0.6236,
+      "step": 11707
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 2.5448747621251868e-05,
+      "loss": 0.6632,
+      "step": 11708
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 2.5446176001645837e-05,
+      "loss": 0.9224,
+      "step": 11709
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 2.544360438203981e-05,
+      "loss": 0.7496,
+      "step": 11710
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 2.544103276243378e-05,
+      "loss": 0.6577,
+      "step": 11711
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 2.5438461142827753e-05,
+      "loss": 1.4478,
+      "step": 11712
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 2.543588952322173e-05,
+      "loss": 0.8335,
+      "step": 11713
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 2.54333179036157e-05,
+      "loss": 0.5935,
+      "step": 11714
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 2.5430746284009672e-05,
+      "loss": 0.6555,
+      "step": 11715
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 2.5428174664403642e-05,
+      "loss": 0.8436,
+      "step": 11716
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 2.5425603044797615e-05,
+      "loss": 0.6456,
+      "step": 11717
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 2.5423031425191585e-05,
+      "loss": 0.6727,
+      "step": 11718
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 2.542045980558556e-05,
+      "loss": 0.9191,
+      "step": 11719
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 2.541788818597953e-05,
+      "loss": 0.7098,
+      "step": 11720
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 2.5415316566373504e-05,
+      "loss": 0.4924,
+      "step": 11721
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 2.5412744946767474e-05,
+      "loss": 0.7206,
+      "step": 11722
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 2.5410173327161447e-05,
+      "loss": 0.7604,
+      "step": 11723
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 2.5407601707555416e-05,
+      "loss": 0.8518,
+      "step": 11724
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 2.5405030087949393e-05,
+      "loss": 0.855,
+      "step": 11725
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 2.5402458468343366e-05,
+      "loss": 0.7122,
+      "step": 11726
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 2.5399886848737336e-05,
+      "loss": 0.7565,
+      "step": 11727
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 2.539731522913131e-05,
+      "loss": 0.9186,
+      "step": 11728
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 2.539474360952528e-05,
+      "loss": 0.4466,
+      "step": 11729
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 2.539217198991925e-05,
+      "loss": 0.9072,
+      "step": 11730
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 2.5389600370313228e-05,
+      "loss": 0.8316,
+      "step": 11731
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 2.5387028750707198e-05,
+      "loss": 0.7505,
+      "step": 11732
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 2.538445713110117e-05,
+      "loss": 0.726,
+      "step": 11733
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 2.538188551149514e-05,
+      "loss": 0.6598,
+      "step": 11734
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 2.537931389188911e-05,
+      "loss": 0.7744,
+      "step": 11735
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 2.5376742272283083e-05,
+      "loss": 0.9216,
+      "step": 11736
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 2.537417065267706e-05,
+      "loss": 0.9932,
+      "step": 11737
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 2.537159903307103e-05,
+      "loss": 0.5858,
+      "step": 11738
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 2.5369027413465002e-05,
+      "loss": 0.3502,
+      "step": 11739
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 2.5366455793858972e-05,
+      "loss": 0.7739,
+      "step": 11740
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 2.5363884174252945e-05,
+      "loss": 0.6807,
+      "step": 11741
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 2.5361312554646915e-05,
+      "loss": 0.7547,
+      "step": 11742
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 2.535874093504089e-05,
+      "loss": 0.7898,
+      "step": 11743
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 2.5356169315434864e-05,
+      "loss": 1.0369,
+      "step": 11744
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 2.5353597695828834e-05,
+      "loss": 0.4159,
+      "step": 11745
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 2.5351026076222807e-05,
+      "loss": 0.6851,
+      "step": 11746
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 2.5348454456616777e-05,
+      "loss": 0.5869,
+      "step": 11747
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 2.5345882837010746e-05,
+      "loss": 1.0391,
+      "step": 11748
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 2.5343311217404726e-05,
+      "loss": 0.6194,
+      "step": 11749
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 2.5340739597798696e-05,
+      "loss": 0.6408,
+      "step": 11750
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 2.5338167978192666e-05,
+      "loss": 0.6384,
+      "step": 11751
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 2.533559635858664e-05,
+      "loss": 0.8507,
+      "step": 11752
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 2.533302473898061e-05,
+      "loss": 0.6395,
+      "step": 11753
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 2.533045311937458e-05,
+      "loss": 0.5716,
+      "step": 11754
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 2.5327881499768558e-05,
+      "loss": 0.7259,
+      "step": 11755
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 2.5325309880162528e-05,
+      "loss": 1.2892,
+      "step": 11756
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 2.53227382605565e-05,
+      "loss": 0.6072,
+      "step": 11757
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 2.532016664095047e-05,
+      "loss": 0.7256,
+      "step": 11758
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 2.5317595021344444e-05,
+      "loss": 0.7771,
+      "step": 11759
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 2.5315023401738413e-05,
+      "loss": 0.7137,
+      "step": 11760
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 2.531245178213239e-05,
+      "loss": 0.7531,
+      "step": 11761
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 2.5309880162526363e-05,
+      "loss": 0.6055,
+      "step": 11762
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 2.5307308542920332e-05,
+      "loss": 0.6348,
+      "step": 11763
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 2.5304736923314306e-05,
+      "loss": 0.7432,
+      "step": 11764
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 2.5302165303708275e-05,
+      "loss": 0.7113,
+      "step": 11765
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 2.5299593684102245e-05,
+      "loss": 0.7851,
+      "step": 11766
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 2.5297022064496225e-05,
+      "loss": 0.6311,
+      "step": 11767
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 2.5294450444890194e-05,
+      "loss": 0.7745,
+      "step": 11768
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 2.5291878825284164e-05,
+      "loss": 0.6623,
+      "step": 11769
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 2.5289307205678137e-05,
+      "loss": 0.8982,
+      "step": 11770
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 2.5286735586072107e-05,
+      "loss": 0.7234,
+      "step": 11771
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 2.528416396646608e-05,
+      "loss": 0.7938,
+      "step": 11772
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 2.5281592346860056e-05,
+      "loss": 0.5919,
+      "step": 11773
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 2.5279020727254026e-05,
+      "loss": 0.6215,
+      "step": 11774
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 2.5276449107648e-05,
+      "loss": 0.8681,
+      "step": 11775
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 2.527387748804197e-05,
+      "loss": 0.5997,
+      "step": 11776
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 2.5271305868435942e-05,
+      "loss": 0.4883,
+      "step": 11777
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 2.526873424882991e-05,
+      "loss": 0.7683,
+      "step": 11778
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 2.5266162629223888e-05,
+      "loss": 0.7318,
+      "step": 11779
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 2.526359100961786e-05,
+      "loss": 0.64,
+      "step": 11780
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 2.526101939001183e-05,
+      "loss": 0.9725,
+      "step": 11781
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 2.52584477704058e-05,
+      "loss": 0.7599,
+      "step": 11782
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 2.5255876150799774e-05,
+      "loss": 0.8994,
+      "step": 11783
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 2.5253304531193743e-05,
+      "loss": 0.8619,
+      "step": 11784
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 2.525073291158772e-05,
+      "loss": 1.0025,
+      "step": 11785
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 2.5248161291981693e-05,
+      "loss": 0.561,
+      "step": 11786
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 2.5245589672375662e-05,
+      "loss": 0.609,
+      "step": 11787
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 2.5243018052769636e-05,
+      "loss": 0.9922,
+      "step": 11788
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 2.5240446433163605e-05,
+      "loss": 0.5881,
+      "step": 11789
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 2.5237874813557578e-05,
+      "loss": 0.9639,
+      "step": 11790
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 2.5235303193951555e-05,
+      "loss": 0.8671,
+      "step": 11791
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 2.5232731574345524e-05,
+      "loss": 0.766,
+      "step": 11792
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 2.5230159954739498e-05,
+      "loss": 0.7901,
+      "step": 11793
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 2.5227588335133467e-05,
+      "loss": 0.6832,
+      "step": 11794
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 2.522501671552744e-05,
+      "loss": 0.8183,
+      "step": 11795
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 2.522244509592141e-05,
+      "loss": 0.6453,
+      "step": 11796
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 2.5219873476315386e-05,
+      "loss": 0.7997,
+      "step": 11797
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 2.5217301856709356e-05,
+      "loss": 0.4305,
+      "step": 11798
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 2.521473023710333e-05,
+      "loss": 0.8572,
+      "step": 11799
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 2.52121586174973e-05,
+      "loss": 0.6008,
+      "step": 11800
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 2.5209586997891272e-05,
+      "loss": 0.5273,
+      "step": 11801
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 2.520701537828524e-05,
+      "loss": 0.9396,
+      "step": 11802
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 2.5204443758679218e-05,
+      "loss": 0.7799,
+      "step": 11803
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 2.520187213907319e-05,
+      "loss": 0.5048,
+      "step": 11804
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 2.519930051946716e-05,
+      "loss": 0.769,
+      "step": 11805
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 2.5196728899861134e-05,
+      "loss": 0.6799,
+      "step": 11806
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 2.5194157280255104e-05,
+      "loss": 0.5961,
+      "step": 11807
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 2.519158566064908e-05,
+      "loss": 0.6278,
+      "step": 11808
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 2.5189014041043053e-05,
+      "loss": 0.5663,
+      "step": 11809
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 2.5186442421437023e-05,
+      "loss": 0.6128,
+      "step": 11810
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 2.5183870801830996e-05,
+      "loss": 0.5373,
+      "step": 11811
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 2.5181299182224966e-05,
+      "loss": 0.8669,
+      "step": 11812
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 2.5178727562618935e-05,
+      "loss": 0.4841,
+      "step": 11813
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 2.5176155943012915e-05,
+      "loss": 0.6011,
+      "step": 11814
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 2.5173584323406885e-05,
+      "loss": 0.6037,
+      "step": 11815
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 2.5171012703800854e-05,
+      "loss": 0.7163,
+      "step": 11816
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 2.5168441084194828e-05,
+      "loss": 0.6163,
+      "step": 11817
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 2.5165869464588797e-05,
+      "loss": 0.6744,
+      "step": 11818
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 2.516329784498277e-05,
+      "loss": 0.5896,
+      "step": 11819
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 2.5160726225376747e-05,
+      "loss": 0.9659,
+      "step": 11820
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 2.5158154605770716e-05,
+      "loss": 0.6483,
+      "step": 11821
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 2.515558298616469e-05,
+      "loss": 1.1113,
+      "step": 11822
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 2.515301136655866e-05,
+      "loss": 0.5592,
+      "step": 11823
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 2.5150439746952632e-05,
+      "loss": 0.4486,
+      "step": 11824
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 2.5147868127346602e-05,
+      "loss": 0.7323,
+      "step": 11825
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 2.514529650774058e-05,
+      "loss": 0.3918,
+      "step": 11826
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 2.514272488813455e-05,
+      "loss": 0.8671,
+      "step": 11827
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 2.514015326852852e-05,
+      "loss": 0.7643,
+      "step": 11828
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 2.513758164892249e-05,
+      "loss": 1.0245,
+      "step": 11829
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 2.5135010029316464e-05,
+      "loss": 0.8184,
+      "step": 11830
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 2.5132438409710434e-05,
+      "loss": 0.5303,
+      "step": 11831
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 2.512986679010441e-05,
+      "loss": 0.8078,
+      "step": 11832
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 2.5127295170498383e-05,
+      "loss": 0.5852,
+      "step": 11833
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 2.5124723550892353e-05,
+      "loss": 1.1349,
+      "step": 11834
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 2.5122151931286326e-05,
+      "loss": 0.5705,
+      "step": 11835
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 2.5119580311680296e-05,
+      "loss": 0.5282,
+      "step": 11836
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 2.511700869207427e-05,
+      "loss": 0.8327,
+      "step": 11837
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 2.5114437072468245e-05,
+      "loss": 0.9072,
+      "step": 11838
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 2.5111865452862215e-05,
+      "loss": 0.6332,
+      "step": 11839
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 2.5109293833256188e-05,
+      "loss": 0.746,
+      "step": 11840
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 2.5106722213650158e-05,
+      "loss": 0.8463,
+      "step": 11841
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 2.510415059404413e-05,
+      "loss": 0.659,
+      "step": 11842
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 2.51015789744381e-05,
+      "loss": 0.6515,
+      "step": 11843
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 2.5099007354832077e-05,
+      "loss": 0.6101,
+      "step": 11844
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 2.5096435735226047e-05,
+      "loss": 1.1806,
+      "step": 11845
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 2.509386411562002e-05,
+      "loss": 0.5774,
+      "step": 11846
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 2.509129249601399e-05,
+      "loss": 0.5618,
+      "step": 11847
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 2.5088720876407962e-05,
+      "loss": 0.6301,
+      "step": 11848
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 2.5086149256801932e-05,
+      "loss": 0.5299,
+      "step": 11849
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 2.508357763719591e-05,
+      "loss": 0.9477,
+      "step": 11850
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 2.508100601758988e-05,
+      "loss": 0.8839,
+      "step": 11851
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 2.507843439798385e-05,
+      "loss": 0.4988,
+      "step": 11852
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 2.5075862778377824e-05,
+      "loss": 0.4897,
+      "step": 11853
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 2.5073291158771794e-05,
+      "loss": 0.5187,
+      "step": 11854
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 2.5070719539165767e-05,
+      "loss": 0.9181,
+      "step": 11855
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 2.5068147919559744e-05,
+      "loss": 0.6497,
+      "step": 11856
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 2.5065576299953713e-05,
+      "loss": 0.8703,
+      "step": 11857
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 2.5063004680347686e-05,
+      "loss": 0.8559,
+      "step": 11858
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 2.5060433060741656e-05,
+      "loss": 0.8288,
+      "step": 11859
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 2.5057861441135626e-05,
+      "loss": 0.8637,
+      "step": 11860
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 2.50552898215296e-05,
+      "loss": 0.526,
+      "step": 11861
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 2.5052718201923575e-05,
+      "loss": 0.5488,
+      "step": 11862
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 2.5050146582317545e-05,
+      "loss": 0.7874,
+      "step": 11863
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 2.5047574962711518e-05,
+      "loss": 0.6998,
+      "step": 11864
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 2.5045003343105488e-05,
+      "loss": 0.9033,
+      "step": 11865
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 2.504243172349946e-05,
+      "loss": 0.887,
+      "step": 11866
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 2.503986010389343e-05,
+      "loss": 0.7288,
+      "step": 11867
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 2.5037288484287407e-05,
+      "loss": 0.6462,
+      "step": 11868
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 2.503471686468138e-05,
+      "loss": 0.5848,
+      "step": 11869
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 2.503214524507535e-05,
+      "loss": 0.8118,
+      "step": 11870
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 2.5029573625469323e-05,
+      "loss": 0.6821,
+      "step": 11871
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 2.5027002005863292e-05,
+      "loss": 0.8057,
+      "step": 11872
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 2.5024430386257262e-05,
+      "loss": 0.9855,
+      "step": 11873
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 2.5021858766651242e-05,
+      "loss": 0.7368,
+      "step": 11874
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 2.501928714704521e-05,
+      "loss": 0.5126,
+      "step": 11875
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 2.501671552743918e-05,
+      "loss": 0.6117,
+      "step": 11876
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 2.5014143907833154e-05,
+      "loss": 0.5824,
+      "step": 11877
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 2.5011572288227124e-05,
+      "loss": 0.5777,
+      "step": 11878
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 2.5009000668621097e-05,
+      "loss": 0.5757,
+      "step": 11879
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 2.5006429049015074e-05,
+      "loss": 0.5055,
+      "step": 11880
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 2.5003857429409043e-05,
+      "loss": 0.6951,
+      "step": 11881
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 2.5001285809803016e-05,
+      "loss": 0.7954,
+      "step": 11882
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 2.4998714190196986e-05,
+      "loss": 0.7547,
+      "step": 11883
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 2.499614257059096e-05,
+      "loss": 0.506,
+      "step": 11884
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 2.4993570950984932e-05,
+      "loss": 0.4346,
+      "step": 11885
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 2.4990999331378902e-05,
+      "loss": 1.3637,
+      "step": 11886
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 2.498842771177288e-05,
+      "loss": 0.5005,
+      "step": 11887
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 2.4985856092166848e-05,
+      "loss": 0.7006,
+      "step": 11888
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 2.498328447256082e-05,
+      "loss": 0.8074,
+      "step": 11889
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 2.4980712852954794e-05,
+      "loss": 0.5316,
+      "step": 11890
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 2.4978141233348764e-05,
+      "loss": 0.7206,
+      "step": 11891
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 2.4975569613742737e-05,
+      "loss": 0.3438,
+      "step": 11892
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 2.497299799413671e-05,
+      "loss": 0.636,
+      "step": 11893
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 2.497042637453068e-05,
+      "loss": 0.7688,
+      "step": 11894
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 2.4967854754924653e-05,
+      "loss": 0.8141,
+      "step": 11895
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 2.4965283135318626e-05,
+      "loss": 0.6087,
+      "step": 11896
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 2.49627115157126e-05,
+      "loss": 0.7026,
+      "step": 11897
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 2.496013989610657e-05,
+      "loss": 0.6772,
+      "step": 11898
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 2.495756827650054e-05,
+      "loss": 0.6991,
+      "step": 11899
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 2.4954996656894515e-05,
+      "loss": 0.8559,
+      "step": 11900
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 2.4952425037288484e-05,
+      "loss": 0.6726,
+      "step": 11901
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 2.4949853417682457e-05,
+      "loss": 0.8102,
+      "step": 11902
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 2.494728179807643e-05,
+      "loss": 0.7904,
+      "step": 11903
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 2.49447101784704e-05,
+      "loss": 0.5839,
+      "step": 11904
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 2.4942138558864377e-05,
+      "loss": 0.7892,
+      "step": 11905
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 2.4939566939258346e-05,
+      "loss": 0.5336,
+      "step": 11906
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 2.4936995319652316e-05,
+      "loss": 0.6241,
+      "step": 11907
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 2.4934423700046293e-05,
+      "loss": 0.6912,
+      "step": 11908
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 2.4931852080440262e-05,
+      "loss": 1.0732,
+      "step": 11909
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 2.4929280460834235e-05,
+      "loss": 0.6193,
+      "step": 11910
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 2.492670884122821e-05,
+      "loss": 0.6153,
+      "step": 11911
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 2.4924137221622178e-05,
+      "loss": 0.5597,
+      "step": 11912
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 2.492156560201615e-05,
+      "loss": 0.3765,
+      "step": 11913
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 2.4918993982410124e-05,
+      "loss": 0.6586,
+      "step": 11914
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 2.4916422362804094e-05,
+      "loss": 0.7202,
+      "step": 11915
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 2.4913850743198067e-05,
+      "loss": 0.6868,
+      "step": 11916
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 2.491127912359204e-05,
+      "loss": 0.6922,
+      "step": 11917
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 2.4908707503986013e-05,
+      "loss": 0.8605,
+      "step": 11918
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 2.4906135884379983e-05,
+      "loss": 0.6813,
+      "step": 11919
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 2.4903564264773956e-05,
+      "loss": 0.8596,
+      "step": 11920
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 2.490099264516793e-05,
+      "loss": 0.7216,
+      "step": 11921
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 2.48984210255619e-05,
+      "loss": 0.488,
+      "step": 11922
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 2.4895849405955872e-05,
+      "loss": 0.8967,
+      "step": 11923
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 2.4893277786349845e-05,
+      "loss": 0.6249,
+      "step": 11924
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 2.4890706166743814e-05,
+      "loss": 0.5473,
+      "step": 11925
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 2.488813454713779e-05,
+      "loss": 0.7543,
+      "step": 11926
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 2.488556292753176e-05,
+      "loss": 0.7511,
+      "step": 11927
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 2.4882991307925734e-05,
+      "loss": 0.8689,
+      "step": 11928
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 2.4880419688319707e-05,
+      "loss": 0.384,
+      "step": 11929
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 2.4877848068713676e-05,
+      "loss": 0.5774,
+      "step": 11930
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 2.487527644910765e-05,
+      "loss": 0.3733,
+      "step": 11931
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 2.4872704829501623e-05,
+      "loss": 0.5866,
+      "step": 11932
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 2.4870133209895592e-05,
+      "loss": 0.6778,
+      "step": 11933
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 2.4867561590289565e-05,
+      "loss": 0.5162,
+      "step": 11934
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 2.486498997068354e-05,
+      "loss": 0.8154,
+      "step": 11935
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 2.486241835107751e-05,
+      "loss": 0.7724,
+      "step": 11936
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 2.485984673147148e-05,
+      "loss": 0.667,
+      "step": 11937
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 2.4857275111865454e-05,
+      "loss": 0.6054,
+      "step": 11938
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 2.4854703492259427e-05,
+      "loss": 0.5487,
+      "step": 11939
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 2.4852131872653397e-05,
+      "loss": 0.4599,
+      "step": 11940
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 2.484956025304737e-05,
+      "loss": 0.7751,
+      "step": 11941
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 2.4846988633441343e-05,
+      "loss": 0.6416,
+      "step": 11942
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 2.4844417013835313e-05,
+      "loss": 0.7063,
+      "step": 11943
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 2.484184539422929e-05,
+      "loss": 1.0019,
+      "step": 11944
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 2.483927377462326e-05,
+      "loss": 1.1075,
+      "step": 11945
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 2.483670215501723e-05,
+      "loss": 0.9167,
+      "step": 11946
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 2.4834130535411205e-05,
+      "loss": 0.4786,
+      "step": 11947
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 2.4831558915805175e-05,
+      "loss": 0.7413,
+      "step": 11948
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 2.4828987296199148e-05,
+      "loss": 0.6433,
+      "step": 11949
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 2.482641567659312e-05,
+      "loss": 0.3532,
+      "step": 11950
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 2.482384405698709e-05,
+      "loss": 0.6554,
+      "step": 11951
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 2.4821272437381064e-05,
+      "loss": 0.6309,
+      "step": 11952
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 2.4818700817775037e-05,
+      "loss": 0.646,
+      "step": 11953
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 2.4816129198169006e-05,
+      "loss": 0.7335,
+      "step": 11954
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 2.481355757856298e-05,
+      "loss": 0.9251,
+      "step": 11955
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 2.4810985958956953e-05,
+      "loss": 0.7376,
+      "step": 11956
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 2.4808414339350926e-05,
+      "loss": 0.603,
+      "step": 11957
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 2.4805842719744895e-05,
+      "loss": 0.614,
+      "step": 11958
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 2.480327110013887e-05,
+      "loss": 0.7081,
+      "step": 11959
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 2.480069948053284e-05,
+      "loss": 0.6298,
+      "step": 11960
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 2.479812786092681e-05,
+      "loss": 0.8915,
+      "step": 11961
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 2.4795556241320784e-05,
+      "loss": 0.7889,
+      "step": 11962
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 2.4792984621714757e-05,
+      "loss": 0.6207,
+      "step": 11963
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 2.4790413002108727e-05,
+      "loss": 0.77,
+      "step": 11964
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 2.4787841382502704e-05,
+      "loss": 0.4935,
+      "step": 11965
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 2.4785269762896673e-05,
+      "loss": 0.688,
+      "step": 11966
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 2.4782698143290646e-05,
+      "loss": 0.665,
+      "step": 11967
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 2.478012652368462e-05,
+      "loss": 0.8443,
+      "step": 11968
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 2.477755490407859e-05,
+      "loss": 0.3984,
+      "step": 11969
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 2.4774983284472562e-05,
+      "loss": 0.8173,
+      "step": 11970
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 2.4772411664866535e-05,
+      "loss": 0.8061,
+      "step": 11971
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 2.4769840045260505e-05,
+      "loss": 0.5724,
+      "step": 11972
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 2.4767268425654478e-05,
+      "loss": 0.6109,
+      "step": 11973
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 2.476469680604845e-05,
+      "loss": 0.9312,
+      "step": 11974
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 2.4762125186442424e-05,
+      "loss": 0.8923,
+      "step": 11975
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 2.4759553566836394e-05,
+      "loss": 0.5256,
+      "step": 11976
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 2.4756981947230367e-05,
+      "loss": 0.7937,
+      "step": 11977
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 2.475441032762434e-05,
+      "loss": 0.7426,
+      "step": 11978
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 2.475183870801831e-05,
+      "loss": 0.6049,
+      "step": 11979
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 2.4749267088412283e-05,
+      "loss": 0.7418,
+      "step": 11980
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 2.4746695468806256e-05,
+      "loss": 0.6809,
+      "step": 11981
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 2.4744123849200225e-05,
+      "loss": 0.5682,
+      "step": 11982
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 2.4741552229594202e-05,
+      "loss": 0.6892,
+      "step": 11983
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 2.473898060998817e-05,
+      "loss": 0.6898,
+      "step": 11984
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 2.473640899038214e-05,
+      "loss": 0.9185,
+      "step": 11985
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 2.4733837370776118e-05,
+      "loss": 0.5101,
+      "step": 11986
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 2.4731265751170087e-05,
+      "loss": 0.8256,
+      "step": 11987
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 2.472869413156406e-05,
+      "loss": 0.4891,
+      "step": 11988
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 2.4726122511958034e-05,
+      "loss": 0.7295,
+      "step": 11989
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 2.4723550892352003e-05,
+      "loss": 0.4875,
+      "step": 11990
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 2.4720979272745976e-05,
+      "loss": 0.7289,
+      "step": 11991
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 2.471840765313995e-05,
+      "loss": 0.8498,
+      "step": 11992
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 2.471583603353392e-05,
+      "loss": 0.8026,
+      "step": 11993
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 2.4713264413927892e-05,
+      "loss": 0.8859,
+      "step": 11994
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 2.4710692794321865e-05,
+      "loss": 0.7849,
+      "step": 11995
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 2.4708121174715838e-05,
+      "loss": 0.744,
+      "step": 11996
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 2.4705549555109808e-05,
+      "loss": 0.7454,
+      "step": 11997
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 2.470297793550378e-05,
+      "loss": 0.9147,
+      "step": 11998
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 2.4700406315897754e-05,
+      "loss": 0.5557,
+      "step": 11999
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 2.4697834696291724e-05,
+      "loss": 0.5094,
+      "step": 12000
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 2.4695263076685697e-05,
+      "loss": 0.5794,
+      "step": 12001
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 2.469269145707967e-05,
+      "loss": 0.8161,
+      "step": 12002
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 2.469011983747364e-05,
+      "loss": 0.5748,
+      "step": 12003
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 2.4687548217867616e-05,
+      "loss": 0.5903,
+      "step": 12004
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 2.4684976598261586e-05,
+      "loss": 0.4276,
+      "step": 12005
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 2.468240497865556e-05,
+      "loss": 0.6024,
+      "step": 12006
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 2.4679833359049532e-05,
+      "loss": 0.5025,
+      "step": 12007
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 2.46772617394435e-05,
+      "loss": 0.8576,
+      "step": 12008
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 2.4674690119837475e-05,
+      "loss": 0.7766,
+      "step": 12009
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 2.4672118500231448e-05,
+      "loss": 0.6105,
+      "step": 12010
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 2.4669546880625417e-05,
+      "loss": 0.4173,
+      "step": 12011
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 2.466697526101939e-05,
+      "loss": 0.6852,
+      "step": 12012
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 2.4664403641413364e-05,
+      "loss": 0.6218,
+      "step": 12013
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 2.4661832021807337e-05,
+      "loss": 0.5882,
+      "step": 12014
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 2.4659260402201306e-05,
+      "loss": 0.5733,
+      "step": 12015
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 2.465668878259528e-05,
+      "loss": 0.6207,
+      "step": 12016
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 2.4654117162989253e-05,
+      "loss": 0.8716,
+      "step": 12017
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 2.4651545543383222e-05,
+      "loss": 0.4051,
+      "step": 12018
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 2.4648973923777195e-05,
+      "loss": 0.6106,
+      "step": 12019
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 2.464640230417117e-05,
+      "loss": 0.6528,
+      "step": 12020
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 2.4643830684565138e-05,
+      "loss": 0.9579,
+      "step": 12021
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 2.4641259064959114e-05,
+      "loss": 0.5796,
+      "step": 12022
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 2.4638687445353084e-05,
+      "loss": 0.9273,
+      "step": 12023
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 2.4636115825747054e-05,
+      "loss": 0.5672,
+      "step": 12024
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 2.463354420614103e-05,
+      "loss": 0.7709,
+      "step": 12025
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 2.4630972586535e-05,
+      "loss": 0.5146,
+      "step": 12026
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 2.4628400966928973e-05,
+      "loss": 1.0163,
+      "step": 12027
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 2.4625829347322946e-05,
+      "loss": 0.5226,
+      "step": 12028
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 2.4623257727716916e-05,
+      "loss": 0.7283,
+      "step": 12029
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 2.462068610811089e-05,
+      "loss": 0.4446,
+      "step": 12030
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 2.4618114488504862e-05,
+      "loss": 0.5548,
+      "step": 12031
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 2.461554286889883e-05,
+      "loss": 0.7913,
+      "step": 12032
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 2.4612971249292805e-05,
+      "loss": 0.6545,
+      "step": 12033
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 2.4610399629686778e-05,
+      "loss": 0.5207,
+      "step": 12034
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 2.460782801008075e-05,
+      "loss": 0.8928,
+      "step": 12035
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 2.4605256390474724e-05,
+      "loss": 0.7314,
+      "step": 12036
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 2.4602684770868694e-05,
+      "loss": 0.6002,
+      "step": 12037
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 2.4600113151262667e-05,
+      "loss": 0.9928,
+      "step": 12038
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 2.459754153165664e-05,
+      "loss": 0.6731,
+      "step": 12039
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 2.459496991205061e-05,
+      "loss": 0.8102,
+      "step": 12040
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 2.4592398292444583e-05,
+      "loss": 0.6028,
+      "step": 12041
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 2.4589826672838556e-05,
+      "loss": 0.7735,
+      "step": 12042
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 2.458725505323253e-05,
+      "loss": 0.9259,
+      "step": 12043
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 2.45846834336265e-05,
+      "loss": 0.8909,
+      "step": 12044
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 2.458211181402047e-05,
+      "loss": 0.5137,
+      "step": 12045
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 2.4579540194414445e-05,
+      "loss": 0.5793,
+      "step": 12046
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 2.4576968574808414e-05,
+      "loss": 0.5874,
+      "step": 12047
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 2.4574396955202387e-05,
+      "loss": 0.6352,
+      "step": 12048
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 2.457182533559636e-05,
+      "loss": 1.0418,
+      "step": 12049
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 2.456925371599033e-05,
+      "loss": 0.6303,
+      "step": 12050
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 2.4566682096384307e-05,
+      "loss": 0.8185,
+      "step": 12051
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 2.4564110476778276e-05,
+      "loss": 1.0446,
+      "step": 12052
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 2.456153885717225e-05,
+      "loss": 0.5859,
+      "step": 12053
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 2.4558967237566222e-05,
+      "loss": 0.5833,
+      "step": 12054
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 2.4556395617960192e-05,
+      "loss": 0.5848,
+      "step": 12055
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 2.4553823998354165e-05,
+      "loss": 0.9339,
+      "step": 12056
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 2.4551252378748138e-05,
+      "loss": 0.6466,
+      "step": 12057
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 2.4548680759142108e-05,
+      "loss": 0.4853,
+      "step": 12058
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 2.454610913953608e-05,
+      "loss": 0.6769,
+      "step": 12059
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 2.4543537519930054e-05,
+      "loss": 0.6682,
+      "step": 12060
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 2.4540965900324027e-05,
+      "loss": 0.8142,
+      "step": 12061
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 2.4538394280717997e-05,
+      "loss": 0.7997,
+      "step": 12062
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 2.453582266111197e-05,
+      "loss": 0.931,
+      "step": 12063
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 2.4533251041505943e-05,
+      "loss": 0.774,
+      "step": 12064
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 2.4530679421899913e-05,
+      "loss": 0.4751,
+      "step": 12065
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 2.4528107802293886e-05,
+      "loss": 0.5074,
+      "step": 12066
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 2.452553618268786e-05,
+      "loss": 0.5717,
+      "step": 12067
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 2.452296456308183e-05,
+      "loss": 0.6053,
+      "step": 12068
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 2.4520392943475805e-05,
+      "loss": 0.5291,
+      "step": 12069
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 2.4517821323869775e-05,
+      "loss": 0.8547,
+      "step": 12070
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 2.4515249704263744e-05,
+      "loss": 1.0265,
+      "step": 12071
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 2.451267808465772e-05,
+      "loss": 0.5383,
+      "step": 12072
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 2.451010646505169e-05,
+      "loss": 0.8916,
+      "step": 12073
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 2.4507534845445663e-05,
+      "loss": 0.6818,
+      "step": 12074
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 2.4504963225839637e-05,
+      "loss": 0.7025,
+      "step": 12075
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 2.4502391606233606e-05,
+      "loss": 0.7729,
+      "step": 12076
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 2.449981998662758e-05,
+      "loss": 0.7964,
+      "step": 12077
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 2.4497248367021552e-05,
+      "loss": 0.4629,
+      "step": 12078
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 2.4494676747415522e-05,
+      "loss": 0.7363,
+      "step": 12079
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 2.4492105127809495e-05,
+      "loss": 0.5806,
+      "step": 12080
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 2.4489533508203468e-05,
+      "loss": 0.6935,
+      "step": 12081
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 2.448696188859744e-05,
+      "loss": 0.7085,
+      "step": 12082
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 2.448439026899141e-05,
+      "loss": 0.481,
+      "step": 12083
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 2.4481818649385384e-05,
+      "loss": 0.9733,
+      "step": 12084
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 2.4479247029779357e-05,
+      "loss": 0.4704,
+      "step": 12085
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 2.4476675410173327e-05,
+      "loss": 0.8017,
+      "step": 12086
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 2.44741037905673e-05,
+      "loss": 0.5924,
+      "step": 12087
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 2.4471532170961273e-05,
+      "loss": 0.731,
+      "step": 12088
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 2.4468960551355243e-05,
+      "loss": 0.9243,
+      "step": 12089
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 2.446638893174922e-05,
+      "loss": 0.6534,
+      "step": 12090
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 2.446381731214319e-05,
+      "loss": 0.7936,
+      "step": 12091
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 2.4461245692537162e-05,
+      "loss": 0.939,
+      "step": 12092
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 2.4458674072931135e-05,
+      "loss": 0.8176,
+      "step": 12093
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 2.4456102453325105e-05,
+      "loss": 0.8739,
+      "step": 12094
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 2.4453530833719078e-05,
+      "loss": 0.7788,
+      "step": 12095
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 2.445095921411305e-05,
+      "loss": 0.5929,
+      "step": 12096
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 2.444838759450702e-05,
+      "loss": 1.1315,
+      "step": 12097
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 2.4445815974900994e-05,
+      "loss": 0.5539,
+      "step": 12098
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 2.4443244355294967e-05,
+      "loss": 0.5429,
+      "step": 12099
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 2.444067273568894e-05,
+      "loss": 0.8985,
+      "step": 12100
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 2.443810111608291e-05,
+      "loss": 0.6558,
+      "step": 12101
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 2.4435529496476882e-05,
+      "loss": 0.7981,
+      "step": 12102
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 2.4432957876870856e-05,
+      "loss": 0.5919,
+      "step": 12103
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 2.4430386257264825e-05,
+      "loss": 0.7752,
+      "step": 12104
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 2.4427814637658798e-05,
+      "loss": 0.2832,
+      "step": 12105
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 2.442524301805277e-05,
+      "loss": 0.7002,
+      "step": 12106
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 2.442267139844674e-05,
+      "loss": 0.8553,
+      "step": 12107
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 2.4420099778840717e-05,
+      "loss": 0.8925,
+      "step": 12108
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 2.4417528159234687e-05,
+      "loss": 0.4042,
+      "step": 12109
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 2.4414956539628657e-05,
+      "loss": 0.9227,
+      "step": 12110
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 2.4412384920022633e-05,
+      "loss": 0.6602,
+      "step": 12111
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 2.4409813300416603e-05,
+      "loss": 0.5845,
+      "step": 12112
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 2.4407241680810576e-05,
+      "loss": 0.8684,
+      "step": 12113
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 2.440467006120455e-05,
+      "loss": 0.465,
+      "step": 12114
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 2.440209844159852e-05,
+      "loss": 1.0243,
+      "step": 12115
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 2.4399526821992492e-05,
+      "loss": 1.0767,
+      "step": 12116
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 2.4396955202386465e-05,
+      "loss": 0.5725,
+      "step": 12117
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 2.4394383582780435e-05,
+      "loss": 0.3319,
+      "step": 12118
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 2.4391811963174408e-05,
+      "loss": 0.7588,
+      "step": 12119
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 2.438924034356838e-05,
+      "loss": 0.7569,
+      "step": 12120
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 2.4386668723962354e-05,
+      "loss": 0.7315,
+      "step": 12121
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 2.4384097104356324e-05,
+      "loss": 0.6479,
+      "step": 12122
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 2.4381525484750297e-05,
+      "loss": 0.4868,
+      "step": 12123
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 2.437895386514427e-05,
+      "loss": 0.665,
+      "step": 12124
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 2.437638224553824e-05,
+      "loss": 0.6631,
+      "step": 12125
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 2.4373810625932212e-05,
+      "loss": 0.6116,
+      "step": 12126
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 2.4371239006326186e-05,
+      "loss": 0.8502,
+      "step": 12127
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 2.4368667386720155e-05,
+      "loss": 0.654,
+      "step": 12128
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 2.4366095767114132e-05,
+      "loss": 0.6108,
+      "step": 12129
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 2.43635241475081e-05,
+      "loss": 0.5858,
+      "step": 12130
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 2.4360952527902074e-05,
+      "loss": 0.8149,
+      "step": 12131
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 2.4358380908296048e-05,
+      "loss": 0.5437,
+      "step": 12132
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 2.4355809288690017e-05,
+      "loss": 0.5909,
+      "step": 12133
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 2.435323766908399e-05,
+      "loss": 0.8395,
+      "step": 12134
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 2.4350666049477963e-05,
+      "loss": 0.4866,
+      "step": 12135
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 2.4348094429871933e-05,
+      "loss": 0.9791,
+      "step": 12136
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 2.4345522810265906e-05,
+      "loss": 0.7223,
+      "step": 12137
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 2.434295119065988e-05,
+      "loss": 0.639,
+      "step": 12138
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 2.4340379571053852e-05,
+      "loss": 0.7155,
+      "step": 12139
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 2.4337807951447822e-05,
+      "loss": 0.6244,
+      "step": 12140
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 2.4335236331841795e-05,
+      "loss": 0.7494,
+      "step": 12141
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 2.4332664712235768e-05,
+      "loss": 0.3824,
+      "step": 12142
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 2.4330093092629738e-05,
+      "loss": 0.6179,
+      "step": 12143
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 2.432752147302371e-05,
+      "loss": 0.6879,
+      "step": 12144
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 2.4324949853417684e-05,
+      "loss": 0.8189,
+      "step": 12145
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 2.4322378233811654e-05,
+      "loss": 0.4861,
+      "step": 12146
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 2.431980661420563e-05,
+      "loss": 0.4684,
+      "step": 12147
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 2.43172349945996e-05,
+      "loss": 1.1212,
+      "step": 12148
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 2.431466337499357e-05,
+      "loss": 0.8128,
+      "step": 12149
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 2.4312091755387546e-05,
+      "loss": 0.6902,
+      "step": 12150
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 2.4309520135781516e-05,
+      "loss": 0.8278,
+      "step": 12151
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 2.430694851617549e-05,
+      "loss": 0.7007,
+      "step": 12152
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 2.4304376896569462e-05,
+      "loss": 0.7934,
+      "step": 12153
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 2.430180527696343e-05,
+      "loss": 0.4544,
+      "step": 12154
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 2.4299233657357405e-05,
+      "loss": 0.4558,
+      "step": 12155
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 2.4296662037751378e-05,
+      "loss": 0.7431,
+      "step": 12156
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 2.4294090418145347e-05,
+      "loss": 0.8831,
+      "step": 12157
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 2.429151879853932e-05,
+      "loss": 0.9539,
+      "step": 12158
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 2.4288947178933293e-05,
+      "loss": 0.504,
+      "step": 12159
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 2.4286375559327266e-05,
+      "loss": 0.9223,
+      "step": 12160
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 2.4283803939721236e-05,
+      "loss": 0.6623,
+      "step": 12161
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 2.428123232011521e-05,
+      "loss": 0.6506,
+      "step": 12162
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 2.4278660700509182e-05,
+      "loss": 0.5147,
+      "step": 12163
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 2.4276089080903152e-05,
+      "loss": 0.8011,
+      "step": 12164
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 2.4273517461297125e-05,
+      "loss": 0.7793,
+      "step": 12165
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 2.4270945841691098e-05,
+      "loss": 0.712,
+      "step": 12166
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 2.4268374222085068e-05,
+      "loss": 0.5123,
+      "step": 12167
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 2.4265802602479044e-05,
+      "loss": 0.753,
+      "step": 12168
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 2.4263230982873014e-05,
+      "loss": 0.9552,
+      "step": 12169
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 2.4260659363266987e-05,
+      "loss": 0.6156,
+      "step": 12170
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 2.425808774366096e-05,
+      "loss": 1.0014,
+      "step": 12171
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 2.425551612405493e-05,
+      "loss": 0.692,
+      "step": 12172
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 2.4252944504448903e-05,
+      "loss": 0.5945,
+      "step": 12173
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 2.4250372884842876e-05,
+      "loss": 0.6811,
+      "step": 12174
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 2.4247801265236846e-05,
+      "loss": 0.7607,
+      "step": 12175
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 2.424522964563082e-05,
+      "loss": 0.6002,
+      "step": 12176
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 2.4242658026024792e-05,
+      "loss": 0.5451,
+      "step": 12177
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 2.4240086406418765e-05,
+      "loss": 0.6446,
+      "step": 12178
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 2.4237514786812735e-05,
+      "loss": 0.7854,
+      "step": 12179
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 2.4234943167206708e-05,
+      "loss": 0.7417,
+      "step": 12180
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 2.423237154760068e-05,
+      "loss": 0.5647,
+      "step": 12181
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 2.422979992799465e-05,
+      "loss": 0.4322,
+      "step": 12182
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 2.4227228308388623e-05,
+      "loss": 0.5275,
+      "step": 12183
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 2.4224656688782597e-05,
+      "loss": 0.9262,
+      "step": 12184
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 2.422208506917657e-05,
+      "loss": 0.744,
+      "step": 12185
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 2.4219513449570543e-05,
+      "loss": 0.9209,
+      "step": 12186
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 2.4216941829964512e-05,
+      "loss": 0.4807,
+      "step": 12187
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 2.4214370210358485e-05,
+      "loss": 0.5908,
+      "step": 12188
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 2.421179859075246e-05,
+      "loss": 0.6785,
+      "step": 12189
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 2.4209226971146428e-05,
+      "loss": 1.0117,
+      "step": 12190
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 2.42066553515404e-05,
+      "loss": 1.0355,
+      "step": 12191
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 2.4204083731934374e-05,
+      "loss": 0.7622,
+      "step": 12192
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 2.4201512112328344e-05,
+      "loss": 0.6245,
+      "step": 12193
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 2.419894049272232e-05,
+      "loss": 0.7116,
+      "step": 12194
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 2.419636887311629e-05,
+      "loss": 0.9361,
+      "step": 12195
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 2.419379725351026e-05,
+      "loss": 1.0666,
+      "step": 12196
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 2.4191225633904236e-05,
+      "loss": 0.2359,
+      "step": 12197
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 2.4188654014298206e-05,
+      "loss": 0.743,
+      "step": 12198
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 2.418608239469218e-05,
+      "loss": 0.7205,
+      "step": 12199
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 2.4183510775086152e-05,
+      "loss": 0.4389,
+      "step": 12200
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 2.4180939155480122e-05,
+      "loss": 0.9633,
+      "step": 12201
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 2.4178367535874095e-05,
+      "loss": 0.9552,
+      "step": 12202
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 2.4175795916268068e-05,
+      "loss": 0.6925,
+      "step": 12203
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 2.4173224296662038e-05,
+      "loss": 1.0037,
+      "step": 12204
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 2.417065267705601e-05,
+      "loss": 0.8779,
+      "step": 12205
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 2.4168081057449984e-05,
+      "loss": 0.7524,
+      "step": 12206
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 2.4165509437843957e-05,
+      "loss": 0.5472,
+      "step": 12207
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 2.4162937818237927e-05,
+      "loss": 0.6536,
+      "step": 12208
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 2.41603661986319e-05,
+      "loss": 0.7887,
+      "step": 12209
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 2.4157794579025873e-05,
+      "loss": 0.7835,
+      "step": 12210
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 2.4155222959419842e-05,
+      "loss": 0.6231,
+      "step": 12211
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 2.4152651339813815e-05,
+      "loss": 0.7021,
+      "step": 12212
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 2.415007972020779e-05,
+      "loss": 0.8554,
+      "step": 12213
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 2.4147508100601758e-05,
+      "loss": 1.0208,
+      "step": 12214
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 2.4144936480995735e-05,
+      "loss": 0.6807,
+      "step": 12215
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 2.4142364861389704e-05,
+      "loss": 0.5521,
+      "step": 12216
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 2.4139793241783677e-05,
+      "loss": 1.0385,
+      "step": 12217
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 2.413722162217765e-05,
+      "loss": 0.8964,
+      "step": 12218
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 2.413465000257162e-05,
+      "loss": 0.6392,
+      "step": 12219
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 2.4132078382965593e-05,
+      "loss": 0.5504,
+      "step": 12220
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 2.4129506763359566e-05,
+      "loss": 0.7189,
+      "step": 12221
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 2.4126935143753536e-05,
+      "loss": 0.6888,
+      "step": 12222
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 2.412436352414751e-05,
+      "loss": 0.7194,
+      "step": 12223
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 2.4121791904541482e-05,
+      "loss": 0.661,
+      "step": 12224
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 2.4119220284935455e-05,
+      "loss": 0.4619,
+      "step": 12225
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 2.4116648665329425e-05,
+      "loss": 0.769,
+      "step": 12226
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 2.4114077045723398e-05,
+      "loss": 0.7018,
+      "step": 12227
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 2.411150542611737e-05,
+      "loss": 0.657,
+      "step": 12228
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 2.410893380651134e-05,
+      "loss": 0.6555,
+      "step": 12229
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 2.4106362186905314e-05,
+      "loss": 0.7682,
+      "step": 12230
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 2.4103790567299287e-05,
+      "loss": 0.5207,
+      "step": 12231
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 2.4101218947693257e-05,
+      "loss": 0.6348,
+      "step": 12232
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 2.4098647328087233e-05,
+      "loss": 0.7775,
+      "step": 12233
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 2.4096075708481203e-05,
+      "loss": 0.8188,
+      "step": 12234
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.4093504088875172e-05,
+      "loss": 0.7953,
+      "step": 12235
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.409093246926915e-05,
+      "loss": 0.6872,
+      "step": 12236
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.408836084966312e-05,
+      "loss": 0.6763,
+      "step": 12237
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.408578923005709e-05,
+      "loss": 0.7456,
+      "step": 12238
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.4083217610451065e-05,
+      "loss": 0.5221,
+      "step": 12239
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.4080645990845034e-05,
+      "loss": 0.9297,
+      "step": 12240
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.4078074371239008e-05,
+      "loss": 0.8457,
+      "step": 12241
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.407550275163298e-05,
+      "loss": 0.8273,
+      "step": 12242
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.407293113202695e-05,
+      "loss": 0.6434,
+      "step": 12243
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.4070359512420923e-05,
+      "loss": 0.4434,
+      "step": 12244
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.4067787892814896e-05,
+      "loss": 0.6321,
+      "step": 12245
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.406521627320887e-05,
+      "loss": 0.9947,
+      "step": 12246
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.406264465360284e-05,
+      "loss": 0.3336,
+      "step": 12247
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.4060073033996812e-05,
+      "loss": 0.6636,
+      "step": 12248
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.4057501414390785e-05,
+      "loss": 0.6106,
+      "step": 12249
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.4054929794784755e-05,
+      "loss": 0.7201,
+      "step": 12250
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.4052358175178728e-05,
+      "loss": 0.7612,
+      "step": 12251
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.40497865555727e-05,
+      "loss": 0.7683,
+      "step": 12252
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.404721493596667e-05,
+      "loss": 0.5297,
+      "step": 12253
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.4044643316360647e-05,
+      "loss": 0.4593,
+      "step": 12254
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.4042071696754617e-05,
+      "loss": 0.7156,
+      "step": 12255
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.403950007714859e-05,
+      "loss": 0.5516,
+      "step": 12256
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.4036928457542563e-05,
+      "loss": 0.5842,
+      "step": 12257
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.4034356837936533e-05,
+      "loss": 0.7525,
+      "step": 12258
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.4031785218330506e-05,
+      "loss": 0.6753,
+      "step": 12259
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.402921359872448e-05,
+      "loss": 0.7851,
+      "step": 12260
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.402664197911845e-05,
+      "loss": 0.8497,
+      "step": 12261
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.4024070359512422e-05,
+      "loss": 1.1083,
+      "step": 12262
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.4021498739906395e-05,
+      "loss": 0.8197,
+      "step": 12263
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.4018927120300368e-05,
+      "loss": 0.6069,
+      "step": 12264
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.4016355500694338e-05,
+      "loss": 0.9272,
+      "step": 12265
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.401378388108831e-05,
+      "loss": 0.8232,
+      "step": 12266
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.4011212261482284e-05,
+      "loss": 0.6596,
+      "step": 12267
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.4008640641876253e-05,
+      "loss": 0.4441,
+      "step": 12268
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.4006069022270226e-05,
+      "loss": 0.9693,
+      "step": 12269
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.40034974026642e-05,
+      "loss": 0.448,
+      "step": 12270
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.400092578305817e-05,
+      "loss": 1.0783,
+      "step": 12271
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.3998354163452146e-05,
+      "loss": 0.648,
+      "step": 12272
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.3995782543846115e-05,
+      "loss": 0.5487,
+      "step": 12273
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.3993210924240085e-05,
+      "loss": 0.5655,
+      "step": 12274
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.399063930463406e-05,
+      "loss": 0.8319,
+      "step": 12275
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.398806768502803e-05,
+      "loss": 0.7405,
+      "step": 12276
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.3985496065422004e-05,
+      "loss": 0.8871,
+      "step": 12277
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.3982924445815977e-05,
+      "loss": 0.7844,
+      "step": 12278
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.3980352826209947e-05,
+      "loss": 0.8839,
+      "step": 12279
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.397778120660392e-05,
+      "loss": 0.6452,
+      "step": 12280
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.3975209586997893e-05,
+      "loss": 0.5691,
+      "step": 12281
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.3972637967391863e-05,
+      "loss": 0.551,
+      "step": 12282
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.3970066347785836e-05,
+      "loss": 0.7357,
+      "step": 12283
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.396749472817981e-05,
+      "loss": 0.7267,
+      "step": 12284
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.3964923108573782e-05,
+      "loss": 0.8062,
+      "step": 12285
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.3962351488967752e-05,
+      "loss": 0.5661,
+      "step": 12286
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.3959779869361725e-05,
+      "loss": 0.9666,
+      "step": 12287
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.3957208249755698e-05,
+      "loss": 0.4364,
+      "step": 12288
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.3954636630149668e-05,
+      "loss": 0.5657,
+      "step": 12289
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.395206501054364e-05,
+      "loss": 0.6634,
+      "step": 12290
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.3949493390937614e-05,
+      "loss": 0.9046,
+      "step": 12291
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.3946921771331583e-05,
+      "loss": 0.5073,
+      "step": 12292
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.394435015172556e-05,
+      "loss": 0.5364,
+      "step": 12293
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.394177853211953e-05,
+      "loss": 0.5908,
+      "step": 12294
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.3939206912513503e-05,
+      "loss": 1.0337,
+      "step": 12295
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.3936635292907476e-05,
+      "loss": 0.7145,
+      "step": 12296
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.3934063673301445e-05,
+      "loss": 0.9456,
+      "step": 12297
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.393149205369542e-05,
+      "loss": 0.6098,
+      "step": 12298
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.392892043408939e-05,
+      "loss": 0.5811,
+      "step": 12299
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.392634881448336e-05,
+      "loss": 0.6544,
+      "step": 12300
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.3923777194877334e-05,
+      "loss": 0.9779,
+      "step": 12301
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.3921205575271307e-05,
+      "loss": 0.7688,
+      "step": 12302
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.391863395566528e-05,
+      "loss": 0.5298,
+      "step": 12303
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.391606233605925e-05,
+      "loss": 0.9013,
+      "step": 12304
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.3913490716453223e-05,
+      "loss": 0.7642,
+      "step": 12305
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.3910919096847196e-05,
+      "loss": 0.7643,
+      "step": 12306
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.3908347477241166e-05,
+      "loss": 0.7362,
+      "step": 12307
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.390577585763514e-05,
+      "loss": 0.7325,
+      "step": 12308
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.3903204238029112e-05,
+      "loss": 0.6902,
+      "step": 12309
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.3900632618423082e-05,
+      "loss": 0.5836,
+      "step": 12310
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.3898060998817058e-05,
+      "loss": 0.4731,
+      "step": 12311
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.3895489379211028e-05,
+      "loss": 0.7075,
+      "step": 12312
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.3892917759604998e-05,
+      "loss": 0.5015,
+      "step": 12313
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.3890346139998974e-05,
+      "loss": 0.9571,
+      "step": 12314
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.3887774520392944e-05,
+      "loss": 0.9062,
+      "step": 12315
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.3885202900786917e-05,
+      "loss": 0.6445,
+      "step": 12316
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.388263128118089e-05,
+      "loss": 0.7012,
+      "step": 12317
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.388005966157486e-05,
+      "loss": 0.5897,
+      "step": 12318
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.3877488041968833e-05,
+      "loss": 0.5728,
+      "step": 12319
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.3874916422362806e-05,
+      "loss": 1.1312,
+      "step": 12320
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.3872344802756775e-05,
+      "loss": 0.8382,
+      "step": 12321
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.386977318315075e-05,
+      "loss": 1.0907,
+      "step": 12322
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.386720156354472e-05,
+      "loss": 0.6766,
+      "step": 12323
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.3864629943938695e-05,
+      "loss": 0.5337,
+      "step": 12324
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.3862058324332664e-05,
+      "loss": 0.6917,
+      "step": 12325
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.3859486704726637e-05,
+      "loss": 0.5924,
+      "step": 12326
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.385691508512061e-05,
+      "loss": 0.5729,
+      "step": 12327
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.385434346551458e-05,
+      "loss": 0.7761,
+      "step": 12328
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.3851771845908553e-05,
+      "loss": 0.9456,
+      "step": 12329
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.3849200226302526e-05,
+      "loss": 0.468,
+      "step": 12330
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.3846628606696496e-05,
+      "loss": 0.5283,
+      "step": 12331
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.3844056987090472e-05,
+      "loss": 0.5405,
+      "step": 12332
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.3841485367484442e-05,
+      "loss": 0.5733,
+      "step": 12333
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.3838913747878415e-05,
+      "loss": 0.8888,
+      "step": 12334
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.383634212827239e-05,
+      "loss": 0.5555,
+      "step": 12335
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.3833770508666358e-05,
+      "loss": 0.6465,
+      "step": 12336
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.383119888906033e-05,
+      "loss": 0.5725,
+      "step": 12337
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.3828627269454304e-05,
+      "loss": 0.6759,
+      "step": 12338
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.3826055649848274e-05,
+      "loss": 0.471,
+      "step": 12339
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.382348403024225e-05,
+      "loss": 0.4775,
+      "step": 12340
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.382091241063622e-05,
+      "loss": 0.6643,
+      "step": 12341
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.3818340791030193e-05,
+      "loss": 0.9049,
+      "step": 12342
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.3815769171424166e-05,
+      "loss": 0.5974,
+      "step": 12343
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.3813197551818136e-05,
+      "loss": 1.0109,
+      "step": 12344
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.381062593221211e-05,
+      "loss": 0.8887,
+      "step": 12345
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.3808054312606082e-05,
+      "loss": 0.759,
+      "step": 12346
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.380548269300005e-05,
+      "loss": 0.7319,
+      "step": 12347
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.3802911073394025e-05,
+      "loss": 0.7136,
+      "step": 12348
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.3800339453787998e-05,
+      "loss": 0.3417,
+      "step": 12349
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.379776783418197e-05,
+      "loss": 0.716,
+      "step": 12350
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.379519621457594e-05,
+      "loss": 0.8625,
+      "step": 12351
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.3792624594969914e-05,
+      "loss": 0.6513,
+      "step": 12352
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.3790052975363887e-05,
+      "loss": 0.707,
+      "step": 12353
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.3787481355757856e-05,
+      "loss": 0.877,
+      "step": 12354
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.378490973615183e-05,
+      "loss": 0.6513,
+      "step": 12355
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.3782338116545803e-05,
+      "loss": 0.5851,
+      "step": 12356
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.3779766496939772e-05,
+      "loss": 0.9021,
+      "step": 12357
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.377719487733375e-05,
+      "loss": 0.8062,
+      "step": 12358
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.377462325772772e-05,
+      "loss": 0.4422,
+      "step": 12359
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.3772051638121688e-05,
+      "loss": 0.4264,
+      "step": 12360
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.3769480018515665e-05,
+      "loss": 1.1396,
+      "step": 12361
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.3766908398909634e-05,
+      "loss": 0.7945,
+      "step": 12362
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.3764336779303607e-05,
+      "loss": 0.5361,
+      "step": 12363
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.376176515969758e-05,
+      "loss": 0.622,
+      "step": 12364
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.375919354009155e-05,
+      "loss": 0.7649,
+      "step": 12365
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.3756621920485523e-05,
+      "loss": 0.7686,
+      "step": 12366
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.3754050300879496e-05,
+      "loss": 0.8529,
+      "step": 12367
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.3751478681273466e-05,
+      "loss": 0.8909,
+      "step": 12368
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.374890706166744e-05,
+      "loss": 0.7591,
+      "step": 12369
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.3746335442061412e-05,
+      "loss": 0.4117,
+      "step": 12370
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.3743763822455385e-05,
+      "loss": 0.5841,
+      "step": 12371
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.3741192202849355e-05,
+      "loss": 0.6278,
+      "step": 12372
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.3738620583243328e-05,
+      "loss": 0.9503,
+      "step": 12373
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.37360489636373e-05,
+      "loss": 0.7765,
+      "step": 12374
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.373347734403127e-05,
+      "loss": 0.7147,
+      "step": 12375
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.3730905724425244e-05,
+      "loss": 0.7665,
+      "step": 12376
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.3728334104819217e-05,
+      "loss": 0.561,
+      "step": 12377
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.3725762485213186e-05,
+      "loss": 0.8241,
+      "step": 12378
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.3723190865607163e-05,
+      "loss": 0.9296,
+      "step": 12379
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.3720619246001133e-05,
+      "loss": 0.5297,
+      "step": 12380
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.3718047626395106e-05,
+      "loss": 0.585,
+      "step": 12381
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.371547600678908e-05,
+      "loss": 0.7251,
+      "step": 12382
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.371290438718305e-05,
+      "loss": 0.8371,
+      "step": 12383
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.371033276757702e-05,
+      "loss": 0.5708,
+      "step": 12384
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.3707761147970995e-05,
+      "loss": 0.8352,
+      "step": 12385
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.3705189528364964e-05,
+      "loss": 0.6271,
+      "step": 12386
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.3702617908758937e-05,
+      "loss": 0.6901,
+      "step": 12387
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.370004628915291e-05,
+      "loss": 0.6396,
+      "step": 12388
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.3697474669546883e-05,
+      "loss": 0.5491,
+      "step": 12389
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.3694903049940853e-05,
+      "loss": 0.9481,
+      "step": 12390
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.3692331430334826e-05,
+      "loss": 0.7886,
+      "step": 12391
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.36897598107288e-05,
+      "loss": 0.5482,
+      "step": 12392
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.368718819112277e-05,
+      "loss": 0.8569,
+      "step": 12393
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.3684616571516742e-05,
+      "loss": 0.5137,
+      "step": 12394
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.3682044951910715e-05,
+      "loss": 0.8787,
+      "step": 12395
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.3679473332304685e-05,
+      "loss": 0.7022,
+      "step": 12396
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.367690171269866e-05,
+      "loss": 1.04,
+      "step": 12397
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.367433009309263e-05,
+      "loss": 0.9889,
+      "step": 12398
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.36717584734866e-05,
+      "loss": 0.9715,
+      "step": 12399
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.3669186853880577e-05,
+      "loss": 0.9243,
+      "step": 12400
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.3666615234274547e-05,
+      "loss": 0.7694,
+      "step": 12401
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.366404361466852e-05,
+      "loss": 0.749,
+      "step": 12402
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.3661471995062493e-05,
+      "loss": 0.45,
+      "step": 12403
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.3658900375456463e-05,
+      "loss": 0.5824,
+      "step": 12404
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.3656328755850436e-05,
+      "loss": 0.6949,
+      "step": 12405
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.365375713624441e-05,
+      "loss": 0.5703,
+      "step": 12406
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.365118551663838e-05,
+      "loss": 0.6449,
+      "step": 12407
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.364861389703235e-05,
+      "loss": 0.6068,
+      "step": 12408
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.3646042277426325e-05,
+      "loss": 0.8131,
+      "step": 12409
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.3643470657820298e-05,
+      "loss": 1.1907,
+      "step": 12410
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.3640899038214267e-05,
+      "loss": 0.8333,
+      "step": 12411
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.363832741860824e-05,
+      "loss": 0.8221,
+      "step": 12412
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.3635755799002214e-05,
+      "loss": 0.3726,
+      "step": 12413
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.3633184179396183e-05,
+      "loss": 0.5416,
+      "step": 12414
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.3630612559790156e-05,
+      "loss": 0.6214,
+      "step": 12415
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.362804094018413e-05,
+      "loss": 0.5117,
+      "step": 12416
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.36254693205781e-05,
+      "loss": 0.778,
+      "step": 12417
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.3622897700972075e-05,
+      "loss": 0.8579,
+      "step": 12418
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.3620326081366045e-05,
+      "loss": 0.7659,
+      "step": 12419
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.3617754461760018e-05,
+      "loss": 0.97,
+      "step": 12420
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.361518284215399e-05,
+      "loss": 0.7099,
+      "step": 12421
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.361261122254796e-05,
+      "loss": 0.4731,
+      "step": 12422
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.3610039602941934e-05,
+      "loss": 0.6298,
+      "step": 12423
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.3607467983335907e-05,
+      "loss": 0.7501,
+      "step": 12424
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.3604896363729877e-05,
+      "loss": 0.8223,
+      "step": 12425
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.360232474412385e-05,
+      "loss": 0.9413,
+      "step": 12426
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.3599753124517823e-05,
+      "loss": 0.3796,
+      "step": 12427
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.3597181504911796e-05,
+      "loss": 0.4213,
+      "step": 12428
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.3594609885305766e-05,
+      "loss": 0.9722,
+      "step": 12429
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.359203826569974e-05,
+      "loss": 0.5161,
+      "step": 12430
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.3589466646093712e-05,
+      "loss": 0.6957,
+      "step": 12431
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.358689502648768e-05,
+      "loss": 0.6216,
+      "step": 12432
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.3584323406881655e-05,
+      "loss": 0.6265,
+      "step": 12433
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.3581751787275628e-05,
+      "loss": 0.9398,
+      "step": 12434
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.3579180167669597e-05,
+      "loss": 0.8942,
+      "step": 12435
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.3576608548063574e-05,
+      "loss": 0.9702,
+      "step": 12436
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.3574036928457544e-05,
+      "loss": 0.5943,
+      "step": 12437
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.3571465308851513e-05,
+      "loss": 0.6858,
+      "step": 12438
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.356889368924549e-05,
+      "loss": 0.7298,
+      "step": 12439
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.356632206963946e-05,
+      "loss": 0.7855,
+      "step": 12440
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.3563750450033432e-05,
+      "loss": 0.7704,
+      "step": 12441
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.3561178830427406e-05,
+      "loss": 0.7961,
+      "step": 12442
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.3558607210821375e-05,
+      "loss": 0.4908,
+      "step": 12443
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.3556035591215348e-05,
+      "loss": 0.519,
+      "step": 12444
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.355346397160932e-05,
+      "loss": 0.8052,
+      "step": 12445
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.355089235200329e-05,
+      "loss": 0.7022,
+      "step": 12446
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.3548320732397264e-05,
+      "loss": 0.6644,
+      "step": 12447
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.3545749112791237e-05,
+      "loss": 0.7752,
+      "step": 12448
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.354317749318521e-05,
+      "loss": 0.5872,
+      "step": 12449
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.354060587357918e-05,
+      "loss": 0.5459,
+      "step": 12450
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.3538034253973153e-05,
+      "loss": 0.3436,
+      "step": 12451
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.3535462634367126e-05,
+      "loss": 0.5805,
+      "step": 12452
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.3532891014761096e-05,
+      "loss": 0.3568,
+      "step": 12453
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.353031939515507e-05,
+      "loss": 0.7094,
+      "step": 12454
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.3527747775549042e-05,
+      "loss": 0.8333,
+      "step": 12455
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.352517615594301e-05,
+      "loss": 0.6582,
+      "step": 12456
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.3522604536336988e-05,
+      "loss": 0.437,
+      "step": 12457
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.3520032916730958e-05,
+      "loss": 0.6961,
+      "step": 12458
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.351746129712493e-05,
+      "loss": 0.607,
+      "step": 12459
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.3514889677518904e-05,
+      "loss": 0.692,
+      "step": 12460
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.3512318057912874e-05,
+      "loss": 0.6969,
+      "step": 12461
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.3509746438306847e-05,
+      "loss": 0.8986,
+      "step": 12462
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.350717481870082e-05,
+      "loss": 0.6928,
+      "step": 12463
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.350460319909479e-05,
+      "loss": 0.5415,
+      "step": 12464
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.3502031579488762e-05,
+      "loss": 0.5733,
+      "step": 12465
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.3499459959882736e-05,
+      "loss": 0.7233,
+      "step": 12466
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.349688834027671e-05,
+      "loss": 0.6964,
+      "step": 12467
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.349431672067068e-05,
+      "loss": 0.3793,
+      "step": 12468
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.349174510106465e-05,
+      "loss": 0.505,
+      "step": 12469
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.3489173481458624e-05,
+      "loss": 0.4535,
+      "step": 12470
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.3486601861852594e-05,
+      "loss": 0.4525,
+      "step": 12471
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.3484030242246567e-05,
+      "loss": 0.6429,
+      "step": 12472
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.348145862264054e-05,
+      "loss": 0.7832,
+      "step": 12473
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.347888700303451e-05,
+      "loss": 0.5582,
+      "step": 12474
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.3476315383428486e-05,
+      "loss": 0.7022,
+      "step": 12475
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.3473743763822456e-05,
+      "loss": 0.6364,
+      "step": 12476
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.3471172144216426e-05,
+      "loss": 0.7789,
+      "step": 12477
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.3468600524610402e-05,
+      "loss": 0.6863,
+      "step": 12478
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.3466028905004372e-05,
+      "loss": 0.9133,
+      "step": 12479
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.3463457285398345e-05,
+      "loss": 0.767,
+      "step": 12480
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.3460885665792318e-05,
+      "loss": 0.6166,
+      "step": 12481
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.3458314046186288e-05,
+      "loss": 0.7518,
+      "step": 12482
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.345574242658026e-05,
+      "loss": 0.4447,
+      "step": 12483
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.3453170806974234e-05,
+      "loss": 0.8253,
+      "step": 12484
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.3450599187368204e-05,
+      "loss": 0.6427,
+      "step": 12485
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.344802756776218e-05,
+      "loss": 0.7666,
+      "step": 12486
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.344545594815615e-05,
+      "loss": 0.5346,
+      "step": 12487
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.3442884328550123e-05,
+      "loss": 0.7445,
+      "step": 12488
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.3440312708944096e-05,
+      "loss": 0.6784,
+      "step": 12489
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.3437741089338066e-05,
+      "loss": 0.6016,
+      "step": 12490
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.343516946973204e-05,
+      "loss": 0.7533,
+      "step": 12491
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.3432597850126012e-05,
+      "loss": 0.5537,
+      "step": 12492
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.343002623051998e-05,
+      "loss": 0.6121,
+      "step": 12493
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.3427454610913955e-05,
+      "loss": 0.9665,
+      "step": 12494
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.3424882991307928e-05,
+      "loss": 0.4866,
+      "step": 12495
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.34223113717019e-05,
+      "loss": 0.6257,
+      "step": 12496
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.341973975209587e-05,
+      "loss": 0.5489,
+      "step": 12497
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.3417168132489843e-05,
+      "loss": 0.3807,
+      "step": 12498
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.3414596512883817e-05,
+      "loss": 1.0639,
+      "step": 12499
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.3412024893277786e-05,
+      "loss": 0.8044,
+      "step": 12500
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.340945327367176e-05,
+      "loss": 0.6959,
+      "step": 12501
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.3406881654065732e-05,
+      "loss": 0.6117,
+      "step": 12502
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.3404310034459702e-05,
+      "loss": 0.6871,
+      "step": 12503
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.340173841485368e-05,
+      "loss": 0.6467,
+      "step": 12504
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.3399166795247648e-05,
+      "loss": 1.0194,
+      "step": 12505
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.339659517564162e-05,
+      "loss": 0.4569,
+      "step": 12506
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.3394023556035594e-05,
+      "loss": 0.6724,
+      "step": 12507
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.3391451936429564e-05,
+      "loss": 0.8482,
+      "step": 12508
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.3388880316823537e-05,
+      "loss": 0.8256,
+      "step": 12509
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.338630869721751e-05,
+      "loss": 0.5012,
+      "step": 12510
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.338373707761148e-05,
+      "loss": 0.6249,
+      "step": 12511
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.3381165458005453e-05,
+      "loss": 0.4022,
+      "step": 12512
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.3378593838399426e-05,
+      "loss": 0.4298,
+      "step": 12513
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.33760222187934e-05,
+      "loss": 0.5664,
+      "step": 12514
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.337345059918737e-05,
+      "loss": 0.612,
+      "step": 12515
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.3370878979581342e-05,
+      "loss": 0.763,
+      "step": 12516
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.3368307359975315e-05,
+      "loss": 0.777,
+      "step": 12517
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.3365735740369285e-05,
+      "loss": 0.5512,
+      "step": 12518
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.3363164120763258e-05,
+      "loss": 0.3546,
+      "step": 12519
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.336059250115723e-05,
+      "loss": 0.8443,
+      "step": 12520
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.33580208815512e-05,
+      "loss": 0.4839,
+      "step": 12521
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.3355449261945177e-05,
+      "loss": 0.686,
+      "step": 12522
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.3352877642339147e-05,
+      "loss": 0.68,
+      "step": 12523
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.3350306022733116e-05,
+      "loss": 0.8591,
+      "step": 12524
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.3347734403127093e-05,
+      "loss": 0.65,
+      "step": 12525
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.3345162783521062e-05,
+      "loss": 0.6962,
+      "step": 12526
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.3342591163915035e-05,
+      "loss": 0.546,
+      "step": 12527
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.334001954430901e-05,
+      "loss": 0.6296,
+      "step": 12528
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.3337447924702978e-05,
+      "loss": 0.9836,
+      "step": 12529
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.333487630509695e-05,
+      "loss": 0.8181,
+      "step": 12530
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.3332304685490924e-05,
+      "loss": 0.6892,
+      "step": 12531
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.3329733065884894e-05,
+      "loss": 0.3899,
+      "step": 12532
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.3327161446278867e-05,
+      "loss": 0.8489,
+      "step": 12533
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.332458982667284e-05,
+      "loss": 0.4739,
+      "step": 12534
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.3322018207066813e-05,
+      "loss": 0.6108,
+      "step": 12535
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.3319446587460783e-05,
+      "loss": 0.7309,
+      "step": 12536
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.3316874967854756e-05,
+      "loss": 0.4576,
+      "step": 12537
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.331430334824873e-05,
+      "loss": 0.8141,
+      "step": 12538
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.33117317286427e-05,
+      "loss": 0.6565,
+      "step": 12539
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.3309160109036672e-05,
+      "loss": 0.5025,
+      "step": 12540
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.3306588489430645e-05,
+      "loss": 0.9392,
+      "step": 12541
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.3304016869824615e-05,
+      "loss": 0.6324,
+      "step": 12542
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.330144525021859e-05,
+      "loss": 0.8762,
+      "step": 12543
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.329887363061256e-05,
+      "loss": 0.6995,
+      "step": 12544
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.3296302011006534e-05,
+      "loss": 0.7437,
+      "step": 12545
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.3293730391400507e-05,
+      "loss": 1.0583,
+      "step": 12546
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.3291158771794477e-05,
+      "loss": 0.9464,
+      "step": 12547
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.328858715218845e-05,
+      "loss": 0.5422,
+      "step": 12548
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.3286015532582423e-05,
+      "loss": 0.6026,
+      "step": 12549
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.3283443912976392e-05,
+      "loss": 0.7085,
+      "step": 12550
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.3280872293370365e-05,
+      "loss": 0.7362,
+      "step": 12551
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.327830067376434e-05,
+      "loss": 0.5397,
+      "step": 12552
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.327572905415831e-05,
+      "loss": 0.6062,
+      "step": 12553
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.327315743455228e-05,
+      "loss": 0.5744,
+      "step": 12554
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.3270585814946254e-05,
+      "loss": 1.1112,
+      "step": 12555
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.3268014195340227e-05,
+      "loss": 0.7434,
+      "step": 12556
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.3265442575734197e-05,
+      "loss": 1.4379,
+      "step": 12557
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.326287095612817e-05,
+      "loss": 0.6388,
+      "step": 12558
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.3260299336522143e-05,
+      "loss": 0.6184,
+      "step": 12559
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.3257727716916113e-05,
+      "loss": 0.8794,
+      "step": 12560
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.325515609731009e-05,
+      "loss": 0.5818,
+      "step": 12561
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.325258447770406e-05,
+      "loss": 0.6628,
+      "step": 12562
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.325001285809803e-05,
+      "loss": 0.7109,
+      "step": 12563
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.3247441238492005e-05,
+      "loss": 0.5292,
+      "step": 12564
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.3244869618885975e-05,
+      "loss": 0.5762,
+      "step": 12565
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.3242297999279948e-05,
+      "loss": 0.3571,
+      "step": 12566
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.323972637967392e-05,
+      "loss": 0.7691,
+      "step": 12567
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.323715476006789e-05,
+      "loss": 0.5536,
+      "step": 12568
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.3234583140461864e-05,
+      "loss": 0.6169,
+      "step": 12569
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.3232011520855837e-05,
+      "loss": 0.7483,
+      "step": 12570
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.3229439901249807e-05,
+      "loss": 0.5595,
+      "step": 12571
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.322686828164378e-05,
+      "loss": 1.0421,
+      "step": 12572
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.3224296662037753e-05,
+      "loss": 0.7169,
+      "step": 12573
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.3221725042431726e-05,
+      "loss": 0.8574,
+      "step": 12574
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.3219153422825696e-05,
+      "loss": 0.6242,
+      "step": 12575
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.321658180321967e-05,
+      "loss": 0.5499,
+      "step": 12576
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.321401018361364e-05,
+      "loss": 0.74,
+      "step": 12577
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.321143856400761e-05,
+      "loss": 0.6924,
+      "step": 12578
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.3208866944401584e-05,
+      "loss": 0.8276,
+      "step": 12579
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.3206295324795558e-05,
+      "loss": 0.698,
+      "step": 12580
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.3203723705189527e-05,
+      "loss": 0.8296,
+      "step": 12581
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.3201152085583504e-05,
+      "loss": 0.6308,
+      "step": 12582
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.3198580465977473e-05,
+      "loss": 1.0988,
+      "step": 12583
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.3196008846371446e-05,
+      "loss": 0.9837,
+      "step": 12584
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.319343722676542e-05,
+      "loss": 0.7073,
+      "step": 12585
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.319086560715939e-05,
+      "loss": 0.4828,
+      "step": 12586
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.3188293987553362e-05,
+      "loss": 0.9596,
+      "step": 12587
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.3185722367947335e-05,
+      "loss": 0.7577,
+      "step": 12588
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.3183150748341305e-05,
+      "loss": 0.8767,
+      "step": 12589
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.3180579128735278e-05,
+      "loss": 0.7827,
+      "step": 12590
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.317800750912925e-05,
+      "loss": 0.6228,
+      "step": 12591
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.3175435889523224e-05,
+      "loss": 0.5352,
+      "step": 12592
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.3172864269917194e-05,
+      "loss": 0.9868,
+      "step": 12593
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.3170292650311167e-05,
+      "loss": 0.7296,
+      "step": 12594
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.316772103070514e-05,
+      "loss": 0.8603,
+      "step": 12595
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.316514941109911e-05,
+      "loss": 0.6537,
+      "step": 12596
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.3162577791493083e-05,
+      "loss": 0.422,
+      "step": 12597
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.3160006171887056e-05,
+      "loss": 0.3369,
+      "step": 12598
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.3157434552281026e-05,
+      "loss": 0.4202,
+      "step": 12599
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.3154862932675002e-05,
+      "loss": 0.6682,
+      "step": 12600
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.3152291313068972e-05,
+      "loss": 0.7997,
+      "step": 12601
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.314971969346294e-05,
+      "loss": 0.5943,
+      "step": 12602
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.3147148073856918e-05,
+      "loss": 0.759,
+      "step": 12603
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.3144576454250888e-05,
+      "loss": 0.7562,
+      "step": 12604
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.314200483464486e-05,
+      "loss": 0.8526,
+      "step": 12605
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.3139433215038834e-05,
+      "loss": 0.7725,
+      "step": 12606
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.3136861595432803e-05,
+      "loss": 0.584,
+      "step": 12607
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.3134289975826776e-05,
+      "loss": 0.7723,
+      "step": 12608
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.313171835622075e-05,
+      "loss": 0.7489,
+      "step": 12609
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.312914673661472e-05,
+      "loss": 0.5266,
+      "step": 12610
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.3126575117008692e-05,
+      "loss": 0.6895,
+      "step": 12611
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.3124003497402665e-05,
+      "loss": 0.4081,
+      "step": 12612
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.312143187779664e-05,
+      "loss": 0.5758,
+      "step": 12613
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.3118860258190608e-05,
+      "loss": 0.5261,
+      "step": 12614
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.311628863858458e-05,
+      "loss": 0.6478,
+      "step": 12615
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.3113717018978554e-05,
+      "loss": 0.8542,
+      "step": 12616
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.3111145399372524e-05,
+      "loss": 0.3779,
+      "step": 12617
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.3108573779766497e-05,
+      "loss": 0.7704,
+      "step": 12618
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.310600216016047e-05,
+      "loss": 0.5199,
+      "step": 12619
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.310343054055444e-05,
+      "loss": 0.7397,
+      "step": 12620
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.3100858920948416e-05,
+      "loss": 0.4263,
+      "step": 12621
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.3098287301342386e-05,
+      "loss": 0.5453,
+      "step": 12622
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.309571568173636e-05,
+      "loss": 0.8194,
+      "step": 12623
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.3093144062130332e-05,
+      "loss": 0.59,
+      "step": 12624
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.3090572442524302e-05,
+      "loss": 0.788,
+      "step": 12625
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.3088000822918275e-05,
+      "loss": 0.5904,
+      "step": 12626
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.3085429203312248e-05,
+      "loss": 0.8193,
+      "step": 12627
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.3082857583706218e-05,
+      "loss": 0.8105,
+      "step": 12628
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.308028596410019e-05,
+      "loss": 0.621,
+      "step": 12629
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.3077714344494164e-05,
+      "loss": 0.9249,
+      "step": 12630
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.3075142724888137e-05,
+      "loss": 0.8402,
+      "step": 12631
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.3072571105282107e-05,
+      "loss": 0.5121,
+      "step": 12632
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.306999948567608e-05,
+      "loss": 0.7269,
+      "step": 12633
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.3067427866070053e-05,
+      "loss": 0.5444,
+      "step": 12634
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.3064856246464022e-05,
+      "loss": 0.7378,
+      "step": 12635
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.3062284626857995e-05,
+      "loss": 0.5046,
+      "step": 12636
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.305971300725197e-05,
+      "loss": 0.6767,
+      "step": 12637
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.305714138764594e-05,
+      "loss": 0.5517,
+      "step": 12638
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.3054569768039915e-05,
+      "loss": 0.695,
+      "step": 12639
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.3051998148433884e-05,
+      "loss": 0.4296,
+      "step": 12640
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.3049426528827857e-05,
+      "loss": 0.9868,
+      "step": 12641
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.304685490922183e-05,
+      "loss": 1.0613,
+      "step": 12642
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.30442832896158e-05,
+      "loss": 1.0799,
+      "step": 12643
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.3041711670009773e-05,
+      "loss": 0.9125,
+      "step": 12644
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.3039140050403746e-05,
+      "loss": 0.4188,
+      "step": 12645
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.3036568430797716e-05,
+      "loss": 0.9804,
+      "step": 12646
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.3033996811191692e-05,
+      "loss": 0.5125,
+      "step": 12647
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.3031425191585662e-05,
+      "loss": 0.5017,
+      "step": 12648
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.3028853571979632e-05,
+      "loss": 0.9022,
+      "step": 12649
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.3026281952373608e-05,
+      "loss": 0.7644,
+      "step": 12650
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.3023710332767578e-05,
+      "loss": 0.8136,
+      "step": 12651
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.302113871316155e-05,
+      "loss": 0.9316,
+      "step": 12652
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.3018567093555524e-05,
+      "loss": 0.7381,
+      "step": 12653
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.3015995473949494e-05,
+      "loss": 0.7309,
+      "step": 12654
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.3013423854343467e-05,
+      "loss": 0.597,
+      "step": 12655
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.301085223473744e-05,
+      "loss": 0.6383,
+      "step": 12656
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.300828061513141e-05,
+      "loss": 0.6779,
+      "step": 12657
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.3005708995525383e-05,
+      "loss": 0.8023,
+      "step": 12658
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.3003137375919356e-05,
+      "loss": 0.9352,
+      "step": 12659
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.300056575631333e-05,
+      "loss": 0.9787,
+      "step": 12660
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.29979941367073e-05,
+      "loss": 0.648,
+      "step": 12661
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.299542251710127e-05,
+      "loss": 0.8763,
+      "step": 12662
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.2992850897495245e-05,
+      "loss": 0.6105,
+      "step": 12663
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.2990279277889214e-05,
+      "loss": 0.6416,
+      "step": 12664
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.2987707658283187e-05,
+      "loss": 0.8278,
+      "step": 12665
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.298513603867716e-05,
+      "loss": 0.7331,
+      "step": 12666
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.298256441907113e-05,
+      "loss": 0.9895,
+      "step": 12667
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.2979992799465107e-05,
+      "loss": 0.7103,
+      "step": 12668
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.2977421179859076e-05,
+      "loss": 0.5963,
+      "step": 12669
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.297484956025305e-05,
+      "loss": 0.5374,
+      "step": 12670
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.2972277940647022e-05,
+      "loss": 0.8537,
+      "step": 12671
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.2969706321040992e-05,
+      "loss": 1.2525,
+      "step": 12672
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.2967134701434965e-05,
+      "loss": 0.6378,
+      "step": 12673
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.296456308182894e-05,
+      "loss": 0.8664,
+      "step": 12674
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.2961991462222908e-05,
+      "loss": 0.8062,
+      "step": 12675
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.295941984261688e-05,
+      "loss": 0.7078,
+      "step": 12676
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.2956848223010854e-05,
+      "loss": 0.6592,
+      "step": 12677
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.2954276603404827e-05,
+      "loss": 0.998,
+      "step": 12678
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.2951704983798797e-05,
+      "loss": 0.8935,
+      "step": 12679
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.294913336419277e-05,
+      "loss": 0.6653,
+      "step": 12680
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.2946561744586743e-05,
+      "loss": 0.6182,
+      "step": 12681
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.2943990124980713e-05,
+      "loss": 0.9354,
+      "step": 12682
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.2941418505374686e-05,
+      "loss": 0.5666,
+      "step": 12683
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.293884688576866e-05,
+      "loss": 0.9944,
+      "step": 12684
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.293627526616263e-05,
+      "loss": 0.6133,
+      "step": 12685
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.2933703646556605e-05,
+      "loss": 0.5389,
+      "step": 12686
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.2931132026950575e-05,
+      "loss": 0.7186,
+      "step": 12687
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.2928560407344544e-05,
+      "loss": 0.6117,
+      "step": 12688
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.292598878773852e-05,
+      "loss": 0.4638,
+      "step": 12689
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.292341716813249e-05,
+      "loss": 0.6892,
+      "step": 12690
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.2920845548526464e-05,
+      "loss": 0.8824,
+      "step": 12691
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.2918273928920437e-05,
+      "loss": 0.5904,
+      "step": 12692
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.2915702309314406e-05,
+      "loss": 0.5641,
+      "step": 12693
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.291313068970838e-05,
+      "loss": 0.6812,
+      "step": 12694
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.2910559070102353e-05,
+      "loss": 0.9408,
+      "step": 12695
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.2907987450496322e-05,
+      "loss": 0.8407,
+      "step": 12696
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.2905415830890295e-05,
+      "loss": 0.6821,
+      "step": 12697
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.290284421128427e-05,
+      "loss": 0.5403,
+      "step": 12698
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.290027259167824e-05,
+      "loss": 0.7643,
+      "step": 12699
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.289770097207221e-05,
+      "loss": 0.9396,
+      "step": 12700
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.2895129352466184e-05,
+      "loss": 0.8017,
+      "step": 12701
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.2892557732860157e-05,
+      "loss": 0.6254,
+      "step": 12702
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.2889986113254127e-05,
+      "loss": 0.5786,
+      "step": 12703
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.28874144936481e-05,
+      "loss": 0.5955,
+      "step": 12704
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.2884842874042073e-05,
+      "loss": 0.5461,
+      "step": 12705
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.2882271254436043e-05,
+      "loss": 0.6963,
+      "step": 12706
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.287969963483002e-05,
+      "loss": 0.74,
+      "step": 12707
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.287712801522399e-05,
+      "loss": 0.9107,
+      "step": 12708
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.2874556395617962e-05,
+      "loss": 0.7779,
+      "step": 12709
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.2871984776011935e-05,
+      "loss": 0.6484,
+      "step": 12710
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.2869413156405905e-05,
+      "loss": 0.7151,
+      "step": 12711
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.2866841536799878e-05,
+      "loss": 0.7242,
+      "step": 12712
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.286426991719385e-05,
+      "loss": 0.3803,
+      "step": 12713
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.286169829758782e-05,
+      "loss": 0.5574,
+      "step": 12714
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.2859126677981794e-05,
+      "loss": 0.8286,
+      "step": 12715
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.2856555058375767e-05,
+      "loss": 0.5409,
+      "step": 12716
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.285398343876974e-05,
+      "loss": 0.5185,
+      "step": 12717
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.285141181916371e-05,
+      "loss": 0.4121,
+      "step": 12718
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.2848840199557683e-05,
+      "loss": 0.4211,
+      "step": 12719
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.2846268579951656e-05,
+      "loss": 0.5347,
+      "step": 12720
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.2843696960345625e-05,
+      "loss": 0.5469,
+      "step": 12721
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.28411253407396e-05,
+      "loss": 0.8277,
+      "step": 12722
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.283855372113357e-05,
+      "loss": 0.7852,
+      "step": 12723
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.283598210152754e-05,
+      "loss": 0.9439,
+      "step": 12724
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.2833410481921518e-05,
+      "loss": 0.8307,
+      "step": 12725
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.2830838862315487e-05,
+      "loss": 0.5662,
+      "step": 12726
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.2828267242709457e-05,
+      "loss": 0.7126,
+      "step": 12727
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.2825695623103433e-05,
+      "loss": 1.0611,
+      "step": 12728
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.2823124003497403e-05,
+      "loss": 0.9105,
+      "step": 12729
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.2820552383891376e-05,
+      "loss": 0.7,
+      "step": 12730
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.281798076428535e-05,
+      "loss": 0.822,
+      "step": 12731
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.281540914467932e-05,
+      "loss": 0.6417,
+      "step": 12732
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.2812837525073292e-05,
+      "loss": 0.6676,
+      "step": 12733
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.2810265905467265e-05,
+      "loss": 0.4857,
+      "step": 12734
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.2807694285861235e-05,
+      "loss": 0.9018,
+      "step": 12735
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.2805122666255208e-05,
+      "loss": 0.5518,
+      "step": 12736
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.280255104664918e-05,
+      "loss": 0.7134,
+      "step": 12737
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.2799979427043154e-05,
+      "loss": 0.818,
+      "step": 12738
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.2797407807437124e-05,
+      "loss": 0.9532,
+      "step": 12739
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.2794836187831097e-05,
+      "loss": 0.5354,
+      "step": 12740
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.279226456822507e-05,
+      "loss": 0.7286,
+      "step": 12741
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.278969294861904e-05,
+      "loss": 0.8393,
+      "step": 12742
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.2787121329013013e-05,
+      "loss": 0.5797,
+      "step": 12743
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.2784549709406986e-05,
+      "loss": 0.5721,
+      "step": 12744
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.2781978089800955e-05,
+      "loss": 0.6179,
+      "step": 12745
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.2779406470194932e-05,
+      "loss": 0.4341,
+      "step": 12746
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.27768348505889e-05,
+      "loss": 0.7113,
+      "step": 12747
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.2774263230982875e-05,
+      "loss": 0.6795,
+      "step": 12748
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.2771691611376848e-05,
+      "loss": 0.8071,
+      "step": 12749
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.2769119991770817e-05,
+      "loss": 0.7149,
+      "step": 12750
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.276654837216479e-05,
+      "loss": 0.4981,
+      "step": 12751
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.2763976752558764e-05,
+      "loss": 0.6627,
+      "step": 12752
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.2761405132952733e-05,
+      "loss": 0.5762,
+      "step": 12753
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.2758833513346706e-05,
+      "loss": 0.84,
+      "step": 12754
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.275626189374068e-05,
+      "loss": 0.5242,
+      "step": 12755
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.2753690274134652e-05,
+      "loss": 0.4734,
+      "step": 12756
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.2751118654528622e-05,
+      "loss": 0.5223,
+      "step": 12757
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.2748547034922595e-05,
+      "loss": 0.8069,
+      "step": 12758
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.2745975415316568e-05,
+      "loss": 0.6928,
+      "step": 12759
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.2743403795710538e-05,
+      "loss": 0.8921,
+      "step": 12760
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.274083217610451e-05,
+      "loss": 0.8488,
+      "step": 12761
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.2738260556498484e-05,
+      "loss": 0.6629,
+      "step": 12762
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.2735688936892454e-05,
+      "loss": 0.9258,
+      "step": 12763
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.273311731728643e-05,
+      "loss": 0.7267,
+      "step": 12764
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.27305456976804e-05,
+      "loss": 0.9892,
+      "step": 12765
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.272797407807437e-05,
+      "loss": 1.0474,
+      "step": 12766
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.2725402458468346e-05,
+      "loss": 0.9079,
+      "step": 12767
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.2722830838862316e-05,
+      "loss": 0.6883,
+      "step": 12768
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.272025921925629e-05,
+      "loss": 0.6543,
+      "step": 12769
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.2717687599650262e-05,
+      "loss": 0.6802,
+      "step": 12770
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.271511598004423e-05,
+      "loss": 0.4557,
+      "step": 12771
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.2712544360438205e-05,
+      "loss": 0.7589,
+      "step": 12772
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.2709972740832178e-05,
+      "loss": 1.0518,
+      "step": 12773
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.2707401121226147e-05,
+      "loss": 0.5193,
+      "step": 12774
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.270482950162012e-05,
+      "loss": 0.4559,
+      "step": 12775
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.2702257882014094e-05,
+      "loss": 0.8622,
+      "step": 12776
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.2699686262408067e-05,
+      "loss": 0.6636,
+      "step": 12777
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.2697114642802036e-05,
+      "loss": 0.9291,
+      "step": 12778
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.269454302319601e-05,
+      "loss": 0.6179,
+      "step": 12779
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.2691971403589982e-05,
+      "loss": 0.7015,
+      "step": 12780
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.2689399783983952e-05,
+      "loss": 0.7541,
+      "step": 12781
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.2686828164377925e-05,
+      "loss": 0.7093,
+      "step": 12782
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.2684256544771898e-05,
+      "loss": 0.683,
+      "step": 12783
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.2681684925165868e-05,
+      "loss": 0.6139,
+      "step": 12784
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.2679113305559844e-05,
+      "loss": 0.6026,
+      "step": 12785
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.2676541685953814e-05,
+      "loss": 0.5533,
+      "step": 12786
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.2673970066347787e-05,
+      "loss": 0.8759,
+      "step": 12787
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.267139844674176e-05,
+      "loss": 0.8353,
+      "step": 12788
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.266882682713573e-05,
+      "loss": 0.8161,
+      "step": 12789
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.2666255207529703e-05,
+      "loss": 0.7504,
+      "step": 12790
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.2663683587923676e-05,
+      "loss": 0.8615,
+      "step": 12791
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.2661111968317646e-05,
+      "loss": 0.7813,
+      "step": 12792
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.2658540348711622e-05,
+      "loss": 0.6547,
+      "step": 12793
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.2655968729105592e-05,
+      "loss": 0.7568,
+      "step": 12794
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.2653397109499565e-05,
+      "loss": 0.7717,
+      "step": 12795
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.2650825489893538e-05,
+      "loss": 1.0401,
+      "step": 12796
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.2648253870287508e-05,
+      "loss": 0.6938,
+      "step": 12797
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.264568225068148e-05,
+      "loss": 0.5056,
+      "step": 12798
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.2643110631075454e-05,
+      "loss": 0.694,
+      "step": 12799
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.2640539011469424e-05,
+      "loss": 0.6072,
+      "step": 12800
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.2637967391863397e-05,
+      "loss": 0.8387,
+      "step": 12801
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.263539577225737e-05,
+      "loss": 0.8721,
+      "step": 12802
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.2632824152651343e-05,
+      "loss": 0.6587,
+      "step": 12803
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.2630252533045313e-05,
+      "loss": 0.6606,
+      "step": 12804
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.2627680913439286e-05,
+      "loss": 0.9912,
+      "step": 12805
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.262510929383326e-05,
+      "loss": 0.6371,
+      "step": 12806
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.262253767422723e-05,
+      "loss": 0.9834,
+      "step": 12807
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.26199660546212e-05,
+      "loss": 0.5647,
+      "step": 12808
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.2617394435015174e-05,
+      "loss": 0.7496,
+      "step": 12809
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.2614822815409144e-05,
+      "loss": 0.6317,
+      "step": 12810
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.261225119580312e-05,
+      "loss": 0.7632,
+      "step": 12811
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.260967957619709e-05,
+      "loss": 0.5448,
+      "step": 12812
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.260710795659106e-05,
+      "loss": 0.7745,
+      "step": 12813
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.2604536336985036e-05,
+      "loss": 0.9262,
+      "step": 12814
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.2601964717379006e-05,
+      "loss": 0.7912,
+      "step": 12815
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.259939309777298e-05,
+      "loss": 0.6009,
+      "step": 12816
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.2596821478166952e-05,
+      "loss": 0.4694,
+      "step": 12817
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.2594249858560922e-05,
+      "loss": 1.1098,
+      "step": 12818
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.2591678238954895e-05,
+      "loss": 0.7382,
+      "step": 12819
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.2589106619348868e-05,
+      "loss": 0.7399,
+      "step": 12820
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.2586534999742838e-05,
+      "loss": 0.6374,
+      "step": 12821
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.258396338013681e-05,
+      "loss": 0.7732,
+      "step": 12822
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.2581391760530784e-05,
+      "loss": 0.9183,
+      "step": 12823
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.2578820140924757e-05,
+      "loss": 1.0251,
+      "step": 12824
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.2576248521318727e-05,
+      "loss": 0.7021,
+      "step": 12825
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.25736769017127e-05,
+      "loss": 0.7099,
+      "step": 12826
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.2571105282106673e-05,
+      "loss": 0.4741,
+      "step": 12827
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.2568533662500643e-05,
+      "loss": 0.8739,
+      "step": 12828
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.2565962042894616e-05,
+      "loss": 0.4193,
+      "step": 12829
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.256339042328859e-05,
+      "loss": 0.641,
+      "step": 12830
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.256081880368256e-05,
+      "loss": 0.6117,
+      "step": 12831
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.2558247184076535e-05,
+      "loss": 0.7569,
+      "step": 12832
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.2555675564470505e-05,
+      "loss": 0.6262,
+      "step": 12833
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.2553103944864478e-05,
+      "loss": 0.6851,
+      "step": 12834
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.255053232525845e-05,
+      "loss": 0.8155,
+      "step": 12835
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.254796070565242e-05,
+      "loss": 0.9776,
+      "step": 12836
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.2545389086046393e-05,
+      "loss": 0.754,
+      "step": 12837
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.2542817466440367e-05,
+      "loss": 0.6727,
+      "step": 12838
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.2540245846834336e-05,
+      "loss": 0.558,
+      "step": 12839
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.253767422722831e-05,
+      "loss": 0.5794,
+      "step": 12840
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.2535102607622282e-05,
+      "loss": 0.7289,
+      "step": 12841
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.2532530988016255e-05,
+      "loss": 0.6087,
+      "step": 12842
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.2529959368410225e-05,
+      "loss": 0.4525,
+      "step": 12843
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.2527387748804198e-05,
+      "loss": 0.8262,
+      "step": 12844
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.252481612919817e-05,
+      "loss": 0.904,
+      "step": 12845
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.252224450959214e-05,
+      "loss": 0.4711,
+      "step": 12846
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.2519672889986114e-05,
+      "loss": 0.6294,
+      "step": 12847
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.2517101270380087e-05,
+      "loss": 0.7025,
+      "step": 12848
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.2514529650774057e-05,
+      "loss": 0.3779,
+      "step": 12849
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.2511958031168033e-05,
+      "loss": 0.4624,
+      "step": 12850
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.2509386411562003e-05,
+      "loss": 0.5806,
+      "step": 12851
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.2506814791955973e-05,
+      "loss": 0.9118,
+      "step": 12852
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.250424317234995e-05,
+      "loss": 0.6254,
+      "step": 12853
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.250167155274392e-05,
+      "loss": 0.5323,
+      "step": 12854
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.2499099933137892e-05,
+      "loss": 0.6085,
+      "step": 12855
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.2496528313531865e-05,
+      "loss": 0.6551,
+      "step": 12856
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.2493956693925835e-05,
+      "loss": 0.697,
+      "step": 12857
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.2491385074319808e-05,
+      "loss": 0.8528,
+      "step": 12858
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.248881345471378e-05,
+      "loss": 0.4313,
+      "step": 12859
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.248624183510775e-05,
+      "loss": 0.6323,
+      "step": 12860
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.2483670215501723e-05,
+      "loss": 0.4594,
+      "step": 12861
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.2481098595895697e-05,
+      "loss": 0.714,
+      "step": 12862
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.247852697628967e-05,
+      "loss": 0.742,
+      "step": 12863
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.247595535668364e-05,
+      "loss": 0.4716,
+      "step": 12864
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.2473383737077612e-05,
+      "loss": 0.8184,
+      "step": 12865
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.2470812117471585e-05,
+      "loss": 0.821,
+      "step": 12866
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.2468240497865555e-05,
+      "loss": 0.5771,
+      "step": 12867
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.2465668878259528e-05,
+      "loss": 0.7718,
+      "step": 12868
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.24630972586535e-05,
+      "loss": 0.6336,
+      "step": 12869
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.246052563904747e-05,
+      "loss": 0.6024,
+      "step": 12870
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.2457954019441447e-05,
+      "loss": 0.5575,
+      "step": 12871
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.2455382399835417e-05,
+      "loss": 0.6444,
+      "step": 12872
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.245281078022939e-05,
+      "loss": 1.1402,
+      "step": 12873
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.2450239160623363e-05,
+      "loss": 0.5731,
+      "step": 12874
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.2447667541017333e-05,
+      "loss": 0.5617,
+      "step": 12875
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.2445095921411306e-05,
+      "loss": 0.5453,
+      "step": 12876
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.244252430180528e-05,
+      "loss": 0.6098,
+      "step": 12877
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.243995268219925e-05,
+      "loss": 1.0829,
+      "step": 12878
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.2437381062593222e-05,
+      "loss": 1.0145,
+      "step": 12879
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.2434809442987195e-05,
+      "loss": 0.6607,
+      "step": 12880
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.2432237823381168e-05,
+      "loss": 0.6221,
+      "step": 12881
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.2429666203775138e-05,
+      "loss": 1.0089,
+      "step": 12882
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.242709458416911e-05,
+      "loss": 0.7766,
+      "step": 12883
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.2424522964563084e-05,
+      "loss": 0.7007,
+      "step": 12884
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.2421951344957054e-05,
+      "loss": 0.8571,
+      "step": 12885
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.2419379725351027e-05,
+      "loss": 0.5327,
+      "step": 12886
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.2416808105745e-05,
+      "loss": 0.5904,
+      "step": 12887
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.241423648613897e-05,
+      "loss": 0.6574,
+      "step": 12888
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.2411664866532946e-05,
+      "loss": 0.5834,
+      "step": 12889
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.2409093246926916e-05,
+      "loss": 0.6598,
+      "step": 12890
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.2406521627320885e-05,
+      "loss": 0.6084,
+      "step": 12891
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.240395000771486e-05,
+      "loss": 0.8021,
+      "step": 12892
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.240137838810883e-05,
+      "loss": 0.9806,
+      "step": 12893
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.2398806768502804e-05,
+      "loss": 0.9293,
+      "step": 12894
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.2396235148896777e-05,
+      "loss": 0.4204,
+      "step": 12895
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.2393663529290747e-05,
+      "loss": 0.6075,
+      "step": 12896
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.239109190968472e-05,
+      "loss": 0.602,
+      "step": 12897
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.2388520290078693e-05,
+      "loss": 0.4826,
+      "step": 12898
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.2385948670472663e-05,
+      "loss": 0.9165,
+      "step": 12899
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.2383377050866636e-05,
+      "loss": 0.8087,
+      "step": 12900
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.238080543126061e-05,
+      "loss": 0.6145,
+      "step": 12901
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.2378233811654582e-05,
+      "loss": 0.9283,
+      "step": 12902
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.2375662192048552e-05,
+      "loss": 0.6665,
+      "step": 12903
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.2373090572442525e-05,
+      "loss": 0.6231,
+      "step": 12904
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.2370518952836498e-05,
+      "loss": 0.7303,
+      "step": 12905
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.2367947333230468e-05,
+      "loss": 1.0724,
+      "step": 12906
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.236537571362444e-05,
+      "loss": 0.8068,
+      "step": 12907
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.2362804094018414e-05,
+      "loss": 0.8552,
+      "step": 12908
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.2360232474412384e-05,
+      "loss": 0.5749,
+      "step": 12909
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.235766085480636e-05,
+      "loss": 0.5906,
+      "step": 12910
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.235508923520033e-05,
+      "loss": 0.696,
+      "step": 12911
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.2352517615594303e-05,
+      "loss": 0.8669,
+      "step": 12912
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.2349945995988276e-05,
+      "loss": 0.7437,
+      "step": 12913
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.2347374376382246e-05,
+      "loss": 0.7966,
+      "step": 12914
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.234480275677622e-05,
+      "loss": 0.8622,
+      "step": 12915
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.2342231137170192e-05,
+      "loss": 0.6873,
+      "step": 12916
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.233965951756416e-05,
+      "loss": 0.8084,
+      "step": 12917
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.2337087897958134e-05,
+      "loss": 0.5765,
+      "step": 12918
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.2334516278352108e-05,
+      "loss": 0.7401,
+      "step": 12919
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.233194465874608e-05,
+      "loss": 0.5885,
+      "step": 12920
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.232937303914005e-05,
+      "loss": 0.729,
+      "step": 12921
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.2326801419534023e-05,
+      "loss": 0.8679,
+      "step": 12922
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.2324229799927996e-05,
+      "loss": 0.4512,
+      "step": 12923
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.2321658180321966e-05,
+      "loss": 0.9578,
+      "step": 12924
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.231908656071594e-05,
+      "loss": 0.6031,
+      "step": 12925
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.2316514941109912e-05,
+      "loss": 0.8794,
+      "step": 12926
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.2313943321503882e-05,
+      "loss": 0.7363,
+      "step": 12927
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.231137170189786e-05,
+      "loss": 0.6995,
+      "step": 12928
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.2308800082291828e-05,
+      "loss": 0.3704,
+      "step": 12929
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.2306228462685798e-05,
+      "loss": 0.7491,
+      "step": 12930
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.2303656843079774e-05,
+      "loss": 0.8354,
+      "step": 12931
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.2301085223473744e-05,
+      "loss": 0.8646,
+      "step": 12932
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.2298513603867717e-05,
+      "loss": 0.715,
+      "step": 12933
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.229594198426169e-05,
+      "loss": 0.8209,
+      "step": 12934
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.229337036465566e-05,
+      "loss": 1.0111,
+      "step": 12935
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.2290798745049633e-05,
+      "loss": 0.8245,
+      "step": 12936
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.2288227125443606e-05,
+      "loss": 0.6081,
+      "step": 12937
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.2285655505837576e-05,
+      "loss": 0.6319,
+      "step": 12938
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.2283083886231552e-05,
+      "loss": 0.9403,
+      "step": 12939
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.2280512266625522e-05,
+      "loss": 0.6311,
+      "step": 12940
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.2277940647019495e-05,
+      "loss": 0.7097,
+      "step": 12941
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.2275369027413468e-05,
+      "loss": 0.8424,
+      "step": 12942
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.2272797407807438e-05,
+      "loss": 0.5608,
+      "step": 12943
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.227022578820141e-05,
+      "loss": 0.8661,
+      "step": 12944
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.2267654168595384e-05,
+      "loss": 0.7283,
+      "step": 12945
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.2265082548989353e-05,
+      "loss": 0.8466,
+      "step": 12946
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.2262510929383326e-05,
+      "loss": 0.6731,
+      "step": 12947
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.22599393097773e-05,
+      "loss": 0.7762,
+      "step": 12948
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.2257367690171273e-05,
+      "loss": 0.5384,
+      "step": 12949
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.2254796070565242e-05,
+      "loss": 0.9401,
+      "step": 12950
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.2252224450959215e-05,
+      "loss": 0.7792,
+      "step": 12951
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.224965283135319e-05,
+      "loss": 0.6917,
+      "step": 12952
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.2247081211747158e-05,
+      "loss": 0.9477,
+      "step": 12953
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.224450959214113e-05,
+      "loss": 0.8346,
+      "step": 12954
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.2241937972535104e-05,
+      "loss": 0.6515,
+      "step": 12955
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.2239366352929074e-05,
+      "loss": 0.7025,
+      "step": 12956
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.223679473332305e-05,
+      "loss": 0.7099,
+      "step": 12957
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.223422311371702e-05,
+      "loss": 0.9076,
+      "step": 12958
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.2231651494110993e-05,
+      "loss": 0.5598,
+      "step": 12959
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.2229079874504966e-05,
+      "loss": 0.8715,
+      "step": 12960
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.2226508254898936e-05,
+      "loss": 0.7485,
+      "step": 12961
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.222393663529291e-05,
+      "loss": 0.8501,
+      "step": 12962
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.2221365015686882e-05,
+      "loss": 0.5552,
+      "step": 12963
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.2218793396080852e-05,
+      "loss": 0.7803,
+      "step": 12964
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.2216221776474825e-05,
+      "loss": 0.815,
+      "step": 12965
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.2213650156868798e-05,
+      "loss": 0.7023,
+      "step": 12966
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.221107853726277e-05,
+      "loss": 0.5953,
+      "step": 12967
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.220850691765674e-05,
+      "loss": 0.5854,
+      "step": 12968
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.2205935298050714e-05,
+      "loss": 0.8068,
+      "step": 12969
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.2203363678444687e-05,
+      "loss": 0.9589,
+      "step": 12970
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.2200792058838657e-05,
+      "loss": 0.7405,
+      "step": 12971
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.219822043923263e-05,
+      "loss": 0.4564,
+      "step": 12972
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.2195648819626603e-05,
+      "loss": 0.7425,
+      "step": 12973
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.2193077200020572e-05,
+      "loss": 0.5518,
+      "step": 12974
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.219050558041455e-05,
+      "loss": 0.8349,
+      "step": 12975
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.218793396080852e-05,
+      "loss": 0.381,
+      "step": 12976
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.2185362341202488e-05,
+      "loss": 0.4448,
+      "step": 12977
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.2182790721596465e-05,
+      "loss": 0.3617,
+      "step": 12978
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.2180219101990434e-05,
+      "loss": 0.5121,
+      "step": 12979
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.2177647482384407e-05,
+      "loss": 0.6625,
+      "step": 12980
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.217507586277838e-05,
+      "loss": 0.597,
+      "step": 12981
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.217250424317235e-05,
+      "loss": 0.4859,
+      "step": 12982
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.2169932623566323e-05,
+      "loss": 0.6398,
+      "step": 12983
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.2167361003960296e-05,
+      "loss": 0.6079,
+      "step": 12984
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.2164789384354266e-05,
+      "loss": 0.7407,
+      "step": 12985
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.216221776474824e-05,
+      "loss": 1.1346,
+      "step": 12986
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.2159646145142212e-05,
+      "loss": 0.6194,
+      "step": 12987
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.2157074525536185e-05,
+      "loss": 0.4915,
+      "step": 12988
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.2154502905930155e-05,
+      "loss": 0.769,
+      "step": 12989
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.2151931286324128e-05,
+      "loss": 0.774,
+      "step": 12990
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 2.21493596667181e-05,
+      "loss": 1.0161,
+      "step": 12991
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 2.214678804711207e-05,
+      "loss": 0.8025,
+      "step": 12992
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 2.2144216427506044e-05,
+      "loss": 0.6682,
+      "step": 12993
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 2.2141644807900017e-05,
+      "loss": 0.7778,
+      "step": 12994
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 2.2139073188293987e-05,
+      "loss": 0.7703,
+      "step": 12995
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 2.2136501568687963e-05,
+      "loss": 0.6345,
+      "step": 12996
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 2.2133929949081933e-05,
+      "loss": 0.6062,
+      "step": 12997
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 2.2131358329475906e-05,
+      "loss": 0.5619,
+      "step": 12998
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 2.212878670986988e-05,
+      "loss": 0.4496,
+      "step": 12999
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 2.212621509026385e-05,
+      "loss": 0.9381,
+      "step": 13000
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 2.212364347065782e-05,
+      "loss": 0.8467,
+      "step": 13001
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 2.2121071851051795e-05,
+      "loss": 0.7107,
+      "step": 13002
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 2.2118500231445764e-05,
+      "loss": 0.6925,
+      "step": 13003
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 2.2115928611839737e-05,
+      "loss": 0.3033,
+      "step": 13004
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 2.211335699223371e-05,
+      "loss": 0.5475,
+      "step": 13005
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 2.2110785372627684e-05,
+      "loss": 0.6809,
+      "step": 13006
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 2.2108213753021653e-05,
+      "loss": 0.6599,
+      "step": 13007
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 2.2105642133415626e-05,
+      "loss": 1.1288,
+      "step": 13008
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 2.21030705138096e-05,
+      "loss": 0.6893,
+      "step": 13009
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 2.210049889420357e-05,
+      "loss": 0.4862,
+      "step": 13010
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 2.2097927274597542e-05,
+      "loss": 0.5957,
+      "step": 13011
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 2.2095355654991515e-05,
+      "loss": 0.5215,
+      "step": 13012
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 2.2092784035385485e-05,
+      "loss": 0.477,
+      "step": 13013
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 2.209021241577946e-05,
+      "loss": 0.8672,
+      "step": 13014
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 2.208764079617343e-05,
+      "loss": 0.6251,
+      "step": 13015
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 2.20850691765674e-05,
+      "loss": 0.5565,
+      "step": 13016
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 2.2082497556961377e-05,
+      "loss": 0.8205,
+      "step": 13017
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 2.2079925937355347e-05,
+      "loss": 0.9504,
+      "step": 13018
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 2.207735431774932e-05,
+      "loss": 1.1116,
+      "step": 13019
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 2.2074782698143293e-05,
+      "loss": 0.9163,
+      "step": 13020
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 2.2072211078537263e-05,
+      "loss": 0.7142,
+      "step": 13021
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 2.2069639458931236e-05,
+      "loss": 0.5401,
+      "step": 13022
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 2.206706783932521e-05,
+      "loss": 0.6257,
+      "step": 13023
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 2.206449621971918e-05,
+      "loss": 0.6308,
+      "step": 13024
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 2.206192460011315e-05,
+      "loss": 0.7695,
+      "step": 13025
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 2.2059352980507125e-05,
+      "loss": 0.6069,
+      "step": 13026
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 2.2056781360901098e-05,
+      "loss": 0.7111,
+      "step": 13027
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 2.2054209741295068e-05,
+      "loss": 0.4515,
+      "step": 13028
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 2.205163812168904e-05,
+      "loss": 0.551,
+      "step": 13029
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 2.2049066502083014e-05,
+      "loss": 0.708,
+      "step": 13030
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 2.2046494882476983e-05,
+      "loss": 0.8679,
+      "step": 13031
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 2.2043923262870956e-05,
+      "loss": 0.4989,
+      "step": 13032
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 2.204135164326493e-05,
+      "loss": 0.6502,
+      "step": 13033
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 2.20387800236589e-05,
+      "loss": 0.7036,
+      "step": 13034
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 2.2036208404052876e-05,
+      "loss": 0.6883,
+      "step": 13035
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 2.2033636784446845e-05,
+      "loss": 0.4978,
+      "step": 13036
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 2.203106516484082e-05,
+      "loss": 0.8722,
+      "step": 13037
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 2.202849354523479e-05,
+      "loss": 0.6651,
+      "step": 13038
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 2.202592192562876e-05,
+      "loss": 0.7936,
+      "step": 13039
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 2.2023350306022734e-05,
+      "loss": 0.5941,
+      "step": 13040
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 2.2020778686416707e-05,
+      "loss": 0.8867,
+      "step": 13041
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 2.2018207066810677e-05,
+      "loss": 1.0079,
+      "step": 13042
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 2.201563544720465e-05,
+      "loss": 1.0901,
+      "step": 13043
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 2.2013063827598623e-05,
+      "loss": 0.7558,
+      "step": 13044
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 2.2010492207992596e-05,
+      "loss": 0.7218,
+      "step": 13045
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 2.2007920588386566e-05,
+      "loss": 0.5993,
+      "step": 13046
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 2.200534896878054e-05,
+      "loss": 0.7354,
+      "step": 13047
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 2.2002777349174512e-05,
+      "loss": 0.8864,
+      "step": 13048
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 2.2000205729568482e-05,
+      "loss": 0.6439,
+      "step": 13049
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 2.1997634109962455e-05,
+      "loss": 0.7038,
+      "step": 13050
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 2.1995062490356428e-05,
+      "loss": 0.7016,
+      "step": 13051
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 2.1992490870750398e-05,
+      "loss": 0.8457,
+      "step": 13052
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 2.1989919251144374e-05,
+      "loss": 0.8743,
+      "step": 13053
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 2.1987347631538344e-05,
+      "loss": 0.611,
+      "step": 13054
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 2.1984776011932313e-05,
+      "loss": 0.6214,
+      "step": 13055
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 2.198220439232629e-05,
+      "loss": 0.7378,
+      "step": 13056
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 2.197963277272026e-05,
+      "loss": 0.8294,
+      "step": 13057
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 2.1977061153114233e-05,
+      "loss": 0.756,
+      "step": 13058
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 2.1974489533508206e-05,
+      "loss": 0.5709,
+      "step": 13059
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 2.1971917913902175e-05,
+      "loss": 0.7557,
+      "step": 13060
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 2.196934629429615e-05,
+      "loss": 0.9377,
+      "step": 13061
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 2.196677467469012e-05,
+      "loss": 0.6989,
+      "step": 13062
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 2.196420305508409e-05,
+      "loss": 0.7176,
+      "step": 13063
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 2.1961631435478064e-05,
+      "loss": 0.8796,
+      "step": 13064
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 2.1959059815872037e-05,
+      "loss": 0.7031,
+      "step": 13065
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 2.195648819626601e-05,
+      "loss": 0.5552,
+      "step": 13066
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 2.195391657665998e-05,
+      "loss": 0.5906,
+      "step": 13067
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 2.1951344957053953e-05,
+      "loss": 1.1062,
+      "step": 13068
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 2.1948773337447926e-05,
+      "loss": 0.7573,
+      "step": 13069
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 2.1946201717841896e-05,
+      "loss": 0.7464,
+      "step": 13070
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 2.194363009823587e-05,
+      "loss": 0.7458,
+      "step": 13071
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 2.1941058478629842e-05,
+      "loss": 0.9201,
+      "step": 13072
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 2.1938486859023812e-05,
+      "loss": 0.6815,
+      "step": 13073
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 2.1935915239417788e-05,
+      "loss": 0.8812,
+      "step": 13074
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 2.1933343619811758e-05,
+      "loss": 0.6455,
+      "step": 13075
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 2.193077200020573e-05,
+      "loss": 0.9522,
+      "step": 13076
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 2.1928200380599704e-05,
+      "loss": 0.6405,
+      "step": 13077
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 2.1925628760993674e-05,
+      "loss": 1.0513,
+      "step": 13078
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 2.1923057141387647e-05,
+      "loss": 0.8413,
+      "step": 13079
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 2.192048552178162e-05,
+      "loss": 0.6372,
+      "step": 13080
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 2.191791390217559e-05,
+      "loss": 0.692,
+      "step": 13081
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 2.1915342282569563e-05,
+      "loss": 0.7374,
+      "step": 13082
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 2.1912770662963536e-05,
+      "loss": 0.7394,
+      "step": 13083
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 2.191019904335751e-05,
+      "loss": 0.3589,
+      "step": 13084
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 2.190762742375148e-05,
+      "loss": 0.6297,
+      "step": 13085
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 2.190505580414545e-05,
+      "loss": 0.5047,
+      "step": 13086
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 2.1902484184539425e-05,
+      "loss": 0.675,
+      "step": 13087
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 2.1899912564933394e-05,
+      "loss": 1.0671,
+      "step": 13088
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 2.1897340945327367e-05,
+      "loss": 0.7559,
+      "step": 13089
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 2.189476932572134e-05,
+      "loss": 0.5067,
+      "step": 13090
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 2.1892197706115314e-05,
+      "loss": 0.7804,
+      "step": 13091
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 2.1889626086509287e-05,
+      "loss": 0.6596,
+      "step": 13092
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 2.1887054466903256e-05,
+      "loss": 0.9475,
+      "step": 13093
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 2.188448284729723e-05,
+      "loss": 0.5869,
+      "step": 13094
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 2.1881911227691202e-05,
+      "loss": 0.6951,
+      "step": 13095
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 2.1879339608085172e-05,
+      "loss": 1.0801,
+      "step": 13096
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 2.1876767988479145e-05,
+      "loss": 0.6268,
+      "step": 13097
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 2.1874196368873118e-05,
+      "loss": 0.5407,
+      "step": 13098
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 2.1871624749267088e-05,
+      "loss": 0.7725,
+      "step": 13099
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 2.1869053129661064e-05,
+      "loss": 1.224,
+      "step": 13100
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 2.1866481510055034e-05,
+      "loss": 1.2094,
+      "step": 13101
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 2.1863909890449004e-05,
+      "loss": 1.0296,
+      "step": 13102
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 2.186133827084298e-05,
+      "loss": 0.6737,
+      "step": 13103
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 2.185876665123695e-05,
+      "loss": 0.339,
+      "step": 13104
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 2.1856195031630923e-05,
+      "loss": 0.8206,
+      "step": 13105
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 2.1853623412024896e-05,
+      "loss": 0.6625,
+      "step": 13106
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 2.1851051792418866e-05,
+      "loss": 0.9167,
+      "step": 13107
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 2.184848017281284e-05,
+      "loss": 0.6756,
+      "step": 13108
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 2.1845908553206812e-05,
+      "loss": 0.7134,
+      "step": 13109
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 2.184333693360078e-05,
+      "loss": 0.9264,
+      "step": 13110
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 2.1840765313994755e-05,
+      "loss": 0.9039,
+      "step": 13111
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 2.1838193694388728e-05,
+      "loss": 0.7762,
+      "step": 13112
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 2.18356220747827e-05,
+      "loss": 0.8078,
+      "step": 13113
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 2.183305045517667e-05,
+      "loss": 0.5791,
+      "step": 13114
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 2.1830478835570644e-05,
+      "loss": 0.8598,
+      "step": 13115
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 2.1827907215964617e-05,
+      "loss": 0.5363,
+      "step": 13116
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 2.1825335596358586e-05,
+      "loss": 0.6977,
+      "step": 13117
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 2.182276397675256e-05,
+      "loss": 0.5635,
+      "step": 13118
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 2.1820192357146532e-05,
+      "loss": 0.7098,
+      "step": 13119
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 2.1817620737540502e-05,
+      "loss": 0.5841,
+      "step": 13120
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 2.181504911793448e-05,
+      "loss": 0.581,
+      "step": 13121
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 2.181247749832845e-05,
+      "loss": 0.9067,
+      "step": 13122
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 2.180990587872242e-05,
+      "loss": 0.764,
+      "step": 13123
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 2.1807334259116394e-05,
+      "loss": 0.6937,
+      "step": 13124
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 2.1804762639510364e-05,
+      "loss": 0.8215,
+      "step": 13125
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 2.1802191019904337e-05,
+      "loss": 0.801,
+      "step": 13126
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 2.179961940029831e-05,
+      "loss": 0.7946,
+      "step": 13127
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 2.179704778069228e-05,
+      "loss": 0.7973,
+      "step": 13128
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 2.1794476161086253e-05,
+      "loss": 0.7538,
+      "step": 13129
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 2.1791904541480226e-05,
+      "loss": 0.7183,
+      "step": 13130
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 2.17893329218742e-05,
+      "loss": 0.678,
+      "step": 13131
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 2.178676130226817e-05,
+      "loss": 0.4636,
+      "step": 13132
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 2.1784189682662142e-05,
+      "loss": 0.859,
+      "step": 13133
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 2.1781618063056115e-05,
+      "loss": 0.8163,
+      "step": 13134
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 2.1779046443450085e-05,
+      "loss": 0.8219,
+      "step": 13135
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 2.1776474823844058e-05,
+      "loss": 0.4584,
+      "step": 13136
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 2.177390320423803e-05,
+      "loss": 0.6821,
+      "step": 13137
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 2.1771331584632e-05,
+      "loss": 0.7983,
+      "step": 13138
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 2.1768759965025977e-05,
+      "loss": 0.9091,
+      "step": 13139
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 2.1766188345419947e-05,
+      "loss": 0.7709,
+      "step": 13140
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 2.1763616725813916e-05,
+      "loss": 0.8883,
+      "step": 13141
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 2.1761045106207893e-05,
+      "loss": 1.0514,
+      "step": 13142
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 2.1758473486601863e-05,
+      "loss": 0.5777,
+      "step": 13143
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 2.1755901866995836e-05,
+      "loss": 0.877,
+      "step": 13144
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 2.175333024738981e-05,
+      "loss": 0.6488,
+      "step": 13145
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 2.175075862778378e-05,
+      "loss": 0.442,
+      "step": 13146
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 2.174818700817775e-05,
+      "loss": 0.6727,
+      "step": 13147
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 2.1745615388571725e-05,
+      "loss": 0.7671,
+      "step": 13148
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 2.1743043768965694e-05,
+      "loss": 0.5075,
+      "step": 13149
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 2.1740472149359667e-05,
+      "loss": 0.7813,
+      "step": 13150
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 2.173790052975364e-05,
+      "loss": 0.535,
+      "step": 13151
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 2.1735328910147613e-05,
+      "loss": 0.6305,
+      "step": 13152
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 2.1732757290541583e-05,
+      "loss": 0.7858,
+      "step": 13153
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 2.1730185670935556e-05,
+      "loss": 0.7542,
+      "step": 13154
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 2.172761405132953e-05,
+      "loss": 0.6693,
+      "step": 13155
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 2.17250424317235e-05,
+      "loss": 0.4821,
+      "step": 13156
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 2.1722470812117472e-05,
+      "loss": 0.52,
+      "step": 13157
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 2.1719899192511445e-05,
+      "loss": 0.5293,
+      "step": 13158
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 2.1717327572905415e-05,
+      "loss": 0.5904,
+      "step": 13159
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 2.171475595329939e-05,
+      "loss": 0.7956,
+      "step": 13160
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 2.171218433369336e-05,
+      "loss": 0.9737,
+      "step": 13161
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 2.1709612714087334e-05,
+      "loss": 0.798,
+      "step": 13162
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 2.1707041094481307e-05,
+      "loss": 0.829,
+      "step": 13163
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 2.1704469474875277e-05,
+      "loss": 0.537,
+      "step": 13164
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 2.170189785526925e-05,
+      "loss": 0.5943,
+      "step": 13165
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 2.1699326235663223e-05,
+      "loss": 0.9645,
+      "step": 13166
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 2.1696754616057193e-05,
+      "loss": 0.7848,
+      "step": 13167
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 2.1694182996451166e-05,
+      "loss": 0.5761,
+      "step": 13168
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 2.169161137684514e-05,
+      "loss": 0.7103,
+      "step": 13169
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 2.1689039757239112e-05,
+      "loss": 0.4713,
+      "step": 13170
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 2.168646813763308e-05,
+      "loss": 0.9887,
+      "step": 13171
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 2.1683896518027055e-05,
+      "loss": 0.5878,
+      "step": 13172
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 2.1681324898421028e-05,
+      "loss": 0.694,
+      "step": 13173
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 2.1678753278814997e-05,
+      "loss": 1.0366,
+      "step": 13174
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 2.167618165920897e-05,
+      "loss": 1.0777,
+      "step": 13175
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 2.1673610039602943e-05,
+      "loss": 0.5563,
+      "step": 13176
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 2.1671038419996913e-05,
+      "loss": 0.9552,
+      "step": 13177
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 2.166846680039089e-05,
+      "loss": 0.7029,
+      "step": 13178
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 2.166589518078486e-05,
+      "loss": 0.5763,
+      "step": 13179
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 2.166332356117883e-05,
+      "loss": 0.6893,
+      "step": 13180
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 2.1660751941572805e-05,
+      "loss": 0.6733,
+      "step": 13181
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 2.1658180321966775e-05,
+      "loss": 0.6548,
+      "step": 13182
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 2.1655608702360748e-05,
+      "loss": 0.7131,
+      "step": 13183
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 2.165303708275472e-05,
+      "loss": 0.6722,
+      "step": 13184
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 2.165046546314869e-05,
+      "loss": 0.7185,
+      "step": 13185
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 2.1647893843542664e-05,
+      "loss": 0.5177,
+      "step": 13186
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 2.1645322223936637e-05,
+      "loss": 0.565,
+      "step": 13187
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 2.1642750604330607e-05,
+      "loss": 0.6357,
+      "step": 13188
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 2.164017898472458e-05,
+      "loss": 0.5341,
+      "step": 13189
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 2.1637607365118553e-05,
+      "loss": 0.7873,
+      "step": 13190
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 2.1635035745512526e-05,
+      "loss": 0.6624,
+      "step": 13191
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 2.1632464125906496e-05,
+      "loss": 0.5512,
+      "step": 13192
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 2.162989250630047e-05,
+      "loss": 0.3878,
+      "step": 13193
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 2.1627320886694442e-05,
+      "loss": 0.6975,
+      "step": 13194
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 2.162474926708841e-05,
+      "loss": 0.6727,
+      "step": 13195
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 2.1622177647482385e-05,
+      "loss": 0.909,
+      "step": 13196
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 2.1619606027876358e-05,
+      "loss": 0.5021,
+      "step": 13197
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 2.1617034408270327e-05,
+      "loss": 0.6406,
+      "step": 13198
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 2.1614462788664304e-05,
+      "loss": 0.6646,
+      "step": 13199
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 2.1611891169058274e-05,
+      "loss": 0.682,
+      "step": 13200
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 2.1609319549452247e-05,
+      "loss": 0.6878,
+      "step": 13201
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 2.160674792984622e-05,
+      "loss": 0.8058,
+      "step": 13202
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 2.160417631024019e-05,
+      "loss": 0.9841,
+      "step": 13203
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 2.1601604690634162e-05,
+      "loss": 0.5546,
+      "step": 13204
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 2.1599033071028135e-05,
+      "loss": 0.7643,
+      "step": 13205
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 2.1596461451422105e-05,
+      "loss": 0.6783,
+      "step": 13206
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 2.1593889831816078e-05,
+      "loss": 0.6612,
+      "step": 13207
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 2.159131821221005e-05,
+      "loss": 0.8726,
+      "step": 13208
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 2.1588746592604024e-05,
+      "loss": 0.7086,
+      "step": 13209
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 2.1586174972997994e-05,
+      "loss": 0.9053,
+      "step": 13210
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 2.1583603353391967e-05,
+      "loss": 0.7188,
+      "step": 13211
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 2.158103173378594e-05,
+      "loss": 0.7376,
+      "step": 13212
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 2.157846011417991e-05,
+      "loss": 0.709,
+      "step": 13213
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 2.1575888494573883e-05,
+      "loss": 0.889,
+      "step": 13214
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 2.1573316874967856e-05,
+      "loss": 0.5798,
+      "step": 13215
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 2.1570745255361826e-05,
+      "loss": 0.9242,
+      "step": 13216
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 2.1568173635755802e-05,
+      "loss": 0.6319,
+      "step": 13217
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 2.1565602016149772e-05,
+      "loss": 0.7294,
+      "step": 13218
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 2.156303039654374e-05,
+      "loss": 0.6936,
+      "step": 13219
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 2.1560458776937718e-05,
+      "loss": 0.311,
+      "step": 13220
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 2.1557887157331688e-05,
+      "loss": 0.6735,
+      "step": 13221
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 2.155531553772566e-05,
+      "loss": 0.778,
+      "step": 13222
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 2.1552743918119634e-05,
+      "loss": 0.6859,
+      "step": 13223
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 2.1550172298513604e-05,
+      "loss": 0.6794,
+      "step": 13224
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 2.1547600678907577e-05,
+      "loss": 0.6602,
+      "step": 13225
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 2.154502905930155e-05,
+      "loss": 0.3955,
+      "step": 13226
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 2.154245743969552e-05,
+      "loss": 0.7169,
+      "step": 13227
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 2.1539885820089492e-05,
+      "loss": 0.5091,
+      "step": 13228
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 2.1537314200483466e-05,
+      "loss": 0.6667,
+      "step": 13229
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 2.153474258087744e-05,
+      "loss": 0.5914,
+      "step": 13230
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 2.1532170961271408e-05,
+      "loss": 0.6911,
+      "step": 13231
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 2.152959934166538e-05,
+      "loss": 0.9472,
+      "step": 13232
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 2.1527027722059354e-05,
+      "loss": 0.9943,
+      "step": 13233
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 2.1524456102453324e-05,
+      "loss": 0.7665,
+      "step": 13234
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 2.1521884482847297e-05,
+      "loss": 0.6325,
+      "step": 13235
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 2.151931286324127e-05,
+      "loss": 0.6,
+      "step": 13236
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 2.151674124363524e-05,
+      "loss": 0.5531,
+      "step": 13237
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 2.1514169624029216e-05,
+      "loss": 0.8657,
+      "step": 13238
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 2.1511598004423186e-05,
+      "loss": 0.6148,
+      "step": 13239
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 2.150902638481716e-05,
+      "loss": 0.5096,
+      "step": 13240
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 2.1506454765211132e-05,
+      "loss": 0.6128,
+      "step": 13241
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 2.1503883145605102e-05,
+      "loss": 0.7127,
+      "step": 13242
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 2.1501311525999075e-05,
+      "loss": 0.3435,
+      "step": 13243
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 2.1498739906393048e-05,
+      "loss": 0.8129,
+      "step": 13244
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 2.1496168286787018e-05,
+      "loss": 0.5606,
+      "step": 13245
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 2.1493596667180994e-05,
+      "loss": 0.6547,
+      "step": 13246
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 2.1491025047574964e-05,
+      "loss": 0.924,
+      "step": 13247
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 2.1488453427968937e-05,
+      "loss": 0.6921,
+      "step": 13248
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 2.148588180836291e-05,
+      "loss": 0.884,
+      "step": 13249
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 2.148331018875688e-05,
+      "loss": 0.6874,
+      "step": 13250
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 2.1480738569150853e-05,
+      "loss": 0.7309,
+      "step": 13251
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 2.1478166949544826e-05,
+      "loss": 0.6497,
+      "step": 13252
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 2.1475595329938796e-05,
+      "loss": 0.713,
+      "step": 13253
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 2.147302371033277e-05,
+      "loss": 0.5629,
+      "step": 13254
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 2.1470452090726742e-05,
+      "loss": 0.8724,
+      "step": 13255
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 2.1467880471120715e-05,
+      "loss": 0.6531,
+      "step": 13256
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 2.1465308851514684e-05,
+      "loss": 0.7807,
+      "step": 13257
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 2.1462737231908658e-05,
+      "loss": 0.6577,
+      "step": 13258
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 2.146016561230263e-05,
+      "loss": 0.61,
+      "step": 13259
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 2.14575939926966e-05,
+      "loss": 0.5128,
+      "step": 13260
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 2.1455022373090573e-05,
+      "loss": 0.7267,
+      "step": 13261
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 2.1452450753484546e-05,
+      "loss": 0.4917,
+      "step": 13262
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 2.1449879133878516e-05,
+      "loss": 0.6401,
+      "step": 13263
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 2.1447307514272493e-05,
+      "loss": 1.0981,
+      "step": 13264
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 2.1444735894666462e-05,
+      "loss": 0.8121,
+      "step": 13265
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 2.1442164275060432e-05,
+      "loss": 0.838,
+      "step": 13266
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 2.143959265545441e-05,
+      "loss": 0.8839,
+      "step": 13267
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 2.1437021035848378e-05,
+      "loss": 0.4173,
+      "step": 13268
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 2.143444941624235e-05,
+      "loss": 0.4744,
+      "step": 13269
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 2.1431877796636324e-05,
+      "loss": 0.8096,
+      "step": 13270
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 2.1429306177030294e-05,
+      "loss": 0.537,
+      "step": 13271
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 2.1426734557424267e-05,
+      "loss": 0.7669,
+      "step": 13272
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 2.142416293781824e-05,
+      "loss": 0.6226,
+      "step": 13273
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 2.142159131821221e-05,
+      "loss": 0.6736,
+      "step": 13274
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 2.1419019698606183e-05,
+      "loss": 0.536,
+      "step": 13275
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 2.1416448079000156e-05,
+      "loss": 0.535,
+      "step": 13276
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 2.141387645939413e-05,
+      "loss": 0.8133,
+      "step": 13277
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 2.14113048397881e-05,
+      "loss": 0.611,
+      "step": 13278
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 2.1408733220182072e-05,
+      "loss": 0.6536,
+      "step": 13279
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 2.1406161600576045e-05,
+      "loss": 0.4695,
+      "step": 13280
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 2.1403589980970015e-05,
+      "loss": 1.1679,
+      "step": 13281
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 2.1401018361363988e-05,
+      "loss": 0.7858,
+      "step": 13282
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 2.139844674175796e-05,
+      "loss": 0.635,
+      "step": 13283
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 2.139587512215193e-05,
+      "loss": 0.6527,
+      "step": 13284
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 2.1393303502545907e-05,
+      "loss": 0.582,
+      "step": 13285
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 2.1390731882939876e-05,
+      "loss": 0.3856,
+      "step": 13286
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 2.138816026333385e-05,
+      "loss": 0.9354,
+      "step": 13287
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 2.1385588643727823e-05,
+      "loss": 0.5667,
+      "step": 13288
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 2.1383017024121792e-05,
+      "loss": 0.6996,
+      "step": 13289
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 2.1380445404515765e-05,
+      "loss": 0.6043,
+      "step": 13290
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 2.137787378490974e-05,
+      "loss": 0.814,
+      "step": 13291
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 2.1375302165303708e-05,
+      "loss": 0.9288,
+      "step": 13292
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 2.137273054569768e-05,
+      "loss": 0.5988,
+      "step": 13293
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 2.1370158926091654e-05,
+      "loss": 0.7912,
+      "step": 13294
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 2.1367587306485627e-05,
+      "loss": 0.6464,
+      "step": 13295
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 2.1365015686879597e-05,
+      "loss": 0.7877,
+      "step": 13296
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 2.136244406727357e-05,
+      "loss": 0.8113,
+      "step": 13297
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 2.1359872447667543e-05,
+      "loss": 0.7636,
+      "step": 13298
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 2.1357300828061513e-05,
+      "loss": 0.6659,
+      "step": 13299
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 2.1354729208455486e-05,
+      "loss": 0.5669,
+      "step": 13300
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 2.135215758884946e-05,
+      "loss": 0.762,
+      "step": 13301
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 2.134958596924343e-05,
+      "loss": 0.625,
+      "step": 13302
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 2.1347014349637405e-05,
+      "loss": 0.838,
+      "step": 13303
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 2.1344442730031375e-05,
+      "loss": 0.6204,
+      "step": 13304
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 2.1341871110425345e-05,
+      "loss": 0.8283,
+      "step": 13305
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 2.133929949081932e-05,
+      "loss": 0.5994,
+      "step": 13306
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 2.133672787121329e-05,
+      "loss": 0.6898,
+      "step": 13307
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 2.1334156251607264e-05,
+      "loss": 0.8434,
+      "step": 13308
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 2.1331584632001237e-05,
+      "loss": 0.3621,
+      "step": 13309
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 2.1329013012395207e-05,
+      "loss": 0.5137,
+      "step": 13310
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 2.132644139278918e-05,
+      "loss": 0.4556,
+      "step": 13311
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 2.1323869773183153e-05,
+      "loss": 0.7226,
+      "step": 13312
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 2.1321298153577122e-05,
+      "loss": 0.5793,
+      "step": 13313
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 2.1318726533971095e-05,
+      "loss": 0.7187,
+      "step": 13314
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 2.131615491436507e-05,
+      "loss": 0.79,
+      "step": 13315
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 2.131358329475904e-05,
+      "loss": 0.5971,
+      "step": 13316
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 2.131101167515301e-05,
+      "loss": 0.7182,
+      "step": 13317
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 2.1308440055546984e-05,
+      "loss": 1.041,
+      "step": 13318
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 2.1305868435940957e-05,
+      "loss": 0.7273,
+      "step": 13319
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 2.1303296816334927e-05,
+      "loss": 0.6984,
+      "step": 13320
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 2.13007251967289e-05,
+      "loss": 0.7996,
+      "step": 13321
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 2.1298153577122873e-05,
+      "loss": 0.735,
+      "step": 13322
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 2.1295581957516843e-05,
+      "loss": 1.034,
+      "step": 13323
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 2.129301033791082e-05,
+      "loss": 0.5751,
+      "step": 13324
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 2.129043871830479e-05,
+      "loss": 0.6846,
+      "step": 13325
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 2.1287867098698762e-05,
+      "loss": 0.6006,
+      "step": 13326
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 2.1285295479092735e-05,
+      "loss": 0.6705,
+      "step": 13327
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 2.1282723859486705e-05,
+      "loss": 0.7499,
+      "step": 13328
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 2.1280152239880678e-05,
+      "loss": 0.8234,
+      "step": 13329
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 2.127758062027465e-05,
+      "loss": 0.6461,
+      "step": 13330
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 2.127500900066862e-05,
+      "loss": 0.6775,
+      "step": 13331
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 2.1272437381062594e-05,
+      "loss": 0.3529,
+      "step": 13332
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 2.1269865761456567e-05,
+      "loss": 0.6401,
+      "step": 13333
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 2.126729414185054e-05,
+      "loss": 0.6054,
+      "step": 13334
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 2.126472252224451e-05,
+      "loss": 0.7937,
+      "step": 13335
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 2.1262150902638483e-05,
+      "loss": 0.8401,
+      "step": 13336
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 2.1259579283032456e-05,
+      "loss": 0.4605,
+      "step": 13337
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 2.1257007663426425e-05,
+      "loss": 0.649,
+      "step": 13338
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 2.12544360438204e-05,
+      "loss": 0.6435,
+      "step": 13339
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 2.125186442421437e-05,
+      "loss": 0.6973,
+      "step": 13340
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 2.124929280460834e-05,
+      "loss": 1.1395,
+      "step": 13341
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 2.1246721185002318e-05,
+      "loss": 0.7984,
+      "step": 13342
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 2.1244149565396287e-05,
+      "loss": 0.7589,
+      "step": 13343
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 2.1241577945790257e-05,
+      "loss": 0.6367,
+      "step": 13344
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 2.1239006326184234e-05,
+      "loss": 0.6189,
+      "step": 13345
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 2.1236434706578203e-05,
+      "loss": 0.9831,
+      "step": 13346
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 2.1233863086972176e-05,
+      "loss": 0.5341,
+      "step": 13347
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 2.123129146736615e-05,
+      "loss": 0.8243,
+      "step": 13348
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 2.122871984776012e-05,
+      "loss": 0.476,
+      "step": 13349
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 2.1226148228154092e-05,
+      "loss": 0.9664,
+      "step": 13350
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 2.1223576608548065e-05,
+      "loss": 0.5176,
+      "step": 13351
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 2.1221004988942035e-05,
+      "loss": 0.9266,
+      "step": 13352
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 2.1218433369336008e-05,
+      "loss": 0.3575,
+      "step": 13353
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 2.121586174972998e-05,
+      "loss": 1.1628,
+      "step": 13354
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 2.1213290130123954e-05,
+      "loss": 0.673,
+      "step": 13355
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 2.1210718510517924e-05,
+      "loss": 0.7484,
+      "step": 13356
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 2.1208146890911897e-05,
+      "loss": 0.7826,
+      "step": 13357
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 2.120557527130587e-05,
+      "loss": 0.3692,
+      "step": 13358
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 2.120300365169984e-05,
+      "loss": 0.334,
+      "step": 13359
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 2.1200432032093813e-05,
+      "loss": 1.1044,
+      "step": 13360
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 2.1197860412487786e-05,
+      "loss": 0.2627,
+      "step": 13361
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 2.1195288792881756e-05,
+      "loss": 0.4272,
+      "step": 13362
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 2.1192717173275732e-05,
+      "loss": 0.6583,
+      "step": 13363
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 2.11901455536697e-05,
+      "loss": 0.6772,
+      "step": 13364
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 2.1187573934063675e-05,
+      "loss": 0.6844,
+      "step": 13365
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 2.1185002314457648e-05,
+      "loss": 0.5749,
+      "step": 13366
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 2.1182430694851618e-05,
+      "loss": 0.5971,
+      "step": 13367
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 2.117985907524559e-05,
+      "loss": 0.8876,
+      "step": 13368
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 2.1177287455639564e-05,
+      "loss": 0.5903,
+      "step": 13369
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 2.1174715836033533e-05,
+      "loss": 0.5421,
+      "step": 13370
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 2.1172144216427506e-05,
+      "loss": 0.4376,
+      "step": 13371
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 2.116957259682148e-05,
+      "loss": 0.6801,
+      "step": 13372
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 2.1167000977215453e-05,
+      "loss": 1.0801,
+      "step": 13373
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 2.1164429357609422e-05,
+      "loss": 0.784,
+      "step": 13374
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 2.1161857738003395e-05,
+      "loss": 0.6785,
+      "step": 13375
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 2.115928611839737e-05,
+      "loss": 0.9316,
+      "step": 13376
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 2.1156714498791338e-05,
+      "loss": 0.7479,
+      "step": 13377
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 2.115414287918531e-05,
+      "loss": 0.8655,
+      "step": 13378
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 2.1151571259579284e-05,
+      "loss": 0.5261,
+      "step": 13379
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 2.1148999639973254e-05,
+      "loss": 0.8233,
+      "step": 13380
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 2.114642802036723e-05,
+      "loss": 0.5679,
+      "step": 13381
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 2.11438564007612e-05,
+      "loss": 0.4447,
+      "step": 13382
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 2.114128478115517e-05,
+      "loss": 0.8239,
+      "step": 13383
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 2.1138713161549146e-05,
+      "loss": 0.5183,
+      "step": 13384
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 2.1136141541943116e-05,
+      "loss": 0.7336,
+      "step": 13385
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 2.113356992233709e-05,
+      "loss": 0.5438,
+      "step": 13386
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 2.1130998302731062e-05,
+      "loss": 0.6169,
+      "step": 13387
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 2.1128426683125032e-05,
+      "loss": 0.65,
+      "step": 13388
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 2.1125855063519005e-05,
+      "loss": 0.9493,
+      "step": 13389
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 2.1123283443912978e-05,
+      "loss": 0.4857,
+      "step": 13390
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 2.1120711824306948e-05,
+      "loss": 0.5029,
+      "step": 13391
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 2.1118140204700924e-05,
+      "loss": 0.7276,
+      "step": 13392
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 2.1115568585094894e-05,
+      "loss": 0.9505,
+      "step": 13393
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 2.1112996965488867e-05,
+      "loss": 1.0568,
+      "step": 13394
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 2.111042534588284e-05,
+      "loss": 0.9693,
+      "step": 13395
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 2.110785372627681e-05,
+      "loss": 0.5305,
+      "step": 13396
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 2.1105282106670783e-05,
+      "loss": 0.8126,
+      "step": 13397
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 2.1102710487064756e-05,
+      "loss": 0.5112,
+      "step": 13398
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 2.1100138867458725e-05,
+      "loss": 0.5082,
+      "step": 13399
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 2.10975672478527e-05,
+      "loss": 0.4596,
+      "step": 13400
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 2.109499562824667e-05,
+      "loss": 0.3958,
+      "step": 13401
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 2.1092424008640645e-05,
+      "loss": 0.5701,
+      "step": 13402
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 2.1089852389034614e-05,
+      "loss": 0.7014,
+      "step": 13403
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 2.1087280769428587e-05,
+      "loss": 0.73,
+      "step": 13404
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 2.108470914982256e-05,
+      "loss": 0.6613,
+      "step": 13405
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 2.108213753021653e-05,
+      "loss": 0.7925,
+      "step": 13406
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 2.1079565910610503e-05,
+      "loss": 0.8151,
+      "step": 13407
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 2.1076994291004476e-05,
+      "loss": 0.7025,
+      "step": 13408
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 2.1074422671398446e-05,
+      "loss": 0.4395,
+      "step": 13409
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 2.1071851051792422e-05,
+      "loss": 0.8704,
+      "step": 13410
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 2.1069279432186392e-05,
+      "loss": 0.6734,
+      "step": 13411
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 2.1066707812580365e-05,
+      "loss": 0.8482,
+      "step": 13412
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 2.1064136192974338e-05,
+      "loss": 0.6241,
+      "step": 13413
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 2.1061564573368308e-05,
+      "loss": 0.7616,
+      "step": 13414
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 2.105899295376228e-05,
+      "loss": 0.5662,
+      "step": 13415
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 2.1056421334156254e-05,
+      "loss": 0.7599,
+      "step": 13416
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 2.1053849714550224e-05,
+      "loss": 0.5661,
+      "step": 13417
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 2.1051278094944197e-05,
+      "loss": 0.8674,
+      "step": 13418
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 2.104870647533817e-05,
+      "loss": 0.6448,
+      "step": 13419
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 2.1046134855732143e-05,
+      "loss": 0.6193,
+      "step": 13420
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 2.1043563236126113e-05,
+      "loss": 0.8819,
+      "step": 13421
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 2.1040991616520086e-05,
+      "loss": 0.6321,
+      "step": 13422
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 2.103841999691406e-05,
+      "loss": 0.5128,
+      "step": 13423
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 2.103584837730803e-05,
+      "loss": 1.2569,
+      "step": 13424
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 2.1033276757702e-05,
+      "loss": 0.4541,
+      "step": 13425
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 2.1030705138095975e-05,
+      "loss": 0.8921,
+      "step": 13426
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 2.1028133518489944e-05,
+      "loss": 0.5992,
+      "step": 13427
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 2.102556189888392e-05,
+      "loss": 0.4189,
+      "step": 13428
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 2.102299027927789e-05,
+      "loss": 1.0421,
+      "step": 13429
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 2.102041865967186e-05,
+      "loss": 0.6366,
+      "step": 13430
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 2.1017847040065837e-05,
+      "loss": 0.8267,
+      "step": 13431
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 2.1015275420459806e-05,
+      "loss": 0.5772,
+      "step": 13432
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 2.101270380085378e-05,
+      "loss": 0.41,
+      "step": 13433
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 2.1010132181247752e-05,
+      "loss": 0.7819,
+      "step": 13434
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 2.1007560561641722e-05,
+      "loss": 0.5914,
+      "step": 13435
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 2.1004988942035695e-05,
+      "loss": 0.5135,
+      "step": 13436
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 2.1002417322429668e-05,
+      "loss": 0.6414,
+      "step": 13437
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 2.0999845702823638e-05,
+      "loss": 0.6617,
+      "step": 13438
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 2.099727408321761e-05,
+      "loss": 0.8801,
+      "step": 13439
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 2.0994702463611584e-05,
+      "loss": 0.5528,
+      "step": 13440
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 2.0992130844005557e-05,
+      "loss": 0.7376,
+      "step": 13441
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 2.0989559224399527e-05,
+      "loss": 0.6921,
+      "step": 13442
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 2.09869876047935e-05,
+      "loss": 0.4983,
+      "step": 13443
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 2.0984415985187473e-05,
+      "loss": 0.5771,
+      "step": 13444
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 2.0981844365581443e-05,
+      "loss": 0.9513,
+      "step": 13445
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 2.0979272745975416e-05,
+      "loss": 0.7997,
+      "step": 13446
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 2.097670112636939e-05,
+      "loss": 0.8246,
+      "step": 13447
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 2.097412950676336e-05,
+      "loss": 0.5788,
+      "step": 13448
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 2.0971557887157335e-05,
+      "loss": 0.6113,
+      "step": 13449
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 2.0968986267551305e-05,
+      "loss": 0.881,
+      "step": 13450
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 2.0966414647945278e-05,
+      "loss": 0.6052,
+      "step": 13451
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 2.096384302833925e-05,
+      "loss": 0.5892,
+      "step": 13452
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 2.096127140873322e-05,
+      "loss": 0.5361,
+      "step": 13453
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 2.0958699789127194e-05,
+      "loss": 0.6971,
+      "step": 13454
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 2.0956128169521167e-05,
+      "loss": 0.7292,
+      "step": 13455
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 2.0953556549915136e-05,
+      "loss": 0.7295,
+      "step": 13456
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 2.095098493030911e-05,
+      "loss": 0.5218,
+      "step": 13457
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 2.0948413310703082e-05,
+      "loss": 0.6082,
+      "step": 13458
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 2.0945841691097056e-05,
+      "loss": 0.8597,
+      "step": 13459
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 2.0943270071491025e-05,
+      "loss": 0.8602,
+      "step": 13460
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 2.0940698451885e-05,
+      "loss": 0.5094,
+      "step": 13461
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 2.093812683227897e-05,
+      "loss": 0.6384,
+      "step": 13462
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 2.093555521267294e-05,
+      "loss": 0.5709,
+      "step": 13463
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 2.0932983593066914e-05,
+      "loss": 0.842,
+      "step": 13464
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 2.0930411973460887e-05,
+      "loss": 0.4577,
+      "step": 13465
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 2.0927840353854857e-05,
+      "loss": 1.0343,
+      "step": 13466
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 2.0925268734248833e-05,
+      "loss": 0.8211,
+      "step": 13467
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 2.0922697114642803e-05,
+      "loss": 1.0358,
+      "step": 13468
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 2.0920125495036773e-05,
+      "loss": 0.4254,
+      "step": 13469
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 2.091755387543075e-05,
+      "loss": 0.4787,
+      "step": 13470
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 2.091498225582472e-05,
+      "loss": 0.7014,
+      "step": 13471
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 2.0912410636218692e-05,
+      "loss": 0.7623,
+      "step": 13472
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 2.0909839016612665e-05,
+      "loss": 0.5152,
+      "step": 13473
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 2.0907267397006635e-05,
+      "loss": 0.6822,
+      "step": 13474
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 2.0904695777400608e-05,
+      "loss": 0.5801,
+      "step": 13475
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 2.090212415779458e-05,
+      "loss": 0.7522,
+      "step": 13476
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 2.089955253818855e-05,
+      "loss": 0.6623,
+      "step": 13477
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 2.0896980918582524e-05,
+      "loss": 0.7369,
+      "step": 13478
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 2.0894409298976497e-05,
+      "loss": 0.7402,
+      "step": 13479
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 2.089183767937047e-05,
+      "loss": 0.5091,
+      "step": 13480
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 2.088926605976444e-05,
+      "loss": 0.9855,
+      "step": 13481
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 2.0886694440158413e-05,
+      "loss": 0.6152,
+      "step": 13482
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 2.0884122820552386e-05,
+      "loss": 0.6886,
+      "step": 13483
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 2.0881551200946355e-05,
+      "loss": 0.8152,
+      "step": 13484
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 2.087897958134033e-05,
+      "loss": 0.6766,
+      "step": 13485
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 2.08764079617343e-05,
+      "loss": 0.7569,
+      "step": 13486
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 2.087383634212827e-05,
+      "loss": 0.6141,
+      "step": 13487
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 2.0871264722522248e-05,
+      "loss": 0.4915,
+      "step": 13488
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 2.0868693102916217e-05,
+      "loss": 0.9571,
+      "step": 13489
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 2.086612148331019e-05,
+      "loss": 0.7435,
+      "step": 13490
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 2.0863549863704163e-05,
+      "loss": 0.5167,
+      "step": 13491
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 2.0860978244098133e-05,
+      "loss": 0.4343,
+      "step": 13492
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 2.0858406624492106e-05,
+      "loss": 0.5833,
+      "step": 13493
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 2.085583500488608e-05,
+      "loss": 0.81,
+      "step": 13494
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 2.085326338528005e-05,
+      "loss": 0.5237,
+      "step": 13495
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 2.0850691765674022e-05,
+      "loss": 0.5806,
+      "step": 13496
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 2.0848120146067995e-05,
+      "loss": 0.4667,
+      "step": 13497
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 2.0845548526461968e-05,
+      "loss": 0.4657,
+      "step": 13498
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 2.0842976906855938e-05,
+      "loss": 0.7203,
+      "step": 13499
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 2.084040528724991e-05,
+      "loss": 0.6666,
+      "step": 13500
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 2.0837833667643884e-05,
+      "loss": 0.667,
+      "step": 13501
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 2.0835262048037854e-05,
+      "loss": 0.5371,
+      "step": 13502
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 2.0832690428431827e-05,
+      "loss": 0.4465,
+      "step": 13503
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 2.08301188088258e-05,
+      "loss": 0.4751,
+      "step": 13504
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 2.082754718921977e-05,
+      "loss": 0.6387,
+      "step": 13505
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 2.0824975569613746e-05,
+      "loss": 0.726,
+      "step": 13506
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 2.0822403950007716e-05,
+      "loss": 0.4948,
+      "step": 13507
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 2.0819832330401685e-05,
+      "loss": 0.4364,
+      "step": 13508
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 2.0817260710795662e-05,
+      "loss": 0.5171,
+      "step": 13509
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 2.081468909118963e-05,
+      "loss": 0.8982,
+      "step": 13510
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 2.0812117471583605e-05,
+      "loss": 0.7391,
+      "step": 13511
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 2.0809545851977578e-05,
+      "loss": 0.6134,
+      "step": 13512
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 2.0806974232371547e-05,
+      "loss": 0.8366,
+      "step": 13513
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 2.080440261276552e-05,
+      "loss": 0.8552,
+      "step": 13514
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 2.0801830993159493e-05,
+      "loss": 0.263,
+      "step": 13515
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 2.0799259373553463e-05,
+      "loss": 0.7869,
+      "step": 13516
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 2.0796687753947436e-05,
+      "loss": 0.8387,
+      "step": 13517
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 2.079411613434141e-05,
+      "loss": 0.6522,
+      "step": 13518
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 2.0791544514735382e-05,
+      "loss": 0.9653,
+      "step": 13519
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 2.0788972895129352e-05,
+      "loss": 0.7292,
+      "step": 13520
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 2.0786401275523325e-05,
+      "loss": 0.5483,
+      "step": 13521
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 2.0783829655917298e-05,
+      "loss": 0.6831,
+      "step": 13522
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 2.0781258036311268e-05,
+      "loss": 0.7814,
+      "step": 13523
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 2.077868641670524e-05,
+      "loss": 0.6088,
+      "step": 13524
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 2.0776114797099214e-05,
+      "loss": 0.5505,
+      "step": 13525
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 2.0773543177493184e-05,
+      "loss": 0.6704,
+      "step": 13526
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 2.077097155788716e-05,
+      "loss": 0.4905,
+      "step": 13527
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 2.076839993828113e-05,
+      "loss": 0.6274,
+      "step": 13528
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 2.0765828318675103e-05,
+      "loss": 0.5976,
+      "step": 13529
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 2.0763256699069076e-05,
+      "loss": 0.4746,
+      "step": 13530
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 2.0760685079463046e-05,
+      "loss": 0.5511,
+      "step": 13531
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 2.075811345985702e-05,
+      "loss": 0.5055,
+      "step": 13532
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 2.0755541840250992e-05,
+      "loss": 0.7511,
+      "step": 13533
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 2.075297022064496e-05,
+      "loss": 0.6971,
+      "step": 13534
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 2.0750398601038935e-05,
+      "loss": 0.6383,
+      "step": 13535
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 2.0747826981432908e-05,
+      "loss": 0.7086,
+      "step": 13536
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 2.074525536182688e-05,
+      "loss": 0.7454,
+      "step": 13537
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 2.074268374222085e-05,
+      "loss": 0.5469,
+      "step": 13538
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 2.0740112122614824e-05,
+      "loss": 0.6935,
+      "step": 13539
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 2.0737540503008797e-05,
+      "loss": 0.7787,
+      "step": 13540
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 2.073496888340277e-05,
+      "loss": 0.803,
+      "step": 13541
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 2.073239726379674e-05,
+      "loss": 0.5258,
+      "step": 13542
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 2.0729825644190712e-05,
+      "loss": 0.647,
+      "step": 13543
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 2.0727254024584685e-05,
+      "loss": 0.5732,
+      "step": 13544
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 2.072468240497866e-05,
+      "loss": 0.9277,
+      "step": 13545
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 2.0722110785372628e-05,
+      "loss": 0.9181,
+      "step": 13546
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 2.07195391657666e-05,
+      "loss": 0.256,
+      "step": 13547
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 2.0716967546160574e-05,
+      "loss": 0.5902,
+      "step": 13548
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 2.0714395926554544e-05,
+      "loss": 0.7239,
+      "step": 13549
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 2.0711824306948517e-05,
+      "loss": 0.8459,
+      "step": 13550
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 2.070925268734249e-05,
+      "loss": 0.704,
+      "step": 13551
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 2.070668106773646e-05,
+      "loss": 1.0812,
+      "step": 13552
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 2.0704109448130436e-05,
+      "loss": 0.546,
+      "step": 13553
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 2.0701537828524406e-05,
+      "loss": 0.8572,
+      "step": 13554
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 2.0698966208918376e-05,
+      "loss": 0.7537,
+      "step": 13555
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 2.0696394589312352e-05,
+      "loss": 0.6699,
+      "step": 13556
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 2.0693822969706322e-05,
+      "loss": 0.7609,
+      "step": 13557
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 2.0691251350100295e-05,
+      "loss": 0.6915,
+      "step": 13558
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 2.0688679730494268e-05,
+      "loss": 0.5241,
+      "step": 13559
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 2.0686108110888238e-05,
+      "loss": 0.5687,
+      "step": 13560
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 2.068353649128221e-05,
+      "loss": 0.71,
+      "step": 13561
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 2.0680964871676184e-05,
+      "loss": 0.8727,
+      "step": 13562
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 2.0678393252070154e-05,
+      "loss": 0.8532,
+      "step": 13563
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 2.0675821632464127e-05,
+      "loss": 1.1765,
+      "step": 13564
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 2.06732500128581e-05,
+      "loss": 0.5083,
+      "step": 13565
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 2.0670678393252073e-05,
+      "loss": 0.8416,
+      "step": 13566
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 2.0668106773646042e-05,
+      "loss": 0.5872,
+      "step": 13567
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 2.0665535154040016e-05,
+      "loss": 0.5364,
+      "step": 13568
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 2.066296353443399e-05,
+      "loss": 0.5616,
+      "step": 13569
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 2.0660391914827958e-05,
+      "loss": 0.8482,
+      "step": 13570
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 2.065782029522193e-05,
+      "loss": 0.6652,
+      "step": 13571
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 2.0655248675615904e-05,
+      "loss": 0.7395,
+      "step": 13572
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 2.0652677056009874e-05,
+      "loss": 0.6051,
+      "step": 13573
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 2.065010543640385e-05,
+      "loss": 0.7396,
+      "step": 13574
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 2.064753381679782e-05,
+      "loss": 0.8231,
+      "step": 13575
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 2.0644962197191793e-05,
+      "loss": 0.5635,
+      "step": 13576
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 2.0642390577585766e-05,
+      "loss": 0.4566,
+      "step": 13577
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 2.0639818957979736e-05,
+      "loss": 0.7343,
+      "step": 13578
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 2.063724733837371e-05,
+      "loss": 0.5289,
+      "step": 13579
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 2.0634675718767682e-05,
+      "loss": 0.9494,
+      "step": 13580
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 2.0632104099161652e-05,
+      "loss": 0.8856,
+      "step": 13581
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 2.0629532479555625e-05,
+      "loss": 0.5192,
+      "step": 13582
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 2.0626960859949598e-05,
+      "loss": 0.7965,
+      "step": 13583
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 2.062438924034357e-05,
+      "loss": 0.8623,
+      "step": 13584
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 2.062181762073754e-05,
+      "loss": 0.7792,
+      "step": 13585
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 2.0619246001131514e-05,
+      "loss": 0.5858,
+      "step": 13586
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 2.0616674381525487e-05,
+      "loss": 1.2192,
+      "step": 13587
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 2.0614102761919457e-05,
+      "loss": 0.7155,
+      "step": 13588
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 2.061153114231343e-05,
+      "loss": 0.4297,
+      "step": 13589
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 2.0608959522707403e-05,
+      "loss": 0.7208,
+      "step": 13590
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 2.0606387903101373e-05,
+      "loss": 0.8198,
+      "step": 13591
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 2.060381628349535e-05,
+      "loss": 0.8423,
+      "step": 13592
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 2.060124466388932e-05,
+      "loss": 1.0674,
+      "step": 13593
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 2.059867304428329e-05,
+      "loss": 0.823,
+      "step": 13594
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 2.0596101424677265e-05,
+      "loss": 0.5142,
+      "step": 13595
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 2.0593529805071234e-05,
+      "loss": 0.8545,
+      "step": 13596
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 2.0590958185465208e-05,
+      "loss": 0.8179,
+      "step": 13597
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 2.058838656585918e-05,
+      "loss": 1.0235,
+      "step": 13598
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 2.058581494625315e-05,
+      "loss": 0.6086,
+      "step": 13599
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 2.0583243326647123e-05,
+      "loss": 0.9566,
+      "step": 13600
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 2.0580671707041096e-05,
+      "loss": 0.8126,
+      "step": 13601
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 2.0578100087435066e-05,
+      "loss": 0.8651,
+      "step": 13602
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 2.057552846782904e-05,
+      "loss": 0.8444,
+      "step": 13603
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 2.0572956848223012e-05,
+      "loss": 0.9137,
+      "step": 13604
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 2.0570385228616985e-05,
+      "loss": 0.5811,
+      "step": 13605
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 2.0567813609010955e-05,
+      "loss": 0.567,
+      "step": 13606
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 2.0565241989404928e-05,
+      "loss": 0.6843,
+      "step": 13607
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 2.05626703697989e-05,
+      "loss": 0.7174,
+      "step": 13608
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 2.056009875019287e-05,
+      "loss": 0.5302,
+      "step": 13609
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 2.0557527130586844e-05,
+      "loss": 0.5356,
+      "step": 13610
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 2.0554955510980817e-05,
+      "loss": 0.6505,
+      "step": 13611
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 2.0552383891374787e-05,
+      "loss": 0.9283,
+      "step": 13612
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 2.0549812271768763e-05,
+      "loss": 0.716,
+      "step": 13613
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 2.0547240652162733e-05,
+      "loss": 0.6814,
+      "step": 13614
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 2.0544669032556706e-05,
+      "loss": 0.4733,
+      "step": 13615
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 2.054209741295068e-05,
+      "loss": 0.4168,
+      "step": 13616
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 2.053952579334465e-05,
+      "loss": 0.4437,
+      "step": 13617
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 2.0536954173738622e-05,
+      "loss": 0.4706,
+      "step": 13618
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 2.0534382554132595e-05,
+      "loss": 0.8409,
+      "step": 13619
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 2.0531810934526565e-05,
+      "loss": 0.8491,
+      "step": 13620
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 2.0529239314920538e-05,
+      "loss": 0.7613,
+      "step": 13621
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 2.052666769531451e-05,
+      "loss": 0.8983,
+      "step": 13622
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 2.0524096075708484e-05,
+      "loss": 0.8854,
+      "step": 13623
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 2.0521524456102453e-05,
+      "loss": 0.8273,
+      "step": 13624
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 2.0518952836496427e-05,
+      "loss": 0.7543,
+      "step": 13625
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 2.05163812168904e-05,
+      "loss": 0.4667,
+      "step": 13626
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 2.051380959728437e-05,
+      "loss": 0.6825,
+      "step": 13627
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 2.0511237977678342e-05,
+      "loss": 0.7574,
+      "step": 13628
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 2.0508666358072315e-05,
+      "loss": 0.9425,
+      "step": 13629
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 2.0506094738466285e-05,
+      "loss": 0.7506,
+      "step": 13630
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 2.050352311886026e-05,
+      "loss": 0.7102,
+      "step": 13631
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 2.050095149925423e-05,
+      "loss": 0.9011,
+      "step": 13632
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 2.04983798796482e-05,
+      "loss": 0.6922,
+      "step": 13633
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 2.0495808260042177e-05,
+      "loss": 0.8072,
+      "step": 13634
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 2.0493236640436147e-05,
+      "loss": 0.7221,
+      "step": 13635
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 2.049066502083012e-05,
+      "loss": 0.5899,
+      "step": 13636
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 2.0488093401224093e-05,
+      "loss": 1.1901,
+      "step": 13637
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 2.0485521781618063e-05,
+      "loss": 0.8042,
+      "step": 13638
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 2.0482950162012036e-05,
+      "loss": 0.4896,
+      "step": 13639
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 2.048037854240601e-05,
+      "loss": 0.687,
+      "step": 13640
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 2.047780692279998e-05,
+      "loss": 0.5419,
+      "step": 13641
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 2.0475235303193952e-05,
+      "loss": 0.7447,
+      "step": 13642
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 2.0472663683587925e-05,
+      "loss": 0.6778,
+      "step": 13643
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 2.0470092063981898e-05,
+      "loss": 0.7689,
+      "step": 13644
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 2.0467520444375868e-05,
+      "loss": 0.817,
+      "step": 13645
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 2.046494882476984e-05,
+      "loss": 0.8966,
+      "step": 13646
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 2.0462377205163814e-05,
+      "loss": 0.4525,
+      "step": 13647
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 2.0459805585557783e-05,
+      "loss": 0.492,
+      "step": 13648
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 2.0457233965951757e-05,
+      "loss": 0.3208,
+      "step": 13649
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 2.045466234634573e-05,
+      "loss": 0.5134,
+      "step": 13650
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 2.04520907267397e-05,
+      "loss": 0.6788,
+      "step": 13651
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 2.0449519107133676e-05,
+      "loss": 0.7853,
+      "step": 13652
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 2.0446947487527645e-05,
+      "loss": 0.943,
+      "step": 13653
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 2.044437586792162e-05,
+      "loss": 0.7033,
+      "step": 13654
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 2.044180424831559e-05,
+      "loss": 0.8082,
+      "step": 13655
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 2.043923262870956e-05,
+      "loss": 0.7032,
+      "step": 13656
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 2.0436661009103534e-05,
+      "loss": 0.8084,
+      "step": 13657
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 2.0434089389497507e-05,
+      "loss": 0.5889,
+      "step": 13658
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 2.0431517769891477e-05,
+      "loss": 0.6405,
+      "step": 13659
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 2.042894615028545e-05,
+      "loss": 0.8277,
+      "step": 13660
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 2.0426374530679423e-05,
+      "loss": 0.8235,
+      "step": 13661
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 2.0423802911073396e-05,
+      "loss": 0.6421,
+      "step": 13662
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 2.0421231291467366e-05,
+      "loss": 0.6818,
+      "step": 13663
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 2.041865967186134e-05,
+      "loss": 0.7705,
+      "step": 13664
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 2.0416088052255312e-05,
+      "loss": 0.6979,
+      "step": 13665
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 2.0413516432649282e-05,
+      "loss": 1.0176,
+      "step": 13666
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 2.0410944813043255e-05,
+      "loss": 0.4669,
+      "step": 13667
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 2.0408373193437228e-05,
+      "loss": 0.9159,
+      "step": 13668
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 2.0405801573831198e-05,
+      "loss": 1.0769,
+      "step": 13669
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 2.0403229954225174e-05,
+      "loss": 0.763,
+      "step": 13670
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 2.0400658334619144e-05,
+      "loss": 0.7049,
+      "step": 13671
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 2.0398086715013114e-05,
+      "loss": 0.894,
+      "step": 13672
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 2.039551509540709e-05,
+      "loss": 0.6964,
+      "step": 13673
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 2.039294347580106e-05,
+      "loss": 0.8394,
+      "step": 13674
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 2.0390371856195033e-05,
+      "loss": 0.8576,
+      "step": 13675
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 2.0387800236589006e-05,
+      "loss": 0.8198,
+      "step": 13676
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 2.0385228616982976e-05,
+      "loss": 0.8406,
+      "step": 13677
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 2.038265699737695e-05,
+      "loss": 0.714,
+      "step": 13678
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 2.038008537777092e-05,
+      "loss": 0.5996,
+      "step": 13679
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 2.037751375816489e-05,
+      "loss": 0.4923,
+      "step": 13680
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 2.0374942138558864e-05,
+      "loss": 0.7083,
+      "step": 13681
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 2.0372370518952837e-05,
+      "loss": 0.591,
+      "step": 13682
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 2.036979889934681e-05,
+      "loss": 0.4396,
+      "step": 13683
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 2.036722727974078e-05,
+      "loss": 0.9017,
+      "step": 13684
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 2.0364655660134753e-05,
+      "loss": 0.8445,
+      "step": 13685
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 2.0362084040528726e-05,
+      "loss": 0.5052,
+      "step": 13686
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 2.0359512420922696e-05,
+      "loss": 0.5715,
+      "step": 13687
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 2.035694080131667e-05,
+      "loss": 0.7573,
+      "step": 13688
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 2.0354369181710642e-05,
+      "loss": 0.773,
+      "step": 13689
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 2.0351797562104612e-05,
+      "loss": 0.7712,
+      "step": 13690
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 2.034922594249859e-05,
+      "loss": 0.6756,
+      "step": 13691
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 2.0346654322892558e-05,
+      "loss": 0.6943,
+      "step": 13692
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 2.034408270328653e-05,
+      "loss": 0.9653,
+      "step": 13693
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 2.0341511083680504e-05,
+      "loss": 0.6228,
+      "step": 13694
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 2.0338939464074474e-05,
+      "loss": 0.5024,
+      "step": 13695
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 2.0336367844468447e-05,
+      "loss": 0.6373,
+      "step": 13696
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 2.033379622486242e-05,
+      "loss": 0.6834,
+      "step": 13697
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 2.033122460525639e-05,
+      "loss": 0.85,
+      "step": 13698
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 2.0328652985650366e-05,
+      "loss": 0.6954,
+      "step": 13699
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 2.0326081366044336e-05,
+      "loss": 0.381,
+      "step": 13700
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 2.032350974643831e-05,
+      "loss": 0.7293,
+      "step": 13701
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 2.0320938126832282e-05,
+      "loss": 0.9416,
+      "step": 13702
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 2.0318366507226252e-05,
+      "loss": 0.678,
+      "step": 13703
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 2.0315794887620225e-05,
+      "loss": 0.5292,
+      "step": 13704
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 2.0313223268014198e-05,
+      "loss": 0.7308,
+      "step": 13705
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 2.0310651648408168e-05,
+      "loss": 0.6002,
+      "step": 13706
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 2.030808002880214e-05,
+      "loss": 0.5847,
+      "step": 13707
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 2.0305508409196114e-05,
+      "loss": 0.599,
+      "step": 13708
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 2.0302936789590087e-05,
+      "loss": 0.6813,
+      "step": 13709
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 2.0300365169984056e-05,
+      "loss": 0.6237,
+      "step": 13710
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 2.029779355037803e-05,
+      "loss": 0.6727,
+      "step": 13711
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 2.0295221930772003e-05,
+      "loss": 0.8616,
+      "step": 13712
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 2.0292650311165972e-05,
+      "loss": 0.5951,
+      "step": 13713
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 2.0290078691559945e-05,
+      "loss": 1.0834,
+      "step": 13714
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 2.028750707195392e-05,
+      "loss": 0.9024,
+      "step": 13715
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 2.0284935452347888e-05,
+      "loss": 0.6854,
+      "step": 13716
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 2.0282363832741865e-05,
+      "loss": 1.2268,
+      "step": 13717
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 2.0279792213135834e-05,
+      "loss": 0.69,
+      "step": 13718
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 2.0277220593529804e-05,
+      "loss": 0.8894,
+      "step": 13719
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 2.027464897392378e-05,
+      "loss": 1.0146,
+      "step": 13720
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 2.027207735431775e-05,
+      "loss": 0.6041,
+      "step": 13721
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 2.0269505734711723e-05,
+      "loss": 0.7417,
+      "step": 13722
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 2.0266934115105696e-05,
+      "loss": 0.7216,
+      "step": 13723
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 2.0264362495499666e-05,
+      "loss": 0.5921,
+      "step": 13724
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 2.026179087589364e-05,
+      "loss": 0.7071,
+      "step": 13725
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 2.0259219256287612e-05,
+      "loss": 0.4892,
+      "step": 13726
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 2.0256647636681582e-05,
+      "loss": 0.7289,
+      "step": 13727
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 2.0254076017075555e-05,
+      "loss": 0.8693,
+      "step": 13728
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 2.0251504397469528e-05,
+      "loss": 0.6173,
+      "step": 13729
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 2.02489327778635e-05,
+      "loss": 0.6832,
+      "step": 13730
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 2.024636115825747e-05,
+      "loss": 0.6628,
+      "step": 13731
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 2.0243789538651444e-05,
+      "loss": 0.7795,
+      "step": 13732
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 2.0241217919045417e-05,
+      "loss": 0.4659,
+      "step": 13733
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 2.0238646299439386e-05,
+      "loss": 0.4652,
+      "step": 13734
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 2.023607467983336e-05,
+      "loss": 0.4493,
+      "step": 13735
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 2.0233503060227333e-05,
+      "loss": 0.6728,
+      "step": 13736
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 2.0230931440621302e-05,
+      "loss": 0.9233,
+      "step": 13737
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 2.022835982101528e-05,
+      "loss": 0.8831,
+      "step": 13738
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 2.022578820140925e-05,
+      "loss": 0.8233,
+      "step": 13739
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 2.022321658180322e-05,
+      "loss": 0.5751,
+      "step": 13740
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 2.0220644962197195e-05,
+      "loss": 0.6524,
+      "step": 13741
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 2.0218073342591164e-05,
+      "loss": 0.6582,
+      "step": 13742
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 2.0215501722985137e-05,
+      "loss": 0.8724,
+      "step": 13743
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 2.021293010337911e-05,
+      "loss": 0.5781,
+      "step": 13744
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 2.021035848377308e-05,
+      "loss": 0.5554,
+      "step": 13745
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 2.0207786864167053e-05,
+      "loss": 0.7413,
+      "step": 13746
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 2.0205215244561026e-05,
+      "loss": 0.5989,
+      "step": 13747
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 2.0202643624955e-05,
+      "loss": 0.5966,
+      "step": 13748
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 2.020007200534897e-05,
+      "loss": 0.6945,
+      "step": 13749
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 2.0197500385742942e-05,
+      "loss": 0.5716,
+      "step": 13750
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 2.0194928766136915e-05,
+      "loss": 0.4949,
+      "step": 13751
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 2.0192357146530885e-05,
+      "loss": 0.6175,
+      "step": 13752
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 2.0189785526924858e-05,
+      "loss": 0.6661,
+      "step": 13753
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 2.018721390731883e-05,
+      "loss": 0.8084,
+      "step": 13754
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 2.01846422877128e-05,
+      "loss": 0.8399,
+      "step": 13755
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 2.0182070668106777e-05,
+      "loss": 0.6664,
+      "step": 13756
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 2.0179499048500747e-05,
+      "loss": 0.7635,
+      "step": 13757
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 2.0176927428894717e-05,
+      "loss": 0.705,
+      "step": 13758
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 2.0174355809288693e-05,
+      "loss": 0.5466,
+      "step": 13759
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 2.0171784189682663e-05,
+      "loss": 0.5515,
+      "step": 13760
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 2.0169212570076636e-05,
+      "loss": 0.3465,
+      "step": 13761
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 2.016664095047061e-05,
+      "loss": 0.6553,
+      "step": 13762
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 2.016406933086458e-05,
+      "loss": 0.6479,
+      "step": 13763
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 2.016149771125855e-05,
+      "loss": 0.3225,
+      "step": 13764
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 2.0158926091652525e-05,
+      "loss": 0.7612,
+      "step": 13765
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 2.0156354472046494e-05,
+      "loss": 0.6928,
+      "step": 13766
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 2.0153782852440467e-05,
+      "loss": 0.4781,
+      "step": 13767
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 2.015121123283444e-05,
+      "loss": 0.4165,
+      "step": 13768
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 2.0148639613228414e-05,
+      "loss": 0.3403,
+      "step": 13769
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 2.0146067993622383e-05,
+      "loss": 0.5221,
+      "step": 13770
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 2.0143496374016356e-05,
+      "loss": 0.643,
+      "step": 13771
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 2.014092475441033e-05,
+      "loss": 0.9026,
+      "step": 13772
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 2.01383531348043e-05,
+      "loss": 0.5924,
+      "step": 13773
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 2.0135781515198272e-05,
+      "loss": 0.5533,
+      "step": 13774
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 2.0133209895592245e-05,
+      "loss": 0.6717,
+      "step": 13775
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 2.0130638275986215e-05,
+      "loss": 0.7403,
+      "step": 13776
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 2.012806665638019e-05,
+      "loss": 0.9354,
+      "step": 13777
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 2.012549503677416e-05,
+      "loss": 0.678,
+      "step": 13778
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 2.0122923417168134e-05,
+      "loss": 0.6341,
+      "step": 13779
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 2.0120351797562107e-05,
+      "loss": 0.4504,
+      "step": 13780
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 2.0117780177956077e-05,
+      "loss": 0.863,
+      "step": 13781
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 2.011520855835005e-05,
+      "loss": 0.8015,
+      "step": 13782
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 2.0112636938744023e-05,
+      "loss": 0.3464,
+      "step": 13783
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 2.0110065319137993e-05,
+      "loss": 0.8645,
+      "step": 13784
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 2.0107493699531966e-05,
+      "loss": 0.7464,
+      "step": 13785
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 2.010492207992594e-05,
+      "loss": 0.3717,
+      "step": 13786
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 2.0102350460319912e-05,
+      "loss": 0.523,
+      "step": 13787
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 2.009977884071388e-05,
+      "loss": 0.7805,
+      "step": 13788
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 2.0097207221107855e-05,
+      "loss": 0.7149,
+      "step": 13789
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 2.0094635601501828e-05,
+      "loss": 0.8345,
+      "step": 13790
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 2.0092063981895797e-05,
+      "loss": 1.0122,
+      "step": 13791
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 2.008949236228977e-05,
+      "loss": 0.8587,
+      "step": 13792
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 2.0086920742683744e-05,
+      "loss": 0.713,
+      "step": 13793
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 2.0084349123077713e-05,
+      "loss": 0.6739,
+      "step": 13794
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 2.008177750347169e-05,
+      "loss": 1.0939,
+      "step": 13795
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 2.007920588386566e-05,
+      "loss": 0.8908,
+      "step": 13796
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 2.007663426425963e-05,
+      "loss": 1.1849,
+      "step": 13797
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 2.0074062644653606e-05,
+      "loss": 0.592,
+      "step": 13798
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 2.0071491025047575e-05,
+      "loss": 0.7746,
+      "step": 13799
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 2.006891940544155e-05,
+      "loss": 0.5673,
+      "step": 13800
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 2.006634778583552e-05,
+      "loss": 1.1384,
+      "step": 13801
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 2.006377616622949e-05,
+      "loss": 0.7695,
+      "step": 13802
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 2.0061204546623464e-05,
+      "loss": 0.8002,
+      "step": 13803
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 2.0058632927017437e-05,
+      "loss": 0.6792,
+      "step": 13804
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 2.0056061307411407e-05,
+      "loss": 0.8361,
+      "step": 13805
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 2.005348968780538e-05,
+      "loss": 0.7277,
+      "step": 13806
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 2.0050918068199353e-05,
+      "loss": 1.0326,
+      "step": 13807
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 2.0048346448593326e-05,
+      "loss": 0.713,
+      "step": 13808
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 2.0045774828987296e-05,
+      "loss": 0.605,
+      "step": 13809
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 2.004320320938127e-05,
+      "loss": 1.0105,
+      "step": 13810
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 2.0040631589775242e-05,
+      "loss": 0.3846,
+      "step": 13811
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 2.003805997016921e-05,
+      "loss": 0.614,
+      "step": 13812
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 2.0035488350563185e-05,
+      "loss": 0.9268,
+      "step": 13813
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 2.0032916730957158e-05,
+      "loss": 0.849,
+      "step": 13814
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 2.0030345111351128e-05,
+      "loss": 0.8044,
+      "step": 13815
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 2.0027773491745104e-05,
+      "loss": 0.8536,
+      "step": 13816
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 2.0025201872139074e-05,
+      "loss": 0.3274,
+      "step": 13817
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 2.0022630252533047e-05,
+      "loss": 0.7692,
+      "step": 13818
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 2.002005863292702e-05,
+      "loss": 0.4381,
+      "step": 13819
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 2.001748701332099e-05,
+      "loss": 0.4703,
+      "step": 13820
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 2.0014915393714963e-05,
+      "loss": 0.9603,
+      "step": 13821
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 2.0012343774108936e-05,
+      "loss": 0.9232,
+      "step": 13822
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 2.0009772154502905e-05,
+      "loss": 0.812,
+      "step": 13823
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 2.000720053489688e-05,
+      "loss": 0.6841,
+      "step": 13824
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 2.000462891529085e-05,
+      "loss": 0.8469,
+      "step": 13825
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 2.0002057295684825e-05,
+      "loss": 0.473,
+      "step": 13826
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.9999485676078794e-05,
+      "loss": 0.4835,
+      "step": 13827
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.9996914056472767e-05,
+      "loss": 0.7779,
+      "step": 13828
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.999434243686674e-05,
+      "loss": 0.6075,
+      "step": 13829
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.999177081726071e-05,
+      "loss": 0.8305,
+      "step": 13830
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.9989199197654683e-05,
+      "loss": 0.8591,
+      "step": 13831
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.9986627578048656e-05,
+      "loss": 0.5729,
+      "step": 13832
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.9984055958442626e-05,
+      "loss": 0.4745,
+      "step": 13833
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.9981484338836602e-05,
+      "loss": 0.8078,
+      "step": 13834
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.9978912719230572e-05,
+      "loss": 0.5783,
+      "step": 13835
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.9976341099624542e-05,
+      "loss": 0.6828,
+      "step": 13836
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.9973769480018518e-05,
+      "loss": 0.704,
+      "step": 13837
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.9971197860412488e-05,
+      "loss": 0.6078,
+      "step": 13838
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.996862624080646e-05,
+      "loss": 0.8133,
+      "step": 13839
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.9966054621200434e-05,
+      "loss": 0.7376,
+      "step": 13840
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.9963483001594404e-05,
+      "loss": 0.6036,
+      "step": 13841
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.996091138198838e-05,
+      "loss": 0.609,
+      "step": 13842
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.995833976238235e-05,
+      "loss": 0.8122,
+      "step": 13843
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.995576814277632e-05,
+      "loss": 0.7657,
+      "step": 13844
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.9953196523170296e-05,
+      "loss": 0.8912,
+      "step": 13845
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.9950624903564266e-05,
+      "loss": 0.9064,
+      "step": 13846
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.994805328395824e-05,
+      "loss": 0.5435,
+      "step": 13847
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.9945481664352212e-05,
+      "loss": 0.4509,
+      "step": 13848
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.994291004474618e-05,
+      "loss": 0.4282,
+      "step": 13849
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.9940338425140155e-05,
+      "loss": 0.8915,
+      "step": 13850
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.9937766805534128e-05,
+      "loss": 0.4621,
+      "step": 13851
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.9935195185928097e-05,
+      "loss": 0.8968,
+      "step": 13852
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.993262356632207e-05,
+      "loss": 0.4633,
+      "step": 13853
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.9930051946716043e-05,
+      "loss": 1.1263,
+      "step": 13854
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.9927480327110017e-05,
+      "loss": 0.8194,
+      "step": 13855
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.9924908707503986e-05,
+      "loss": 0.6067,
+      "step": 13856
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.992233708789796e-05,
+      "loss": 0.5183,
+      "step": 13857
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.9919765468291932e-05,
+      "loss": 0.5049,
+      "step": 13858
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.9917193848685902e-05,
+      "loss": 0.7483,
+      "step": 13859
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.9914622229079875e-05,
+      "loss": 1.0974,
+      "step": 13860
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.9912050609473848e-05,
+      "loss": 0.8538,
+      "step": 13861
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.9909478989867818e-05,
+      "loss": 0.5788,
+      "step": 13862
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.9906907370261794e-05,
+      "loss": 0.6453,
+      "step": 13863
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.9904335750655764e-05,
+      "loss": 0.4732,
+      "step": 13864
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.9901764131049737e-05,
+      "loss": 0.4988,
+      "step": 13865
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.989919251144371e-05,
+      "loss": 0.775,
+      "step": 13866
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.989662089183768e-05,
+      "loss": 0.8256,
+      "step": 13867
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.9894049272231653e-05,
+      "loss": 0.3517,
+      "step": 13868
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.9891477652625626e-05,
+      "loss": 0.6037,
+      "step": 13869
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.9888906033019596e-05,
+      "loss": 0.5104,
+      "step": 13870
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.988633441341357e-05,
+      "loss": 0.8116,
+      "step": 13871
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.9883762793807542e-05,
+      "loss": 0.9669,
+      "step": 13872
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.9881191174201515e-05,
+      "loss": 0.4671,
+      "step": 13873
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.9878619554595485e-05,
+      "loss": 0.6341,
+      "step": 13874
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.9876047934989458e-05,
+      "loss": 0.6162,
+      "step": 13875
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.987347631538343e-05,
+      "loss": 0.8645,
+      "step": 13876
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.98709046957774e-05,
+      "loss": 0.8264,
+      "step": 13877
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.9868333076171374e-05,
+      "loss": 0.8801,
+      "step": 13878
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.9865761456565347e-05,
+      "loss": 0.6976,
+      "step": 13879
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.9863189836959316e-05,
+      "loss": 0.7001,
+      "step": 13880
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.9860618217353293e-05,
+      "loss": 0.4589,
+      "step": 13881
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.9858046597747262e-05,
+      "loss": 1.341,
+      "step": 13882
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.9855474978141232e-05,
+      "loss": 0.4274,
+      "step": 13883
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.985290335853521e-05,
+      "loss": 0.6212,
+      "step": 13884
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.9850331738929178e-05,
+      "loss": 0.5048,
+      "step": 13885
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.984776011932315e-05,
+      "loss": 0.6313,
+      "step": 13886
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.9845188499717124e-05,
+      "loss": 0.7343,
+      "step": 13887
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.9842616880111094e-05,
+      "loss": 0.7172,
+      "step": 13888
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.9840045260505067e-05,
+      "loss": 0.4572,
+      "step": 13889
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.983747364089904e-05,
+      "loss": 0.4875,
+      "step": 13890
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.983490202129301e-05,
+      "loss": 0.528,
+      "step": 13891
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.9832330401686983e-05,
+      "loss": 0.7803,
+      "step": 13892
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.9829758782080956e-05,
+      "loss": 0.7012,
+      "step": 13893
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.982718716247493e-05,
+      "loss": 0.4835,
+      "step": 13894
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.98246155428689e-05,
+      "loss": 1.0246,
+      "step": 13895
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.9822043923262872e-05,
+      "loss": 0.6622,
+      "step": 13896
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.9819472303656845e-05,
+      "loss": 0.9561,
+      "step": 13897
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.9816900684050815e-05,
+      "loss": 0.8911,
+      "step": 13898
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.9814329064444788e-05,
+      "loss": 0.7396,
+      "step": 13899
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.981175744483876e-05,
+      "loss": 0.783,
+      "step": 13900
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.980918582523273e-05,
+      "loss": 0.6417,
+      "step": 13901
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.9806614205626707e-05,
+      "loss": 0.7702,
+      "step": 13902
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.9804042586020677e-05,
+      "loss": 0.3585,
+      "step": 13903
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.980147096641465e-05,
+      "loss": 0.7944,
+      "step": 13904
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.9798899346808623e-05,
+      "loss": 0.5737,
+      "step": 13905
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.9796327727202592e-05,
+      "loss": 0.679,
+      "step": 13906
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.9793756107596566e-05,
+      "loss": 0.6094,
+      "step": 13907
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.979118448799054e-05,
+      "loss": 0.7212,
+      "step": 13908
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.978861286838451e-05,
+      "loss": 0.8612,
+      "step": 13909
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.978604124877848e-05,
+      "loss": 0.6082,
+      "step": 13910
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.9783469629172454e-05,
+      "loss": 0.7823,
+      "step": 13911
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.9780898009566428e-05,
+      "loss": 0.6381,
+      "step": 13912
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.9778326389960397e-05,
+      "loss": 0.7062,
+      "step": 13913
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.977575477035437e-05,
+      "loss": 0.4742,
+      "step": 13914
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.9773183150748343e-05,
+      "loss": 0.3676,
+      "step": 13915
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.9770611531142313e-05,
+      "loss": 1.182,
+      "step": 13916
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.9768039911536286e-05,
+      "loss": 0.6287,
+      "step": 13917
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.976546829193026e-05,
+      "loss": 0.6139,
+      "step": 13918
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.976289667232423e-05,
+      "loss": 0.995,
+      "step": 13919
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.9760325052718205e-05,
+      "loss": 0.5554,
+      "step": 13920
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.9757753433112175e-05,
+      "loss": 0.8316,
+      "step": 13921
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.9755181813506145e-05,
+      "loss": 0.8585,
+      "step": 13922
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.975261019390012e-05,
+      "loss": 0.5049,
+      "step": 13923
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.975003857429409e-05,
+      "loss": 0.8441,
+      "step": 13924
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.9747466954688064e-05,
+      "loss": 0.8297,
+      "step": 13925
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.9744895335082037e-05,
+      "loss": 0.6596,
+      "step": 13926
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.9742323715476007e-05,
+      "loss": 0.5013,
+      "step": 13927
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.973975209586998e-05,
+      "loss": 0.5591,
+      "step": 13928
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.9737180476263953e-05,
+      "loss": 0.3471,
+      "step": 13929
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.9734608856657923e-05,
+      "loss": 0.3847,
+      "step": 13930
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.9732037237051896e-05,
+      "loss": 0.7048,
+      "step": 13931
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.972946561744587e-05,
+      "loss": 0.47,
+      "step": 13932
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.9726893997839842e-05,
+      "loss": 1.0947,
+      "step": 13933
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.972432237823381e-05,
+      "loss": 0.9919,
+      "step": 13934
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.9721750758627785e-05,
+      "loss": 0.6734,
+      "step": 13935
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.9719179139021758e-05,
+      "loss": 0.9192,
+      "step": 13936
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.9716607519415727e-05,
+      "loss": 0.6501,
+      "step": 13937
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.97140358998097e-05,
+      "loss": 0.4068,
+      "step": 13938
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.9711464280203673e-05,
+      "loss": 0.607,
+      "step": 13939
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.9708892660597643e-05,
+      "loss": 0.6841,
+      "step": 13940
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.970632104099162e-05,
+      "loss": 0.6657,
+      "step": 13941
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.970374942138559e-05,
+      "loss": 0.9716,
+      "step": 13942
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.9701177801779562e-05,
+      "loss": 0.8147,
+      "step": 13943
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.9698606182173535e-05,
+      "loss": 0.8376,
+      "step": 13944
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.9696034562567505e-05,
+      "loss": 0.8044,
+      "step": 13945
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.9693462942961478e-05,
+      "loss": 0.6869,
+      "step": 13946
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.969089132335545e-05,
+      "loss": 0.8082,
+      "step": 13947
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.968831970374942e-05,
+      "loss": 0.2525,
+      "step": 13948
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.9685748084143394e-05,
+      "loss": 0.4767,
+      "step": 13949
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.9683176464537367e-05,
+      "loss": 0.8741,
+      "step": 13950
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.968060484493134e-05,
+      "loss": 0.481,
+      "step": 13951
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.967803322532531e-05,
+      "loss": 0.8066,
+      "step": 13952
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.9675461605719283e-05,
+      "loss": 0.5539,
+      "step": 13953
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.9672889986113256e-05,
+      "loss": 0.5123,
+      "step": 13954
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.9670318366507226e-05,
+      "loss": 0.7798,
+      "step": 13955
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.96677467469012e-05,
+      "loss": 0.6806,
+      "step": 13956
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.9665175127295172e-05,
+      "loss": 0.7188,
+      "step": 13957
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.966260350768914e-05,
+      "loss": 0.5779,
+      "step": 13958
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.9660031888083118e-05,
+      "loss": 1.1259,
+      "step": 13959
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.9657460268477088e-05,
+      "loss": 0.6865,
+      "step": 13960
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.9654888648871057e-05,
+      "loss": 0.4569,
+      "step": 13961
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.9652317029265034e-05,
+      "loss": 0.7183,
+      "step": 13962
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.9649745409659003e-05,
+      "loss": 0.5713,
+      "step": 13963
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.9647173790052977e-05,
+      "loss": 0.7453,
+      "step": 13964
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.964460217044695e-05,
+      "loss": 0.5608,
+      "step": 13965
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.964203055084092e-05,
+      "loss": 1.08,
+      "step": 13966
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.9639458931234892e-05,
+      "loss": 0.9598,
+      "step": 13967
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.9636887311628865e-05,
+      "loss": 0.6306,
+      "step": 13968
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.9634315692022835e-05,
+      "loss": 0.7636,
+      "step": 13969
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.9631744072416808e-05,
+      "loss": 0.6717,
+      "step": 13970
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.962917245281078e-05,
+      "loss": 0.9539,
+      "step": 13971
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.9626600833204754e-05,
+      "loss": 0.6347,
+      "step": 13972
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.9624029213598724e-05,
+      "loss": 0.4973,
+      "step": 13973
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.9621457593992697e-05,
+      "loss": 0.4681,
+      "step": 13974
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.961888597438667e-05,
+      "loss": 0.6757,
+      "step": 13975
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.961631435478064e-05,
+      "loss": 0.6157,
+      "step": 13976
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.9613742735174613e-05,
+      "loss": 0.608,
+      "step": 13977
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.9611171115568586e-05,
+      "loss": 0.8562,
+      "step": 13978
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.9608599495962556e-05,
+      "loss": 0.8659,
+      "step": 13979
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.9606027876356532e-05,
+      "loss": 0.5596,
+      "step": 13980
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.9603456256750502e-05,
+      "loss": 0.6985,
+      "step": 13981
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.960088463714447e-05,
+      "loss": 0.5688,
+      "step": 13982
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.9598313017538448e-05,
+      "loss": 1.171,
+      "step": 13983
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.9595741397932418e-05,
+      "loss": 0.7486,
+      "step": 13984
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.959316977832639e-05,
+      "loss": 0.8875,
+      "step": 13985
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.9590598158720364e-05,
+      "loss": 0.6713,
+      "step": 13986
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.9588026539114333e-05,
+      "loss": 0.6641,
+      "step": 13987
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.9585454919508307e-05,
+      "loss": 0.7023,
+      "step": 13988
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.958288329990228e-05,
+      "loss": 0.4762,
+      "step": 13989
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.9580311680296253e-05,
+      "loss": 0.8899,
+      "step": 13990
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.9577740060690222e-05,
+      "loss": 0.7651,
+      "step": 13991
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.9575168441084195e-05,
+      "loss": 0.5876,
+      "step": 13992
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.957259682147817e-05,
+      "loss": 0.8564,
+      "step": 13993
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.957002520187214e-05,
+      "loss": 0.7141,
+      "step": 13994
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.956745358226611e-05,
+      "loss": 0.8562,
+      "step": 13995
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.9564881962660084e-05,
+      "loss": 0.7624,
+      "step": 13996
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.9562310343054057e-05,
+      "loss": 0.8609,
+      "step": 13997
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.955973872344803e-05,
+      "loss": 0.6581,
+      "step": 13998
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.9557167103842e-05,
+      "loss": 0.5878,
+      "step": 13999
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.9554595484235973e-05,
+      "loss": 0.6985,
+      "step": 14000
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.9552023864629946e-05,
+      "loss": 0.5716,
+      "step": 14001
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.9549452245023916e-05,
+      "loss": 0.4478,
+      "step": 14002
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.954688062541789e-05,
+      "loss": 0.5996,
+      "step": 14003
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.9544309005811862e-05,
+      "loss": 0.9342,
+      "step": 14004
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.9541737386205832e-05,
+      "loss": 0.4047,
+      "step": 14005
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.953916576659981e-05,
+      "loss": 0.7639,
+      "step": 14006
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.9536594146993778e-05,
+      "loss": 0.6355,
+      "step": 14007
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.9534022527387748e-05,
+      "loss": 0.7562,
+      "step": 14008
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.9531450907781724e-05,
+      "loss": 0.488,
+      "step": 14009
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.9528879288175694e-05,
+      "loss": 0.8285,
+      "step": 14010
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.9526307668569667e-05,
+      "loss": 0.9267,
+      "step": 14011
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.952373604896364e-05,
+      "loss": 0.8053,
+      "step": 14012
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.952116442935761e-05,
+      "loss": 0.537,
+      "step": 14013
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.9518592809751583e-05,
+      "loss": 0.6697,
+      "step": 14014
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.9516021190145556e-05,
+      "loss": 0.7146,
+      "step": 14015
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.9513449570539526e-05,
+      "loss": 0.857,
+      "step": 14016
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.95108779509335e-05,
+      "loss": 0.8262,
+      "step": 14017
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.950830633132747e-05,
+      "loss": 0.7105,
+      "step": 14018
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.9505734711721445e-05,
+      "loss": 0.5394,
+      "step": 14019
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.9503163092115414e-05,
+      "loss": 0.5697,
+      "step": 14020
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.9500591472509388e-05,
+      "loss": 0.8059,
+      "step": 14021
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.949801985290336e-05,
+      "loss": 0.7599,
+      "step": 14022
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.949544823329733e-05,
+      "loss": 0.4689,
+      "step": 14023
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.9492876613691303e-05,
+      "loss": 0.5502,
+      "step": 14024
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.9490304994085276e-05,
+      "loss": 1.082,
+      "step": 14025
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.9487733374479246e-05,
+      "loss": 0.5209,
+      "step": 14026
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.9485161754873223e-05,
+      "loss": 0.8947,
+      "step": 14027
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.9482590135267192e-05,
+      "loss": 0.6751,
+      "step": 14028
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.9480018515661165e-05,
+      "loss": 0.8214,
+      "step": 14029
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.947744689605514e-05,
+      "loss": 0.352,
+      "step": 14030
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.9474875276449108e-05,
+      "loss": 1.0448,
+      "step": 14031
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.947230365684308e-05,
+      "loss": 0.7324,
+      "step": 14032
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.9469732037237054e-05,
+      "loss": 0.7337,
+      "step": 14033
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.9467160417631024e-05,
+      "loss": 0.7063,
+      "step": 14034
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.9464588798024997e-05,
+      "loss": 0.8428,
+      "step": 14035
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.946201717841897e-05,
+      "loss": 0.5368,
+      "step": 14036
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.9459445558812943e-05,
+      "loss": 0.6285,
+      "step": 14037
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.9456873939206913e-05,
+      "loss": 0.7026,
+      "step": 14038
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.9454302319600886e-05,
+      "loss": 0.8837,
+      "step": 14039
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.945173069999486e-05,
+      "loss": 0.7639,
+      "step": 14040
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.944915908038883e-05,
+      "loss": 0.6237,
+      "step": 14041
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.9446587460782802e-05,
+      "loss": 0.7902,
+      "step": 14042
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.9444015841176775e-05,
+      "loss": 0.6906,
+      "step": 14043
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.9441444221570744e-05,
+      "loss": 0.5718,
+      "step": 14044
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.943887260196472e-05,
+      "loss": 0.6331,
+      "step": 14045
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.943630098235869e-05,
+      "loss": 1.0587,
+      "step": 14046
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.943372936275266e-05,
+      "loss": 0.7802,
+      "step": 14047
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.9431157743146637e-05,
+      "loss": 0.4723,
+      "step": 14048
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.9428586123540606e-05,
+      "loss": 0.6642,
+      "step": 14049
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.942601450393458e-05,
+      "loss": 0.505,
+      "step": 14050
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.9423442884328553e-05,
+      "loss": 0.6092,
+      "step": 14051
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.9420871264722522e-05,
+      "loss": 0.513,
+      "step": 14052
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.9418299645116495e-05,
+      "loss": 0.6162,
+      "step": 14053
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.941572802551047e-05,
+      "loss": 0.3728,
+      "step": 14054
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.9413156405904438e-05,
+      "loss": 0.7398,
+      "step": 14055
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.941058478629841e-05,
+      "loss": 0.715,
+      "step": 14056
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.9408013166692384e-05,
+      "loss": 1.0898,
+      "step": 14057
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.9405441547086357e-05,
+      "loss": 0.6243,
+      "step": 14058
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.9402869927480327e-05,
+      "loss": 0.566,
+      "step": 14059
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.94002983078743e-05,
+      "loss": 1.0454,
+      "step": 14060
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.9397726688268273e-05,
+      "loss": 0.6086,
+      "step": 14061
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.9395155068662243e-05,
+      "loss": 0.5468,
+      "step": 14062
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.9392583449056216e-05,
+      "loss": 0.6641,
+      "step": 14063
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.939001182945019e-05,
+      "loss": 0.7395,
+      "step": 14064
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.938744020984416e-05,
+      "loss": 0.5967,
+      "step": 14065
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.9384868590238135e-05,
+      "loss": 0.4975,
+      "step": 14066
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.9382296970632105e-05,
+      "loss": 0.8518,
+      "step": 14067
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.9379725351026078e-05,
+      "loss": 0.6218,
+      "step": 14068
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.937715373142005e-05,
+      "loss": 0.7011,
+      "step": 14069
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.937458211181402e-05,
+      "loss": 0.8832,
+      "step": 14070
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.9372010492207994e-05,
+      "loss": 0.672,
+      "step": 14071
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.9369438872601967e-05,
+      "loss": 0.9763,
+      "step": 14072
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.9366867252995936e-05,
+      "loss": 0.6342,
+      "step": 14073
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.936429563338991e-05,
+      "loss": 0.4272,
+      "step": 14074
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.9361724013783883e-05,
+      "loss": 1.0589,
+      "step": 14075
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.9359152394177856e-05,
+      "loss": 0.5183,
+      "step": 14076
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.9356580774571825e-05,
+      "loss": 0.4701,
+      "step": 14077
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.93540091549658e-05,
+      "loss": 0.5231,
+      "step": 14078
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.935143753535977e-05,
+      "loss": 0.6554,
+      "step": 14079
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.934886591575374e-05,
+      "loss": 0.8075,
+      "step": 14080
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.9346294296147714e-05,
+      "loss": 0.7706,
+      "step": 14081
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.9343722676541687e-05,
+      "loss": 0.8506,
+      "step": 14082
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.9341151056935657e-05,
+      "loss": 0.4117,
+      "step": 14083
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.9338579437329634e-05,
+      "loss": 0.6181,
+      "step": 14084
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.9336007817723603e-05,
+      "loss": 0.476,
+      "step": 14085
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.9333436198117573e-05,
+      "loss": 0.3379,
+      "step": 14086
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.933086457851155e-05,
+      "loss": 0.6011,
+      "step": 14087
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.932829295890552e-05,
+      "loss": 0.833,
+      "step": 14088
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.9325721339299492e-05,
+      "loss": 0.6227,
+      "step": 14089
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.9323149719693465e-05,
+      "loss": 0.7284,
+      "step": 14090
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.9320578100087435e-05,
+      "loss": 0.7571,
+      "step": 14091
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.9318006480481408e-05,
+      "loss": 0.6693,
+      "step": 14092
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.931543486087538e-05,
+      "loss": 0.8067,
+      "step": 14093
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.931286324126935e-05,
+      "loss": 0.4567,
+      "step": 14094
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.9310291621663324e-05,
+      "loss": 0.5123,
+      "step": 14095
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.9307720002057297e-05,
+      "loss": 0.6787,
+      "step": 14096
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.930514838245127e-05,
+      "loss": 0.6169,
+      "step": 14097
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.930257676284524e-05,
+      "loss": 0.7034,
+      "step": 14098
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.9300005143239213e-05,
+      "loss": 0.8776,
+      "step": 14099
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.9297433523633186e-05,
+      "loss": 0.6124,
+      "step": 14100
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.9294861904027155e-05,
+      "loss": 0.871,
+      "step": 14101
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.929229028442113e-05,
+      "loss": 0.4613,
+      "step": 14102
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.92897186648151e-05,
+      "loss": 0.7875,
+      "step": 14103
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.928714704520907e-05,
+      "loss": 0.8751,
+      "step": 14104
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.9284575425603048e-05,
+      "loss": 0.5596,
+      "step": 14105
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.9282003805997017e-05,
+      "loss": 0.6669,
+      "step": 14106
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.9279432186390987e-05,
+      "loss": 0.6541,
+      "step": 14107
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.9276860566784964e-05,
+      "loss": 0.5928,
+      "step": 14108
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.9274288947178933e-05,
+      "loss": 0.7088,
+      "step": 14109
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.9271717327572906e-05,
+      "loss": 0.6519,
+      "step": 14110
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.926914570796688e-05,
+      "loss": 0.4246,
+      "step": 14111
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.926657408836085e-05,
+      "loss": 0.6475,
+      "step": 14112
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.9264002468754822e-05,
+      "loss": 0.9309,
+      "step": 14113
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.9261430849148795e-05,
+      "loss": 0.7327,
+      "step": 14114
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.925885922954277e-05,
+      "loss": 0.7288,
+      "step": 14115
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.9256287609936738e-05,
+      "loss": 0.8672,
+      "step": 14116
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.925371599033071e-05,
+      "loss": 0.5392,
+      "step": 14117
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.9251144370724684e-05,
+      "loss": 1.0425,
+      "step": 14118
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.9248572751118654e-05,
+      "loss": 0.6299,
+      "step": 14119
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.9246001131512627e-05,
+      "loss": 0.6537,
+      "step": 14120
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.92434295119066e-05,
+      "loss": 0.8436,
+      "step": 14121
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.924085789230057e-05,
+      "loss": 0.6316,
+      "step": 14122
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.9238286272694546e-05,
+      "loss": 0.8254,
+      "step": 14123
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.9235714653088516e-05,
+      "loss": 0.7358,
+      "step": 14124
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.9233143033482485e-05,
+      "loss": 0.7498,
+      "step": 14125
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.9230571413876462e-05,
+      "loss": 0.6693,
+      "step": 14126
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.922799979427043e-05,
+      "loss": 0.5881,
+      "step": 14127
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.9225428174664405e-05,
+      "loss": 0.5494,
+      "step": 14128
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.9222856555058378e-05,
+      "loss": 0.5021,
+      "step": 14129
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.9220284935452347e-05,
+      "loss": 0.9273,
+      "step": 14130
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.921771331584632e-05,
+      "loss": 0.5624,
+      "step": 14131
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.9215141696240294e-05,
+      "loss": 0.6897,
+      "step": 14132
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.9212570076634263e-05,
+      "loss": 0.8723,
+      "step": 14133
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.9209998457028236e-05,
+      "loss": 1.0469,
+      "step": 14134
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.920742683742221e-05,
+      "loss": 0.6268,
+      "step": 14135
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.9204855217816183e-05,
+      "loss": 0.3481,
+      "step": 14136
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.9202283598210152e-05,
+      "loss": 0.9027,
+      "step": 14137
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.9199711978604125e-05,
+      "loss": 0.8154,
+      "step": 14138
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.91971403589981e-05,
+      "loss": 0.8737,
+      "step": 14139
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.9194568739392068e-05,
+      "loss": 0.5189,
+      "step": 14140
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.919199711978604e-05,
+      "loss": 0.6134,
+      "step": 14141
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.9189425500180014e-05,
+      "loss": 0.781,
+      "step": 14142
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.9186853880573984e-05,
+      "loss": 0.6384,
+      "step": 14143
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.918428226096796e-05,
+      "loss": 0.7962,
+      "step": 14144
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.918171064136193e-05,
+      "loss": 0.8157,
+      "step": 14145
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.9179139021755903e-05,
+      "loss": 1.0435,
+      "step": 14146
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.9176567402149876e-05,
+      "loss": 0.7123,
+      "step": 14147
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.9173995782543846e-05,
+      "loss": 0.4521,
+      "step": 14148
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.917142416293782e-05,
+      "loss": 0.8075,
+      "step": 14149
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.9168852543331792e-05,
+      "loss": 0.7998,
+      "step": 14150
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.916628092372576e-05,
+      "loss": 0.4284,
+      "step": 14151
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.9163709304119738e-05,
+      "loss": 0.9621,
+      "step": 14152
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.9161137684513708e-05,
+      "loss": 1.0866,
+      "step": 14153
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.915856606490768e-05,
+      "loss": 0.7522,
+      "step": 14154
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.9155994445301654e-05,
+      "loss": 0.9085,
+      "step": 14155
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.9153422825695624e-05,
+      "loss": 0.6506,
+      "step": 14156
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.9150851206089597e-05,
+      "loss": 0.7365,
+      "step": 14157
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.914827958648357e-05,
+      "loss": 0.6209,
+      "step": 14158
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.914570796687754e-05,
+      "loss": 0.5852,
+      "step": 14159
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.9143136347271513e-05,
+      "loss": 0.7273,
+      "step": 14160
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.9140564727665486e-05,
+      "loss": 0.7385,
+      "step": 14161
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.913799310805946e-05,
+      "loss": 0.675,
+      "step": 14162
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.913542148845343e-05,
+      "loss": 0.5943,
+      "step": 14163
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.91328498688474e-05,
+      "loss": 0.5342,
+      "step": 14164
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.9130278249241375e-05,
+      "loss": 0.5947,
+      "step": 14165
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.9127706629635344e-05,
+      "loss": 0.6702,
+      "step": 14166
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.9125135010029317e-05,
+      "loss": 0.5297,
+      "step": 14167
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.912256339042329e-05,
+      "loss": 0.3159,
+      "step": 14168
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.911999177081726e-05,
+      "loss": 0.8299,
+      "step": 14169
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.9117420151211237e-05,
+      "loss": 0.5594,
+      "step": 14170
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.9114848531605206e-05,
+      "loss": 0.5694,
+      "step": 14171
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.9112276911999176e-05,
+      "loss": 0.6771,
+      "step": 14172
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.9109705292393152e-05,
+      "loss": 0.7069,
+      "step": 14173
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.9107133672787122e-05,
+      "loss": 0.8363,
+      "step": 14174
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.9104562053181095e-05,
+      "loss": 0.7597,
+      "step": 14175
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.9101990433575068e-05,
+      "loss": 0.6331,
+      "step": 14176
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.9099418813969038e-05,
+      "loss": 0.5117,
+      "step": 14177
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.909684719436301e-05,
+      "loss": 0.6394,
+      "step": 14178
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.9094275574756984e-05,
+      "loss": 0.8798,
+      "step": 14179
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.9091703955150954e-05,
+      "loss": 0.7624,
+      "step": 14180
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.9089132335544927e-05,
+      "loss": 0.4859,
+      "step": 14181
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.90865607159389e-05,
+      "loss": 0.8412,
+      "step": 14182
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.9083989096332873e-05,
+      "loss": 0.7738,
+      "step": 14183
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.9081417476726843e-05,
+      "loss": 0.4332,
+      "step": 14184
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.9078845857120816e-05,
+      "loss": 0.9382,
+      "step": 14185
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.907627423751479e-05,
+      "loss": 0.4211,
+      "step": 14186
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.907370261790876e-05,
+      "loss": 0.6273,
+      "step": 14187
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.907113099830273e-05,
+      "loss": 0.5827,
+      "step": 14188
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.9068559378696705e-05,
+      "loss": 0.8184,
+      "step": 14189
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.9065987759090674e-05,
+      "loss": 0.7382,
+      "step": 14190
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.906341613948465e-05,
+      "loss": 0.8307,
+      "step": 14191
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.906084451987862e-05,
+      "loss": 0.4089,
+      "step": 14192
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.9058272900272593e-05,
+      "loss": 0.5147,
+      "step": 14193
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.9055701280666567e-05,
+      "loss": 0.7702,
+      "step": 14194
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.9053129661060536e-05,
+      "loss": 0.5219,
+      "step": 14195
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.905055804145451e-05,
+      "loss": 0.46,
+      "step": 14196
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.9047986421848482e-05,
+      "loss": 0.3792,
+      "step": 14197
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.9045414802242452e-05,
+      "loss": 0.657,
+      "step": 14198
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.9042843182636425e-05,
+      "loss": 0.7612,
+      "step": 14199
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.9040271563030398e-05,
+      "loss": 0.5606,
+      "step": 14200
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.903769994342437e-05,
+      "loss": 0.7732,
+      "step": 14201
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.903512832381834e-05,
+      "loss": 0.6292,
+      "step": 14202
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.9032556704212314e-05,
+      "loss": 0.5821,
+      "step": 14203
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.9029985084606287e-05,
+      "loss": 0.773,
+      "step": 14204
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.9027413465000257e-05,
+      "loss": 0.6964,
+      "step": 14205
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.902484184539423e-05,
+      "loss": 0.7059,
+      "step": 14206
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.9022270225788203e-05,
+      "loss": 0.7974,
+      "step": 14207
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.9019698606182173e-05,
+      "loss": 0.6604,
+      "step": 14208
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.901712698657615e-05,
+      "loss": 0.5168,
+      "step": 14209
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.901455536697012e-05,
+      "loss": 0.5888,
+      "step": 14210
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.901198374736409e-05,
+      "loss": 0.6165,
+      "step": 14211
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.9009412127758065e-05,
+      "loss": 0.4505,
+      "step": 14212
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.9006840508152035e-05,
+      "loss": 0.7457,
+      "step": 14213
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.9004268888546008e-05,
+      "loss": 0.7566,
+      "step": 14214
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.900169726893998e-05,
+      "loss": 0.5268,
+      "step": 14215
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.899912564933395e-05,
+      "loss": 0.574,
+      "step": 14216
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.8996554029727924e-05,
+      "loss": 0.6863,
+      "step": 14217
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.8993982410121897e-05,
+      "loss": 0.5561,
+      "step": 14218
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.8991410790515866e-05,
+      "loss": 0.7787,
+      "step": 14219
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.898883917090984e-05,
+      "loss": 0.4671,
+      "step": 14220
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.8986267551303812e-05,
+      "loss": 0.9965,
+      "step": 14221
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.8983695931697786e-05,
+      "loss": 0.5486,
+      "step": 14222
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.8981124312091755e-05,
+      "loss": 0.5923,
+      "step": 14223
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.8978552692485728e-05,
+      "loss": 0.5229,
+      "step": 14224
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.89759810728797e-05,
+      "loss": 0.7838,
+      "step": 14225
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.897340945327367e-05,
+      "loss": 0.6446,
+      "step": 14226
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.8970837833667644e-05,
+      "loss": 0.8497,
+      "step": 14227
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.8968266214061617e-05,
+      "loss": 0.9584,
+      "step": 14228
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.8965694594455587e-05,
+      "loss": 1.2758,
+      "step": 14229
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.8963122974849563e-05,
+      "loss": 0.5418,
+      "step": 14230
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.8960551355243533e-05,
+      "loss": 0.4515,
+      "step": 14231
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.8957979735637506e-05,
+      "loss": 0.6701,
+      "step": 14232
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.895540811603148e-05,
+      "loss": 0.4636,
+      "step": 14233
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.895283649642545e-05,
+      "loss": 0.8032,
+      "step": 14234
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.8950264876819422e-05,
+      "loss": 0.5778,
+      "step": 14235
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.8947693257213395e-05,
+      "loss": 1.0064,
+      "step": 14236
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.8945121637607365e-05,
+      "loss": 0.4557,
+      "step": 14237
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.8942550018001338e-05,
+      "loss": 0.5579,
+      "step": 14238
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.893997839839531e-05,
+      "loss": 1.1092,
+      "step": 14239
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.8937406778789284e-05,
+      "loss": 0.7752,
+      "step": 14240
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.8934835159183254e-05,
+      "loss": 0.8364,
+      "step": 14241
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.8932263539577227e-05,
+      "loss": 0.732,
+      "step": 14242
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.89296919199712e-05,
+      "loss": 0.5395,
+      "step": 14243
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.892712030036517e-05,
+      "loss": 0.5763,
+      "step": 14244
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.8924548680759142e-05,
+      "loss": 0.7021,
+      "step": 14245
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.8921977061153116e-05,
+      "loss": 0.9077,
+      "step": 14246
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.8919405441547085e-05,
+      "loss": 0.8165,
+      "step": 14247
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.8916833821941062e-05,
+      "loss": 0.9908,
+      "step": 14248
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.891426220233503e-05,
+      "loss": 0.616,
+      "step": 14249
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.8911690582729e-05,
+      "loss": 0.8382,
+      "step": 14250
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.8909118963122978e-05,
+      "loss": 0.7362,
+      "step": 14251
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.8906547343516947e-05,
+      "loss": 0.6243,
+      "step": 14252
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.890397572391092e-05,
+      "loss": 0.7907,
+      "step": 14253
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.8901404104304893e-05,
+      "loss": 0.5876,
+      "step": 14254
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.8898832484698863e-05,
+      "loss": 0.5449,
+      "step": 14255
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.8896260865092836e-05,
+      "loss": 0.7036,
+      "step": 14256
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.889368924548681e-05,
+      "loss": 0.8463,
+      "step": 14257
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.889111762588078e-05,
+      "loss": 0.8437,
+      "step": 14258
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.8888546006274752e-05,
+      "loss": 0.6463,
+      "step": 14259
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.8885974386668725e-05,
+      "loss": 0.6253,
+      "step": 14260
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.8883402767062698e-05,
+      "loss": 0.7301,
+      "step": 14261
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.8880831147456668e-05,
+      "loss": 0.6271,
+      "step": 14262
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.887825952785064e-05,
+      "loss": 0.9397,
+      "step": 14263
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.8875687908244614e-05,
+      "loss": 0.8197,
+      "step": 14264
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.8873116288638584e-05,
+      "loss": 0.7949,
+      "step": 14265
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.8870544669032557e-05,
+      "loss": 0.5959,
+      "step": 14266
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.886797304942653e-05,
+      "loss": 1.1909,
+      "step": 14267
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.88654014298205e-05,
+      "loss": 0.677,
+      "step": 14268
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.8862829810214476e-05,
+      "loss": 0.6311,
+      "step": 14269
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.8860258190608446e-05,
+      "loss": 0.6508,
+      "step": 14270
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.8857686571002415e-05,
+      "loss": 0.8542,
+      "step": 14271
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.8855114951396392e-05,
+      "loss": 0.9296,
+      "step": 14272
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.885254333179036e-05,
+      "loss": 0.3372,
+      "step": 14273
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.8849971712184335e-05,
+      "loss": 0.7117,
+      "step": 14274
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.8847400092578308e-05,
+      "loss": 0.7366,
+      "step": 14275
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.8844828472972277e-05,
+      "loss": 0.8378,
+      "step": 14276
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.884225685336625e-05,
+      "loss": 0.8725,
+      "step": 14277
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.8839685233760223e-05,
+      "loss": 0.8545,
+      "step": 14278
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.8837113614154196e-05,
+      "loss": 0.5149,
+      "step": 14279
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.8834541994548166e-05,
+      "loss": 0.6978,
+      "step": 14280
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.883197037494214e-05,
+      "loss": 0.6227,
+      "step": 14281
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.8829398755336112e-05,
+      "loss": 0.5003,
+      "step": 14282
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.8826827135730082e-05,
+      "loss": 0.6314,
+      "step": 14283
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.8824255516124055e-05,
+      "loss": 0.9646,
+      "step": 14284
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.8821683896518028e-05,
+      "loss": 0.5014,
+      "step": 14285
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.8819112276911998e-05,
+      "loss": 0.5431,
+      "step": 14286
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.8816540657305974e-05,
+      "loss": 0.6402,
+      "step": 14287
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.8813969037699944e-05,
+      "loss": 0.9407,
+      "step": 14288
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.8811397418093914e-05,
+      "loss": 0.6391,
+      "step": 14289
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.880882579848789e-05,
+      "loss": 1.0357,
+      "step": 14290
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.880625417888186e-05,
+      "loss": 0.8201,
+      "step": 14291
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.8803682559275833e-05,
+      "loss": 0.8842,
+      "step": 14292
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.8801110939669806e-05,
+      "loss": 0.7038,
+      "step": 14293
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.8798539320063776e-05,
+      "loss": 0.8907,
+      "step": 14294
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.8795967700457752e-05,
+      "loss": 0.6139,
+      "step": 14295
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.8793396080851722e-05,
+      "loss": 0.8415,
+      "step": 14296
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.879082446124569e-05,
+      "loss": 0.6377,
+      "step": 14297
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.8788252841639668e-05,
+      "loss": 0.832,
+      "step": 14298
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.8785681222033638e-05,
+      "loss": 0.7643,
+      "step": 14299
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.878310960242761e-05,
+      "loss": 0.5039,
+      "step": 14300
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.8780537982821584e-05,
+      "loss": 0.5074,
+      "step": 14301
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.8777966363215553e-05,
+      "loss": 0.7689,
+      "step": 14302
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.8775394743609527e-05,
+      "loss": 0.4543,
+      "step": 14303
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.87728231240035e-05,
+      "loss": 0.6873,
+      "step": 14304
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.877025150439747e-05,
+      "loss": 0.4639,
+      "step": 14305
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.8767679884791442e-05,
+      "loss": 0.8619,
+      "step": 14306
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.8765108265185415e-05,
+      "loss": 0.88,
+      "step": 14307
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.876253664557939e-05,
+      "loss": 0.5791,
+      "step": 14308
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.8759965025973358e-05,
+      "loss": 0.7103,
+      "step": 14309
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.875739340636733e-05,
+      "loss": 0.875,
+      "step": 14310
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.8754821786761304e-05,
+      "loss": 0.7433,
+      "step": 14311
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.8752250167155274e-05,
+      "loss": 0.7728,
+      "step": 14312
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.8749678547549247e-05,
+      "loss": 0.8286,
+      "step": 14313
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.874710692794322e-05,
+      "loss": 0.9868,
+      "step": 14314
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.874453530833719e-05,
+      "loss": 0.7879,
+      "step": 14315
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.8741963688731166e-05,
+      "loss": 0.7245,
+      "step": 14316
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.8739392069125136e-05,
+      "loss": 0.5941,
+      "step": 14317
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.873682044951911e-05,
+      "loss": 0.5295,
+      "step": 14318
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.8734248829913082e-05,
+      "loss": 0.6083,
+      "step": 14319
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.8731677210307052e-05,
+      "loss": 0.756,
+      "step": 14320
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.8729105590701025e-05,
+      "loss": 0.8014,
+      "step": 14321
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.8726533971094998e-05,
+      "loss": 0.8774,
+      "step": 14322
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.8723962351488968e-05,
+      "loss": 0.7492,
+      "step": 14323
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.872139073188294e-05,
+      "loss": 0.4361,
+      "step": 14324
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.8718819112276914e-05,
+      "loss": 0.6866,
+      "step": 14325
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.8716247492670887e-05,
+      "loss": 0.9716,
+      "step": 14326
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.8713675873064857e-05,
+      "loss": 0.6022,
+      "step": 14327
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.871110425345883e-05,
+      "loss": 0.6527,
+      "step": 14328
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.8708532633852803e-05,
+      "loss": 0.66,
+      "step": 14329
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.8705961014246772e-05,
+      "loss": 0.7011,
+      "step": 14330
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.8703389394640745e-05,
+      "loss": 0.7709,
+      "step": 14331
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.870081777503472e-05,
+      "loss": 0.7511,
+      "step": 14332
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.8698246155428688e-05,
+      "loss": 0.7561,
+      "step": 14333
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.8695674535822665e-05,
+      "loss": 0.7058,
+      "step": 14334
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.8693102916216634e-05,
+      "loss": 0.4802,
+      "step": 14335
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.8690531296610604e-05,
+      "loss": 0.6492,
+      "step": 14336
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.868795967700458e-05,
+      "loss": 0.7228,
+      "step": 14337
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.868538805739855e-05,
+      "loss": 0.7014,
+      "step": 14338
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.8682816437792523e-05,
+      "loss": 0.8135,
+      "step": 14339
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.8680244818186496e-05,
+      "loss": 1.0478,
+      "step": 14340
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.8677673198580466e-05,
+      "loss": 0.6668,
+      "step": 14341
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.867510157897444e-05,
+      "loss": 0.4622,
+      "step": 14342
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.8672529959368412e-05,
+      "loss": 0.5991,
+      "step": 14343
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.8669958339762382e-05,
+      "loss": 0.5429,
+      "step": 14344
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.8667386720156355e-05,
+      "loss": 0.6902,
+      "step": 14345
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.8664815100550328e-05,
+      "loss": 0.8831,
+      "step": 14346
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.86622434809443e-05,
+      "loss": 0.7159,
+      "step": 14347
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.865967186133827e-05,
+      "loss": 0.6614,
+      "step": 14348
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.8657100241732244e-05,
+      "loss": 0.9166,
+      "step": 14349
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.8654528622126217e-05,
+      "loss": 1.0654,
+      "step": 14350
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.8651957002520187e-05,
+      "loss": 0.5624,
+      "step": 14351
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.864938538291416e-05,
+      "loss": 0.5252,
+      "step": 14352
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.8646813763308133e-05,
+      "loss": 0.7504,
+      "step": 14353
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.8644242143702102e-05,
+      "loss": 0.8025,
+      "step": 14354
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.864167052409608e-05,
+      "loss": 0.7491,
+      "step": 14355
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.863909890449005e-05,
+      "loss": 0.7838,
+      "step": 14356
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.863652728488402e-05,
+      "loss": 0.5425,
+      "step": 14357
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.8633955665277995e-05,
+      "loss": 0.7495,
+      "step": 14358
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.8631384045671964e-05,
+      "loss": 0.5578,
+      "step": 14359
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.8628812426065938e-05,
+      "loss": 0.8172,
+      "step": 14360
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.862624080645991e-05,
+      "loss": 0.7621,
+      "step": 14361
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.862366918685388e-05,
+      "loss": 0.4504,
+      "step": 14362
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.8621097567247853e-05,
+      "loss": 0.9048,
+      "step": 14363
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.8618525947641826e-05,
+      "loss": 0.3083,
+      "step": 14364
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.86159543280358e-05,
+      "loss": 0.6377,
+      "step": 14365
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.861338270842977e-05,
+      "loss": 1.1476,
+      "step": 14366
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.8610811088823742e-05,
+      "loss": 0.5089,
+      "step": 14367
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.8608239469217715e-05,
+      "loss": 0.635,
+      "step": 14368
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.8605667849611685e-05,
+      "loss": 0.6391,
+      "step": 14369
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.8603096230005658e-05,
+      "loss": 0.6125,
+      "step": 14370
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.860052461039963e-05,
+      "loss": 0.5414,
+      "step": 14371
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.85979529907936e-05,
+      "loss": 1.1042,
+      "step": 14372
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.8595381371187577e-05,
+      "loss": 0.68,
+      "step": 14373
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.8592809751581547e-05,
+      "loss": 0.6529,
+      "step": 14374
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.8590238131975517e-05,
+      "loss": 0.5852,
+      "step": 14375
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.8587666512369493e-05,
+      "loss": 0.7537,
+      "step": 14376
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.8585094892763463e-05,
+      "loss": 0.6498,
+      "step": 14377
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.8582523273157436e-05,
+      "loss": 0.8692,
+      "step": 14378
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.857995165355141e-05,
+      "loss": 0.7633,
+      "step": 14379
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.857738003394538e-05,
+      "loss": 0.2499,
+      "step": 14380
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.8574808414339352e-05,
+      "loss": 0.6408,
+      "step": 14381
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.8572236794733325e-05,
+      "loss": 0.6245,
+      "step": 14382
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.8569665175127294e-05,
+      "loss": 0.7087,
+      "step": 14383
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.8567093555521268e-05,
+      "loss": 0.5609,
+      "step": 14384
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.856452193591524e-05,
+      "loss": 0.6932,
+      "step": 14385
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.8561950316309214e-05,
+      "loss": 0.5364,
+      "step": 14386
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.8559378696703183e-05,
+      "loss": 0.6886,
+      "step": 14387
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.8556807077097156e-05,
+      "loss": 0.6486,
+      "step": 14388
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.855423545749113e-05,
+      "loss": 0.5433,
+      "step": 14389
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.85516638378851e-05,
+      "loss": 0.7467,
+      "step": 14390
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.8549092218279072e-05,
+      "loss": 0.737,
+      "step": 14391
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.8546520598673045e-05,
+      "loss": 0.5151,
+      "step": 14392
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.8543948979067015e-05,
+      "loss": 0.5752,
+      "step": 14393
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.854137735946099e-05,
+      "loss": 0.6053,
+      "step": 14394
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.853880573985496e-05,
+      "loss": 0.5673,
+      "step": 14395
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.853623412024893e-05,
+      "loss": 0.6851,
+      "step": 14396
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.8533662500642907e-05,
+      "loss": 0.5801,
+      "step": 14397
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.8531090881036877e-05,
+      "loss": 0.5765,
+      "step": 14398
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.852851926143085e-05,
+      "loss": 0.4202,
+      "step": 14399
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.8525947641824823e-05,
+      "loss": 0.3511,
+      "step": 14400
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.8523376022218793e-05,
+      "loss": 0.3952,
+      "step": 14401
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.8520804402612766e-05,
+      "loss": 0.8092,
+      "step": 14402
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.851823278300674e-05,
+      "loss": 0.5452,
+      "step": 14403
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.8515661163400712e-05,
+      "loss": 1.1779,
+      "step": 14404
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.8513089543794682e-05,
+      "loss": 0.7765,
+      "step": 14405
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.8510517924188655e-05,
+      "loss": 0.5499,
+      "step": 14406
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.8507946304582628e-05,
+      "loss": 0.6214,
+      "step": 14407
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.8505374684976598e-05,
+      "loss": 0.5416,
+      "step": 14408
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.850280306537057e-05,
+      "loss": 0.5861,
+      "step": 14409
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.8500231445764544e-05,
+      "loss": 0.8702,
+      "step": 14410
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.8497659826158513e-05,
+      "loss": 0.3818,
+      "step": 14411
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.849508820655249e-05,
+      "loss": 0.6777,
+      "step": 14412
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.849251658694646e-05,
+      "loss": 0.9388,
+      "step": 14413
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.848994496734043e-05,
+      "loss": 0.7251,
+      "step": 14414
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.8487373347734406e-05,
+      "loss": 0.9956,
+      "step": 14415
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.8484801728128375e-05,
+      "loss": 0.5876,
+      "step": 14416
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.848223010852235e-05,
+      "loss": 0.7469,
+      "step": 14417
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.847965848891632e-05,
+      "loss": 1.0623,
+      "step": 14418
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.847708686931029e-05,
+      "loss": 0.4944,
+      "step": 14419
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.8474515249704264e-05,
+      "loss": 0.6587,
+      "step": 14420
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.8471943630098237e-05,
+      "loss": 0.9555,
+      "step": 14421
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.8469372010492207e-05,
+      "loss": 0.597,
+      "step": 14422
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.846680039088618e-05,
+      "loss": 0.67,
+      "step": 14423
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.8464228771280153e-05,
+      "loss": 0.6947,
+      "step": 14424
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.8461657151674126e-05,
+      "loss": 0.7832,
+      "step": 14425
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.8459085532068096e-05,
+      "loss": 0.6998,
+      "step": 14426
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.845651391246207e-05,
+      "loss": 0.8549,
+      "step": 14427
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.8453942292856042e-05,
+      "loss": 0.8742,
+      "step": 14428
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.8451370673250012e-05,
+      "loss": 0.5204,
+      "step": 14429
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.8448799053643985e-05,
+      "loss": 0.7192,
+      "step": 14430
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.8446227434037958e-05,
+      "loss": 0.6068,
+      "step": 14431
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.8443655814431928e-05,
+      "loss": 0.7476,
+      "step": 14432
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.8441084194825904e-05,
+      "loss": 0.5913,
+      "step": 14433
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.8438512575219874e-05,
+      "loss": 0.452,
+      "step": 14434
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.8435940955613843e-05,
+      "loss": 0.8892,
+      "step": 14435
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.843336933600782e-05,
+      "loss": 0.7832,
+      "step": 14436
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.843079771640179e-05,
+      "loss": 0.7489,
+      "step": 14437
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.8428226096795763e-05,
+      "loss": 0.6797,
+      "step": 14438
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.8425654477189736e-05,
+      "loss": 1.0363,
+      "step": 14439
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.8423082857583705e-05,
+      "loss": 0.7969,
+      "step": 14440
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.842051123797768e-05,
+      "loss": 0.6088,
+      "step": 14441
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.841793961837165e-05,
+      "loss": 0.6629,
+      "step": 14442
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.8415367998765625e-05,
+      "loss": 0.5505,
+      "step": 14443
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.8412796379159594e-05,
+      "loss": 0.5687,
+      "step": 14444
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.8410224759553567e-05,
+      "loss": 0.98,
+      "step": 14445
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.840765313994754e-05,
+      "loss": 0.7268,
+      "step": 14446
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.8405081520341514e-05,
+      "loss": 0.8997,
+      "step": 14447
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.8402509900735483e-05,
+      "loss": 0.5061,
+      "step": 14448
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.8399938281129456e-05,
+      "loss": 0.7436,
+      "step": 14449
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.839736666152343e-05,
+      "loss": 0.7176,
+      "step": 14450
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.8394795041917402e-05,
+      "loss": 0.828,
+      "step": 14451
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.8392223422311372e-05,
+      "loss": 0.4759,
+      "step": 14452
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.8389651802705345e-05,
+      "loss": 0.4862,
+      "step": 14453
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.838708018309932e-05,
+      "loss": 0.6325,
+      "step": 14454
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.8384508563493288e-05,
+      "loss": 0.9558,
+      "step": 14455
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.838193694388726e-05,
+      "loss": 0.7656,
+      "step": 14456
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.8379365324281234e-05,
+      "loss": 0.5418,
+      "step": 14457
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.8376793704675204e-05,
+      "loss": 0.5861,
+      "step": 14458
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.837422208506918e-05,
+      "loss": 0.8057,
+      "step": 14459
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.837165046546315e-05,
+      "loss": 0.5461,
+      "step": 14460
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.836907884585712e-05,
+      "loss": 0.8344,
+      "step": 14461
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.8366507226251096e-05,
+      "loss": 0.5492,
+      "step": 14462
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.8363935606645066e-05,
+      "loss": 0.6968,
+      "step": 14463
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.836136398703904e-05,
+      "loss": 1.0635,
+      "step": 14464
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.8358792367433012e-05,
+      "loss": 0.386,
+      "step": 14465
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.835622074782698e-05,
+      "loss": 0.6605,
+      "step": 14466
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.8353649128220955e-05,
+      "loss": 0.622,
+      "step": 14467
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.8351077508614928e-05,
+      "loss": 0.6692,
+      "step": 14468
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.8348505889008897e-05,
+      "loss": 0.6682,
+      "step": 14469
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.834593426940287e-05,
+      "loss": 0.7805,
+      "step": 14470
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.8343362649796844e-05,
+      "loss": 0.7152,
+      "step": 14471
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.8340791030190817e-05,
+      "loss": 0.2035,
+      "step": 14472
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.8338219410584786e-05,
+      "loss": 0.7894,
+      "step": 14473
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.833564779097876e-05,
+      "loss": 0.6483,
+      "step": 14474
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.8333076171372733e-05,
+      "loss": 0.5942,
+      "step": 14475
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.8330504551766702e-05,
+      "loss": 0.9929,
+      "step": 14476
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.8327932932160675e-05,
+      "loss": 0.4834,
+      "step": 14477
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.832536131255465e-05,
+      "loss": 0.6725,
+      "step": 14478
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.8322789692948618e-05,
+      "loss": 0.7638,
+      "step": 14479
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.8320218073342595e-05,
+      "loss": 0.762,
+      "step": 14480
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.8317646453736564e-05,
+      "loss": 0.4661,
+      "step": 14481
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.8315074834130537e-05,
+      "loss": 0.6984,
+      "step": 14482
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.831250321452451e-05,
+      "loss": 0.7749,
+      "step": 14483
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.830993159491848e-05,
+      "loss": 0.8232,
+      "step": 14484
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.8307359975312453e-05,
+      "loss": 0.6103,
+      "step": 14485
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.8304788355706426e-05,
+      "loss": 0.8443,
+      "step": 14486
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.8302216736100396e-05,
+      "loss": 0.6045,
+      "step": 14487
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.829964511649437e-05,
+      "loss": 0.5625,
+      "step": 14488
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.8297073496888342e-05,
+      "loss": 0.4601,
+      "step": 14489
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.8294501877282315e-05,
+      "loss": 0.7024,
+      "step": 14490
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.8291930257676285e-05,
+      "loss": 0.775,
+      "step": 14491
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.8289358638070258e-05,
+      "loss": 0.6981,
+      "step": 14492
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.828678701846423e-05,
+      "loss": 0.7694,
+      "step": 14493
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.82842153988582e-05,
+      "loss": 0.5379,
+      "step": 14494
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.8281643779252174e-05,
+      "loss": 0.3963,
+      "step": 14495
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.8279072159646147e-05,
+      "loss": 0.6288,
+      "step": 14496
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.8276500540040116e-05,
+      "loss": 0.5288,
+      "step": 14497
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.8273928920434093e-05,
+      "loss": 0.7411,
+      "step": 14498
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.8271357300828063e-05,
+      "loss": 0.7858,
+      "step": 14499
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.8268785681222032e-05,
+      "loss": 0.5165,
+      "step": 14500
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.826621406161601e-05,
+      "loss": 0.5567,
+      "step": 14501
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.826364244200998e-05,
+      "loss": 0.9815,
+      "step": 14502
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.826107082240395e-05,
+      "loss": 1.1042,
+      "step": 14503
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.8258499202797925e-05,
+      "loss": 0.631,
+      "step": 14504
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.8255927583191894e-05,
+      "loss": 0.7883,
+      "step": 14505
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.8253355963585867e-05,
+      "loss": 0.6681,
+      "step": 14506
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.825078434397984e-05,
+      "loss": 0.6602,
+      "step": 14507
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.824821272437381e-05,
+      "loss": 0.8156,
+      "step": 14508
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.8245641104767783e-05,
+      "loss": 0.6264,
+      "step": 14509
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.8243069485161756e-05,
+      "loss": 0.4965,
+      "step": 14510
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.824049786555573e-05,
+      "loss": 0.6231,
+      "step": 14511
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.82379262459497e-05,
+      "loss": 0.9794,
+      "step": 14512
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.8235354626343672e-05,
+      "loss": 0.5634,
+      "step": 14513
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.8232783006737645e-05,
+      "loss": 0.6816,
+      "step": 14514
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.8230211387131615e-05,
+      "loss": 0.7193,
+      "step": 14515
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.8227639767525588e-05,
+      "loss": 0.599,
+      "step": 14516
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.822506814791956e-05,
+      "loss": 0.8999,
+      "step": 14517
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.822249652831353e-05,
+      "loss": 0.5311,
+      "step": 14518
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.8219924908707507e-05,
+      "loss": 0.8046,
+      "step": 14519
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.8217353289101477e-05,
+      "loss": 0.5255,
+      "step": 14520
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.821478166949545e-05,
+      "loss": 0.6289,
+      "step": 14521
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.8212210049889423e-05,
+      "loss": 0.4935,
+      "step": 14522
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.8209638430283393e-05,
+      "loss": 0.6487,
+      "step": 14523
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.8207066810677366e-05,
+      "loss": 0.8789,
+      "step": 14524
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.820449519107134e-05,
+      "loss": 0.6458,
+      "step": 14525
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.820192357146531e-05,
+      "loss": 0.8446,
+      "step": 14526
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.819935195185928e-05,
+      "loss": 0.9364,
+      "step": 14527
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.8196780332253255e-05,
+      "loss": 0.8289,
+      "step": 14528
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.8194208712647228e-05,
+      "loss": 0.7753,
+      "step": 14529
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.8191637093041197e-05,
+      "loss": 0.6275,
+      "step": 14530
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.818906547343517e-05,
+      "loss": 0.5935,
+      "step": 14531
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.8186493853829144e-05,
+      "loss": 1.0261,
+      "step": 14532
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.8183922234223113e-05,
+      "loss": 1.0129,
+      "step": 14533
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.8181350614617086e-05,
+      "loss": 0.6541,
+      "step": 14534
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.817877899501106e-05,
+      "loss": 0.5264,
+      "step": 14535
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.817620737540503e-05,
+      "loss": 0.6172,
+      "step": 14536
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.8173635755799005e-05,
+      "loss": 0.7078,
+      "step": 14537
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.8171064136192975e-05,
+      "loss": 0.5833,
+      "step": 14538
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.8168492516586945e-05,
+      "loss": 0.7301,
+      "step": 14539
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.816592089698092e-05,
+      "loss": 0.7746,
+      "step": 14540
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.816334927737489e-05,
+      "loss": 0.8194,
+      "step": 14541
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.8160777657768864e-05,
+      "loss": 0.8475,
+      "step": 14542
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.8158206038162837e-05,
+      "loss": 0.747,
+      "step": 14543
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.8155634418556807e-05,
+      "loss": 0.6081,
+      "step": 14544
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.815306279895078e-05,
+      "loss": 0.6078,
+      "step": 14545
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.8150491179344753e-05,
+      "loss": 0.5538,
+      "step": 14546
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.8147919559738723e-05,
+      "loss": 0.7638,
+      "step": 14547
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.8145347940132696e-05,
+      "loss": 0.4754,
+      "step": 14548
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.814277632052667e-05,
+      "loss": 0.8793,
+      "step": 14549
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.8140204700920642e-05,
+      "loss": 0.7601,
+      "step": 14550
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.813763308131461e-05,
+      "loss": 0.9802,
+      "step": 14551
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.8135061461708585e-05,
+      "loss": 0.6974,
+      "step": 14552
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.8132489842102558e-05,
+      "loss": 0.7852,
+      "step": 14553
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.8129918222496527e-05,
+      "loss": 0.8881,
+      "step": 14554
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.81273466028905e-05,
+      "loss": 0.4735,
+      "step": 14555
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.8124774983284474e-05,
+      "loss": 0.8165,
+      "step": 14556
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.8122203363678443e-05,
+      "loss": 0.5513,
+      "step": 14557
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.811963174407242e-05,
+      "loss": 0.661,
+      "step": 14558
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.811706012446639e-05,
+      "loss": 0.762,
+      "step": 14559
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.811448850486036e-05,
+      "loss": 0.7841,
+      "step": 14560
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.8111916885254336e-05,
+      "loss": 0.5824,
+      "step": 14561
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.8109345265648305e-05,
+      "loss": 0.8484,
+      "step": 14562
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.8106773646042278e-05,
+      "loss": 0.8039,
+      "step": 14563
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.810420202643625e-05,
+      "loss": 0.7412,
+      "step": 14564
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.810163040683022e-05,
+      "loss": 0.8596,
+      "step": 14565
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.8099058787224194e-05,
+      "loss": 0.6798,
+      "step": 14566
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.8096487167618167e-05,
+      "loss": 0.8018,
+      "step": 14567
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.809391554801214e-05,
+      "loss": 0.5138,
+      "step": 14568
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.809134392840611e-05,
+      "loss": 0.7727,
+      "step": 14569
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.8088772308800083e-05,
+      "loss": 0.7995,
+      "step": 14570
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.8086200689194056e-05,
+      "loss": 0.6907,
+      "step": 14571
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.8083629069588026e-05,
+      "loss": 1.0048,
+      "step": 14572
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.8081057449982e-05,
+      "loss": 0.6197,
+      "step": 14573
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.8078485830375972e-05,
+      "loss": 0.6342,
+      "step": 14574
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.807591421076994e-05,
+      "loss": 0.6795,
+      "step": 14575
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.8073342591163918e-05,
+      "loss": 1.0399,
+      "step": 14576
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.8070770971557888e-05,
+      "loss": 0.5476,
+      "step": 14577
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.8068199351951857e-05,
+      "loss": 0.6489,
+      "step": 14578
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.8065627732345834e-05,
+      "loss": 0.6313,
+      "step": 14579
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.8063056112739804e-05,
+      "loss": 0.7904,
+      "step": 14580
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.8060484493133777e-05,
+      "loss": 0.7402,
+      "step": 14581
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.805791287352775e-05,
+      "loss": 0.793,
+      "step": 14582
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.805534125392172e-05,
+      "loss": 0.8326,
+      "step": 14583
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.8052769634315693e-05,
+      "loss": 0.438,
+      "step": 14584
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.8050198014709666e-05,
+      "loss": 0.7139,
+      "step": 14585
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.8047626395103635e-05,
+      "loss": 0.5103,
+      "step": 14586
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.804505477549761e-05,
+      "loss": 0.6487,
+      "step": 14587
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.804248315589158e-05,
+      "loss": 0.7261,
+      "step": 14588
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.8039911536285554e-05,
+      "loss": 0.4916,
+      "step": 14589
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.8037339916679524e-05,
+      "loss": 0.8461,
+      "step": 14590
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.8034768297073497e-05,
+      "loss": 0.5023,
+      "step": 14591
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.803219667746747e-05,
+      "loss": 0.3404,
+      "step": 14592
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.802962505786144e-05,
+      "loss": 0.5287,
+      "step": 14593
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.8027053438255413e-05,
+      "loss": 0.6732,
+      "step": 14594
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.8024481818649386e-05,
+      "loss": 0.5857,
+      "step": 14595
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.802191019904336e-05,
+      "loss": 1.0436,
+      "step": 14596
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.8019338579437332e-05,
+      "loss": 0.6946,
+      "step": 14597
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.8016766959831302e-05,
+      "loss": 0.5678,
+      "step": 14598
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.8014195340225275e-05,
+      "loss": 0.7772,
+      "step": 14599
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.8011623720619248e-05,
+      "loss": 0.7593,
+      "step": 14600
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.8009052101013218e-05,
+      "loss": 0.6078,
+      "step": 14601
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.800648048140719e-05,
+      "loss": 1.0377,
+      "step": 14602
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.8003908861801164e-05,
+      "loss": 0.7335,
+      "step": 14603
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.8001337242195134e-05,
+      "loss": 0.5726,
+      "step": 14604
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.799876562258911e-05,
+      "loss": 0.982,
+      "step": 14605
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.799619400298308e-05,
+      "loss": 0.577,
+      "step": 14606
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.7993622383377053e-05,
+      "loss": 0.7742,
+      "step": 14607
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.7991050763771026e-05,
+      "loss": 0.6962,
+      "step": 14608
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.7988479144164996e-05,
+      "loss": 0.8535,
+      "step": 14609
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.798590752455897e-05,
+      "loss": 1.0228,
+      "step": 14610
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.7983335904952942e-05,
+      "loss": 0.7073,
+      "step": 14611
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.798076428534691e-05,
+      "loss": 0.381,
+      "step": 14612
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.7978192665740885e-05,
+      "loss": 0.4223,
+      "step": 14613
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.7975621046134858e-05,
+      "loss": 0.6429,
+      "step": 14614
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.797304942652883e-05,
+      "loss": 0.8174,
+      "step": 14615
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.79704778069228e-05,
+      "loss": 0.8134,
+      "step": 14616
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.7967906187316773e-05,
+      "loss": 0.769,
+      "step": 14617
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.7965334567710747e-05,
+      "loss": 0.9644,
+      "step": 14618
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.7962762948104716e-05,
+      "loss": 0.7325,
+      "step": 14619
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.796019132849869e-05,
+      "loss": 0.7906,
+      "step": 14620
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.7957619708892662e-05,
+      "loss": 0.6107,
+      "step": 14621
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.7955048089286632e-05,
+      "loss": 0.4795,
+      "step": 14622
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.795247646968061e-05,
+      "loss": 0.4159,
+      "step": 14623
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.7949904850074578e-05,
+      "loss": 0.5748,
+      "step": 14624
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.7947333230468548e-05,
+      "loss": 0.6884,
+      "step": 14625
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.7944761610862524e-05,
+      "loss": 0.7515,
+      "step": 14626
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.7942189991256494e-05,
+      "loss": 0.4957,
+      "step": 14627
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.7939618371650467e-05,
+      "loss": 0.9997,
+      "step": 14628
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.793704675204444e-05,
+      "loss": 0.5211,
+      "step": 14629
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.793447513243841e-05,
+      "loss": 0.8654,
+      "step": 14630
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.7931903512832383e-05,
+      "loss": 0.8905,
+      "step": 14631
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.7929331893226356e-05,
+      "loss": 0.8828,
+      "step": 14632
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.7926760273620326e-05,
+      "loss": 0.5723,
+      "step": 14633
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.79241886540143e-05,
+      "loss": 0.6479,
+      "step": 14634
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.7921617034408272e-05,
+      "loss": 0.6209,
+      "step": 14635
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.7919045414802245e-05,
+      "loss": 0.6653,
+      "step": 14636
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.7916473795196215e-05,
+      "loss": 0.5826,
+      "step": 14637
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.7913902175590188e-05,
+      "loss": 0.9143,
+      "step": 14638
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.791133055598416e-05,
+      "loss": 0.9822,
+      "step": 14639
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.790875893637813e-05,
+      "loss": 0.547,
+      "step": 14640
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.7906187316772103e-05,
+      "loss": 0.5764,
+      "step": 14641
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.7903615697166077e-05,
+      "loss": 0.6487,
+      "step": 14642
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.7901044077560046e-05,
+      "loss": 0.7517,
+      "step": 14643
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.7898472457954023e-05,
+      "loss": 0.7513,
+      "step": 14644
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.7895900838347992e-05,
+      "loss": 0.4583,
+      "step": 14645
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.7893329218741965e-05,
+      "loss": 0.8752,
+      "step": 14646
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.789075759913594e-05,
+      "loss": 0.9121,
+      "step": 14647
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.7888185979529908e-05,
+      "loss": 0.7611,
+      "step": 14648
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.788561435992388e-05,
+      "loss": 0.5822,
+      "step": 14649
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.7883042740317854e-05,
+      "loss": 0.6548,
+      "step": 14650
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.7880471120711824e-05,
+      "loss": 0.8715,
+      "step": 14651
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.7877899501105797e-05,
+      "loss": 0.5881,
+      "step": 14652
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.787532788149977e-05,
+      "loss": 0.4936,
+      "step": 14653
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.7872756261893743e-05,
+      "loss": 0.7227,
+      "step": 14654
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.7870184642287713e-05,
+      "loss": 0.5556,
+      "step": 14655
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.7867613022681686e-05,
+      "loss": 0.7122,
+      "step": 14656
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.786504140307566e-05,
+      "loss": 0.9398,
+      "step": 14657
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.786246978346963e-05,
+      "loss": 0.6343,
+      "step": 14658
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.7859898163863602e-05,
+      "loss": 0.2919,
+      "step": 14659
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.7857326544257575e-05,
+      "loss": 0.6642,
+      "step": 14660
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.7854754924651545e-05,
+      "loss": 0.4567,
+      "step": 14661
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.785218330504552e-05,
+      "loss": 0.8495,
+      "step": 14662
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.784961168543949e-05,
+      "loss": 0.6494,
+      "step": 14663
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.784704006583346e-05,
+      "loss": 0.744,
+      "step": 14664
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.7844468446227437e-05,
+      "loss": 1.0737,
+      "step": 14665
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.7841896826621407e-05,
+      "loss": 0.9209,
+      "step": 14666
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.783932520701538e-05,
+      "loss": 0.5103,
+      "step": 14667
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.7836753587409353e-05,
+      "loss": 0.7327,
+      "step": 14668
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.7834181967803322e-05,
+      "loss": 0.4358,
+      "step": 14669
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.7831610348197296e-05,
+      "loss": 0.4563,
+      "step": 14670
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.782903872859127e-05,
+      "loss": 1.0751,
+      "step": 14671
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.7826467108985238e-05,
+      "loss": 0.6193,
+      "step": 14672
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.782389548937921e-05,
+      "loss": 0.6038,
+      "step": 14673
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.7821323869773184e-05,
+      "loss": 0.6654,
+      "step": 14674
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.7818752250167157e-05,
+      "loss": 0.8611,
+      "step": 14675
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.7816180630561127e-05,
+      "loss": 0.3869,
+      "step": 14676
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.78136090109551e-05,
+      "loss": 0.4118,
+      "step": 14677
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.7811037391349073e-05,
+      "loss": 0.4534,
+      "step": 14678
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.7808465771743043e-05,
+      "loss": 0.8773,
+      "step": 14679
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.7805894152137016e-05,
+      "loss": 0.5483,
+      "step": 14680
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.780332253253099e-05,
+      "loss": 0.6854,
+      "step": 14681
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.780075091292496e-05,
+      "loss": 0.7747,
+      "step": 14682
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.7798179293318935e-05,
+      "loss": 0.4074,
+      "step": 14683
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.7795607673712905e-05,
+      "loss": 0.8647,
+      "step": 14684
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.7793036054106875e-05,
+      "loss": 0.9585,
+      "step": 14685
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.779046443450085e-05,
+      "loss": 1.0498,
+      "step": 14686
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.778789281489482e-05,
+      "loss": 0.6897,
+      "step": 14687
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.7785321195288794e-05,
+      "loss": 0.7459,
+      "step": 14688
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.7782749575682767e-05,
+      "loss": 0.6741,
+      "step": 14689
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.7780177956076737e-05,
+      "loss": 0.8485,
+      "step": 14690
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.777760633647071e-05,
+      "loss": 0.9794,
+      "step": 14691
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.7775034716864683e-05,
+      "loss": 0.7422,
+      "step": 14692
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.7772463097258656e-05,
+      "loss": 0.6161,
+      "step": 14693
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.7769891477652626e-05,
+      "loss": 0.6403,
+      "step": 14694
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.77673198580466e-05,
+      "loss": 0.788,
+      "step": 14695
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.7764748238440572e-05,
+      "loss": 1.0276,
+      "step": 14696
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.776217661883454e-05,
+      "loss": 0.9066,
+      "step": 14697
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.7759604999228514e-05,
+      "loss": 0.4931,
+      "step": 14698
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.7757033379622488e-05,
+      "loss": 0.7643,
+      "step": 14699
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.7754461760016457e-05,
+      "loss": 0.8386,
+      "step": 14700
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.7751890140410434e-05,
+      "loss": 0.8359,
+      "step": 14701
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.7749318520804403e-05,
+      "loss": 1.0865,
+      "step": 14702
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.7746746901198373e-05,
+      "loss": 0.6133,
+      "step": 14703
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.774417528159235e-05,
+      "loss": 0.5456,
+      "step": 14704
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.774160366198632e-05,
+      "loss": 0.7167,
+      "step": 14705
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.7739032042380292e-05,
+      "loss": 0.5942,
+      "step": 14706
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.7736460422774265e-05,
+      "loss": 0.5979,
+      "step": 14707
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.7733888803168235e-05,
+      "loss": 1.0844,
+      "step": 14708
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.7731317183562208e-05,
+      "loss": 0.9623,
+      "step": 14709
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.772874556395618e-05,
+      "loss": 0.6138,
+      "step": 14710
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.772617394435015e-05,
+      "loss": 0.783,
+      "step": 14711
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.7723602324744124e-05,
+      "loss": 0.4274,
+      "step": 14712
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.7721030705138097e-05,
+      "loss": 0.625,
+      "step": 14713
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.771845908553207e-05,
+      "loss": 0.564,
+      "step": 14714
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.771588746592604e-05,
+      "loss": 0.7126,
+      "step": 14715
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.7713315846320013e-05,
+      "loss": 0.8669,
+      "step": 14716
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.7710744226713986e-05,
+      "loss": 0.5835,
+      "step": 14717
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.7708172607107956e-05,
+      "loss": 0.6408,
+      "step": 14718
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.770560098750193e-05,
+      "loss": 0.8459,
+      "step": 14719
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.7703029367895902e-05,
+      "loss": 0.4389,
+      "step": 14720
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.770045774828987e-05,
+      "loss": 0.7367,
+      "step": 14721
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.7697886128683848e-05,
+      "loss": 0.7652,
+      "step": 14722
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.7695314509077818e-05,
+      "loss": 0.8769,
+      "step": 14723
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.7692742889471787e-05,
+      "loss": 1.0258,
+      "step": 14724
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.7690171269865764e-05,
+      "loss": 0.4841,
+      "step": 14725
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.7687599650259733e-05,
+      "loss": 0.6809,
+      "step": 14726
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.7685028030653706e-05,
+      "loss": 0.9304,
+      "step": 14727
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.768245641104768e-05,
+      "loss": 0.7375,
+      "step": 14728
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.767988479144165e-05,
+      "loss": 0.7079,
+      "step": 14729
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.7677313171835622e-05,
+      "loss": 0.7955,
+      "step": 14730
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.7674741552229595e-05,
+      "loss": 0.4089,
+      "step": 14731
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.767216993262357e-05,
+      "loss": 0.3759,
+      "step": 14732
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.7669598313017538e-05,
+      "loss": 0.7072,
+      "step": 14733
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.766702669341151e-05,
+      "loss": 0.6892,
+      "step": 14734
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.7664455073805484e-05,
+      "loss": 0.7929,
+      "step": 14735
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.7661883454199454e-05,
+      "loss": 0.7219,
+      "step": 14736
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.7659311834593427e-05,
+      "loss": 0.7791,
+      "step": 14737
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.76567402149874e-05,
+      "loss": 0.7836,
+      "step": 14738
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.765416859538137e-05,
+      "loss": 0.5619,
+      "step": 14739
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.7651596975775346e-05,
+      "loss": 0.6111,
+      "step": 14740
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.7649025356169316e-05,
+      "loss": 1.0157,
+      "step": 14741
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.7646453736563286e-05,
+      "loss": 0.459,
+      "step": 14742
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.7643882116957262e-05,
+      "loss": 0.6118,
+      "step": 14743
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.7641310497351232e-05,
+      "loss": 0.6751,
+      "step": 14744
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.7638738877745205e-05,
+      "loss": 0.6408,
+      "step": 14745
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.7636167258139178e-05,
+      "loss": 0.6067,
+      "step": 14746
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.7633595638533148e-05,
+      "loss": 0.5585,
+      "step": 14747
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.7631024018927124e-05,
+      "loss": 0.7498,
+      "step": 14748
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.7628452399321094e-05,
+      "loss": 0.8088,
+      "step": 14749
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.7625880779715063e-05,
+      "loss": 0.6979,
+      "step": 14750
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.762330916010904e-05,
+      "loss": 0.6452,
+      "step": 14751
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.762073754050301e-05,
+      "loss": 0.494,
+      "step": 14752
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.7618165920896983e-05,
+      "loss": 0.7237,
+      "step": 14753
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.7615594301290956e-05,
+      "loss": 0.4708,
+      "step": 14754
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.7613022681684925e-05,
+      "loss": 0.8485,
+      "step": 14755
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.76104510620789e-05,
+      "loss": 0.6548,
+      "step": 14756
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.760787944247287e-05,
+      "loss": 0.7875,
+      "step": 14757
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.760530782286684e-05,
+      "loss": 0.8541,
+      "step": 14758
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.7602736203260814e-05,
+      "loss": 0.8635,
+      "step": 14759
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.7600164583654787e-05,
+      "loss": 0.3678,
+      "step": 14760
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.759759296404876e-05,
+      "loss": 1.0072,
+      "step": 14761
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.759502134444273e-05,
+      "loss": 0.7743,
+      "step": 14762
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.7592449724836703e-05,
+      "loss": 0.5736,
+      "step": 14763
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.7589878105230676e-05,
+      "loss": 0.6689,
+      "step": 14764
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.7587306485624646e-05,
+      "loss": 0.3648,
+      "step": 14765
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.758473486601862e-05,
+      "loss": 0.7026,
+      "step": 14766
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.7582163246412592e-05,
+      "loss": 0.4253,
+      "step": 14767
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.7579591626806562e-05,
+      "loss": 0.4364,
+      "step": 14768
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.7577020007200538e-05,
+      "loss": 0.4088,
+      "step": 14769
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.7574448387594508e-05,
+      "loss": 1.0696,
+      "step": 14770
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.757187676798848e-05,
+      "loss": 0.7182,
+      "step": 14771
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.7569305148382454e-05,
+      "loss": 0.973,
+      "step": 14772
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.7566733528776424e-05,
+      "loss": 0.8333,
+      "step": 14773
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.7564161909170397e-05,
+      "loss": 0.4523,
+      "step": 14774
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.756159028956437e-05,
+      "loss": 0.7646,
+      "step": 14775
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.755901866995834e-05,
+      "loss": 0.8473,
+      "step": 14776
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.7556447050352313e-05,
+      "loss": 0.6568,
+      "step": 14777
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.7553875430746286e-05,
+      "loss": 0.6851,
+      "step": 14778
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.755130381114026e-05,
+      "loss": 0.7867,
+      "step": 14779
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.754873219153423e-05,
+      "loss": 0.762,
+      "step": 14780
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.75461605719282e-05,
+      "loss": 0.8937,
+      "step": 14781
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.7543588952322175e-05,
+      "loss": 0.5878,
+      "step": 14782
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.7541017332716144e-05,
+      "loss": 0.4458,
+      "step": 14783
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.7538445713110117e-05,
+      "loss": 1.0173,
+      "step": 14784
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.753587409350409e-05,
+      "loss": 0.9308,
+      "step": 14785
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.753330247389806e-05,
+      "loss": 0.7327,
+      "step": 14786
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.7530730854292037e-05,
+      "loss": 0.4547,
+      "step": 14787
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.7528159234686006e-05,
+      "loss": 0.7539,
+      "step": 14788
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.7525587615079976e-05,
+      "loss": 0.7847,
+      "step": 14789
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.7523015995473953e-05,
+      "loss": 0.7258,
+      "step": 14790
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.7520444375867922e-05,
+      "loss": 0.5392,
+      "step": 14791
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.7517872756261895e-05,
+      "loss": 0.4079,
+      "step": 14792
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.751530113665587e-05,
+      "loss": 0.5376,
+      "step": 14793
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.7512729517049838e-05,
+      "loss": 0.7633,
+      "step": 14794
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.751015789744381e-05,
+      "loss": 0.7721,
+      "step": 14795
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.7507586277837784e-05,
+      "loss": 0.9289,
+      "step": 14796
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.7505014658231754e-05,
+      "loss": 0.7284,
+      "step": 14797
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.7502443038625727e-05,
+      "loss": 0.7619,
+      "step": 14798
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.74998714190197e-05,
+      "loss": 0.7919,
+      "step": 14799
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.7497299799413673e-05,
+      "loss": 0.9027,
+      "step": 14800
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.7494728179807643e-05,
+      "loss": 0.7086,
+      "step": 14801
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.7492156560201616e-05,
+      "loss": 1.1104,
+      "step": 14802
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.748958494059559e-05,
+      "loss": 0.7939,
+      "step": 14803
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.748701332098956e-05,
+      "loss": 0.8623,
+      "step": 14804
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.748444170138353e-05,
+      "loss": 0.9092,
+      "step": 14805
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.7481870081777505e-05,
+      "loss": 0.8932,
+      "step": 14806
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.7479298462171474e-05,
+      "loss": 0.5664,
+      "step": 14807
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.747672684256545e-05,
+      "loss": 0.757,
+      "step": 14808
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.747415522295942e-05,
+      "loss": 0.7478,
+      "step": 14809
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.747158360335339e-05,
+      "loss": 0.6713,
+      "step": 14810
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.7469011983747367e-05,
+      "loss": 0.5356,
+      "step": 14811
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.7466440364141336e-05,
+      "loss": 0.6005,
+      "step": 14812
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.746386874453531e-05,
+      "loss": 0.666,
+      "step": 14813
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.7461297124929283e-05,
+      "loss": 0.6149,
+      "step": 14814
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.7458725505323252e-05,
+      "loss": 0.6444,
+      "step": 14815
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.7456153885717225e-05,
+      "loss": 0.9977,
+      "step": 14816
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.74535822661112e-05,
+      "loss": 0.6152,
+      "step": 14817
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.745101064650517e-05,
+      "loss": 0.678,
+      "step": 14818
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.744843902689914e-05,
+      "loss": 0.5837,
+      "step": 14819
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.7445867407293114e-05,
+      "loss": 0.5367,
+      "step": 14820
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.7443295787687087e-05,
+      "loss": 0.4419,
+      "step": 14821
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.7440724168081057e-05,
+      "loss": 0.6644,
+      "step": 14822
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.743815254847503e-05,
+      "loss": 0.7514,
+      "step": 14823
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.7435580928869003e-05,
+      "loss": 0.3551,
+      "step": 14824
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.7433009309262973e-05,
+      "loss": 0.5339,
+      "step": 14825
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.743043768965695e-05,
+      "loss": 0.516,
+      "step": 14826
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.742786607005092e-05,
+      "loss": 0.9062,
+      "step": 14827
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.742529445044489e-05,
+      "loss": 0.7984,
+      "step": 14828
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.7422722830838865e-05,
+      "loss": 0.6166,
+      "step": 14829
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.7420151211232835e-05,
+      "loss": 0.4941,
+      "step": 14830
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.7417579591626808e-05,
+      "loss": 0.6067,
+      "step": 14831
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.741500797202078e-05,
+      "loss": 0.6518,
+      "step": 14832
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.741243635241475e-05,
+      "loss": 0.4774,
+      "step": 14833
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.7409864732808724e-05,
+      "loss": 0.4943,
+      "step": 14834
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.7407293113202697e-05,
+      "loss": 1.0729,
+      "step": 14835
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.7404721493596666e-05,
+      "loss": 0.7374,
+      "step": 14836
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.740214987399064e-05,
+      "loss": 0.5462,
+      "step": 14837
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.7399578254384613e-05,
+      "loss": 0.754,
+      "step": 14838
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.7397006634778586e-05,
+      "loss": 0.4848,
+      "step": 14839
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.7394435015172555e-05,
+      "loss": 1.0443,
+      "step": 14840
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.739186339556653e-05,
+      "loss": 0.8629,
+      "step": 14841
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.73892917759605e-05,
+      "loss": 1.0052,
+      "step": 14842
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.738672015635447e-05,
+      "loss": 0.7406,
+      "step": 14843
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.7384148536748444e-05,
+      "loss": 0.5087,
+      "step": 14844
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.7381576917142417e-05,
+      "loss": 0.6982,
+      "step": 14845
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.7379005297536387e-05,
+      "loss": 0.6155,
+      "step": 14846
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.7376433677930363e-05,
+      "loss": 0.8333,
+      "step": 14847
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.7373862058324333e-05,
+      "loss": 0.7602,
+      "step": 14848
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.7371290438718303e-05,
+      "loss": 0.6481,
+      "step": 14849
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.736871881911228e-05,
+      "loss": 0.3845,
+      "step": 14850
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.736614719950625e-05,
+      "loss": 0.8727,
+      "step": 14851
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.7363575579900222e-05,
+      "loss": 0.2779,
+      "step": 14852
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.7361003960294195e-05,
+      "loss": 0.6958,
+      "step": 14853
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.7358432340688165e-05,
+      "loss": 0.7269,
+      "step": 14854
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.7355860721082138e-05,
+      "loss": 0.6195,
+      "step": 14855
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.735328910147611e-05,
+      "loss": 0.8992,
+      "step": 14856
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.7350717481870084e-05,
+      "loss": 0.7463,
+      "step": 14857
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.7348145862264054e-05,
+      "loss": 0.6062,
+      "step": 14858
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.7345574242658027e-05,
+      "loss": 0.5011,
+      "step": 14859
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.7343002623052e-05,
+      "loss": 0.5787,
+      "step": 14860
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.734043100344597e-05,
+      "loss": 0.5303,
+      "step": 14861
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.7337859383839943e-05,
+      "loss": 1.0118,
+      "step": 14862
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.7335287764233916e-05,
+      "loss": 0.3764,
+      "step": 14863
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.7332716144627885e-05,
+      "loss": 0.511,
+      "step": 14864
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.7330144525021862e-05,
+      "loss": 0.988,
+      "step": 14865
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.732757290541583e-05,
+      "loss": 0.6634,
+      "step": 14866
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.73250012858098e-05,
+      "loss": 0.5976,
+      "step": 14867
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.7322429666203778e-05,
+      "loss": 0.3965,
+      "step": 14868
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.7319858046597747e-05,
+      "loss": 0.5845,
+      "step": 14869
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.731728642699172e-05,
+      "loss": 0.6554,
+      "step": 14870
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.7314714807385694e-05,
+      "loss": 0.6097,
+      "step": 14871
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.7312143187779663e-05,
+      "loss": 0.9161,
+      "step": 14872
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.7309571568173636e-05,
+      "loss": 0.6706,
+      "step": 14873
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.730699994856761e-05,
+      "loss": 0.6996,
+      "step": 14874
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.730442832896158e-05,
+      "loss": 0.7587,
+      "step": 14875
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.7301856709355552e-05,
+      "loss": 0.4815,
+      "step": 14876
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.7299285089749525e-05,
+      "loss": 0.7068,
+      "step": 14877
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.7296713470143498e-05,
+      "loss": 0.7708,
+      "step": 14878
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.7294141850537468e-05,
+      "loss": 0.6604,
+      "step": 14879
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.729157023093144e-05,
+      "loss": 0.777,
+      "step": 14880
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.7288998611325414e-05,
+      "loss": 0.4454,
+      "step": 14881
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.7286426991719384e-05,
+      "loss": 0.8907,
+      "step": 14882
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.7283855372113357e-05,
+      "loss": 0.9363,
+      "step": 14883
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.728128375250733e-05,
+      "loss": 0.5049,
+      "step": 14884
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.72787121329013e-05,
+      "loss": 0.7126,
+      "step": 14885
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.7276140513295276e-05,
+      "loss": 0.7486,
+      "step": 14886
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.7273568893689246e-05,
+      "loss": 0.5978,
+      "step": 14887
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.7270997274083215e-05,
+      "loss": 0.9043,
+      "step": 14888
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.7268425654477192e-05,
+      "loss": 0.6248,
+      "step": 14889
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.726585403487116e-05,
+      "loss": 0.5282,
+      "step": 14890
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.7263282415265135e-05,
+      "loss": 0.9799,
+      "step": 14891
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.7260710795659108e-05,
+      "loss": 0.7697,
+      "step": 14892
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.7258139176053077e-05,
+      "loss": 0.5962,
+      "step": 14893
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.725556755644705e-05,
+      "loss": 1.0261,
+      "step": 14894
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.7252995936841024e-05,
+      "loss": 0.7079,
+      "step": 14895
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.7250424317234997e-05,
+      "loss": 0.6444,
+      "step": 14896
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.724785269762897e-05,
+      "loss": 0.8155,
+      "step": 14897
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.724528107802294e-05,
+      "loss": 0.7822,
+      "step": 14898
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.7242709458416912e-05,
+      "loss": 0.6518,
+      "step": 14899
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.7240137838810886e-05,
+      "loss": 0.5331,
+      "step": 14900
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.7237566219204855e-05,
+      "loss": 0.6204,
+      "step": 14901
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.7234994599598828e-05,
+      "loss": 1.0196,
+      "step": 14902
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.72324229799928e-05,
+      "loss": 0.7798,
+      "step": 14903
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.7229851360386774e-05,
+      "loss": 0.8026,
+      "step": 14904
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.7227279740780744e-05,
+      "loss": 0.6394,
+      "step": 14905
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.7224708121174717e-05,
+      "loss": 0.8586,
+      "step": 14906
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.722213650156869e-05,
+      "loss": 0.7289,
+      "step": 14907
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.721956488196266e-05,
+      "loss": 0.7944,
+      "step": 14908
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.7216993262356633e-05,
+      "loss": 0.8737,
+      "step": 14909
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.7214421642750606e-05,
+      "loss": 1.0103,
+      "step": 14910
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.7211850023144576e-05,
+      "loss": 0.7089,
+      "step": 14911
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.7209278403538552e-05,
+      "loss": 1.0167,
+      "step": 14912
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.7206706783932522e-05,
+      "loss": 0.5654,
+      "step": 14913
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.720413516432649e-05,
+      "loss": 0.5135,
+      "step": 14914
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.7201563544720468e-05,
+      "loss": 0.691,
+      "step": 14915
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.7198991925114438e-05,
+      "loss": 0.737,
+      "step": 14916
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.719642030550841e-05,
+      "loss": 0.9956,
+      "step": 14917
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.7193848685902384e-05,
+      "loss": 0.6702,
+      "step": 14918
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.7191277066296354e-05,
+      "loss": 0.6701,
+      "step": 14919
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.7188705446690327e-05,
+      "loss": 0.8445,
+      "step": 14920
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.71861338270843e-05,
+      "loss": 1.091,
+      "step": 14921
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.718356220747827e-05,
+      "loss": 0.8406,
+      "step": 14922
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.7180990587872243e-05,
+      "loss": 0.5782,
+      "step": 14923
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.7178418968266216e-05,
+      "loss": 0.4656,
+      "step": 14924
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.717584734866019e-05,
+      "loss": 0.6449,
+      "step": 14925
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.717327572905416e-05,
+      "loss": 0.7157,
+      "step": 14926
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.717070410944813e-05,
+      "loss": 0.65,
+      "step": 14927
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.7168132489842104e-05,
+      "loss": 0.7053,
+      "step": 14928
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.7165560870236074e-05,
+      "loss": 0.7718,
+      "step": 14929
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.7162989250630047e-05,
+      "loss": 0.7593,
+      "step": 14930
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.716041763102402e-05,
+      "loss": 0.5579,
+      "step": 14931
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.715784601141799e-05,
+      "loss": 0.7642,
+      "step": 14932
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.7155274391811966e-05,
+      "loss": 0.609,
+      "step": 14933
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.7152702772205936e-05,
+      "loss": 0.6012,
+      "step": 14934
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.715013115259991e-05,
+      "loss": 0.7673,
+      "step": 14935
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.7147559532993882e-05,
+      "loss": 0.7141,
+      "step": 14936
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.7144987913387852e-05,
+      "loss": 1.0126,
+      "step": 14937
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.7142416293781825e-05,
+      "loss": 0.6307,
+      "step": 14938
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.7139844674175798e-05,
+      "loss": 0.5997,
+      "step": 14939
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.7137273054569768e-05,
+      "loss": 0.5583,
+      "step": 14940
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.713470143496374e-05,
+      "loss": 0.477,
+      "step": 14941
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.7132129815357714e-05,
+      "loss": 0.7659,
+      "step": 14942
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.7129558195751687e-05,
+      "loss": 0.9467,
+      "step": 14943
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.7126986576145657e-05,
+      "loss": 0.6677,
+      "step": 14944
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.712441495653963e-05,
+      "loss": 0.7745,
+      "step": 14945
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.7121843336933603e-05,
+      "loss": 0.8368,
+      "step": 14946
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.7119271717327573e-05,
+      "loss": 0.7311,
+      "step": 14947
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.7116700097721546e-05,
+      "loss": 0.7143,
+      "step": 14948
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.711412847811552e-05,
+      "loss": 0.6551,
+      "step": 14949
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.711155685850949e-05,
+      "loss": 0.5506,
+      "step": 14950
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.7108985238903465e-05,
+      "loss": 0.7436,
+      "step": 14951
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.7106413619297435e-05,
+      "loss": 1.1139,
+      "step": 14952
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.7103841999691404e-05,
+      "loss": 0.8258,
+      "step": 14953
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.710127038008538e-05,
+      "loss": 0.6912,
+      "step": 14954
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.709869876047935e-05,
+      "loss": 0.8331,
+      "step": 14955
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.7096127140873323e-05,
+      "loss": 0.6408,
+      "step": 14956
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.7093555521267297e-05,
+      "loss": 0.5186,
+      "step": 14957
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.7090983901661266e-05,
+      "loss": 0.436,
+      "step": 14958
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.708841228205524e-05,
+      "loss": 0.707,
+      "step": 14959
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.7085840662449212e-05,
+      "loss": 0.7018,
+      "step": 14960
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.7083269042843182e-05,
+      "loss": 0.9321,
+      "step": 14961
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.7080697423237155e-05,
+      "loss": 0.8643,
+      "step": 14962
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.7078125803631128e-05,
+      "loss": 0.5866,
+      "step": 14963
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.70755541840251e-05,
+      "loss": 0.9294,
+      "step": 14964
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.707298256441907e-05,
+      "loss": 0.7924,
+      "step": 14965
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.7070410944813044e-05,
+      "loss": 0.8146,
+      "step": 14966
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.7067839325207017e-05,
+      "loss": 0.8272,
+      "step": 14967
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.7065267705600987e-05,
+      "loss": 0.9389,
+      "step": 14968
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.706269608599496e-05,
+      "loss": 0.4476,
+      "step": 14969
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.7060124466388933e-05,
+      "loss": 0.8037,
+      "step": 14970
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.7057552846782903e-05,
+      "loss": 0.8415,
+      "step": 14971
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.705498122717688e-05,
+      "loss": 0.5377,
+      "step": 14972
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.705240960757085e-05,
+      "loss": 0.7143,
+      "step": 14973
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.704983798796482e-05,
+      "loss": 0.7107,
+      "step": 14974
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.7047266368358795e-05,
+      "loss": 0.5571,
+      "step": 14975
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.7044694748752765e-05,
+      "loss": 0.6,
+      "step": 14976
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.7042123129146738e-05,
+      "loss": 1.0345,
+      "step": 14977
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.703955150954071e-05,
+      "loss": 0.7386,
+      "step": 14978
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.703697988993468e-05,
+      "loss": 0.6417,
+      "step": 14979
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.7034408270328653e-05,
+      "loss": 0.9285,
+      "step": 14980
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.7031836650722627e-05,
+      "loss": 0.3742,
+      "step": 14981
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.70292650311166e-05,
+      "loss": 0.4665,
+      "step": 14982
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.702669341151057e-05,
+      "loss": 0.6088,
+      "step": 14983
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.7024121791904542e-05,
+      "loss": 0.4471,
+      "step": 14984
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.7021550172298515e-05,
+      "loss": 0.7089,
+      "step": 14985
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.7018978552692485e-05,
+      "loss": 0.6573,
+      "step": 14986
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.7016406933086458e-05,
+      "loss": 0.657,
+      "step": 14987
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.701383531348043e-05,
+      "loss": 0.4724,
+      "step": 14988
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.70112636938744e-05,
+      "loss": 0.6632,
+      "step": 14989
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.7008692074268377e-05,
+      "loss": 0.7489,
+      "step": 14990
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.7006120454662347e-05,
+      "loss": 0.6434,
+      "step": 14991
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.7003548835056317e-05,
+      "loss": 0.4703,
+      "step": 14992
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.7000977215450293e-05,
+      "loss": 0.8109,
+      "step": 14993
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.6998405595844263e-05,
+      "loss": 0.6175,
+      "step": 14994
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.6995833976238236e-05,
+      "loss": 0.7487,
+      "step": 14995
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.699326235663221e-05,
+      "loss": 0.817,
+      "step": 14996
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.699069073702618e-05,
+      "loss": 0.9326,
+      "step": 14997
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.6988119117420152e-05,
+      "loss": 0.5589,
+      "step": 14998
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.6985547497814125e-05,
+      "loss": 0.6484,
+      "step": 14999
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.6982975878208095e-05,
+      "loss": 0.7014,
+      "step": 15000
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.6980404258602068e-05,
+      "loss": 0.965,
+      "step": 15001
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.697783263899604e-05,
+      "loss": 0.7796,
+      "step": 15002
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.6975261019390014e-05,
+      "loss": 0.8012,
+      "step": 15003
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.6972689399783984e-05,
+      "loss": 0.6684,
+      "step": 15004
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.6970117780177957e-05,
+      "loss": 0.8151,
+      "step": 15005
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.696754616057193e-05,
+      "loss": 0.675,
+      "step": 15006
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.69649745409659e-05,
+      "loss": 0.5609,
+      "step": 15007
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.6962402921359872e-05,
+      "loss": 0.6529,
+      "step": 15008
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.6959831301753846e-05,
+      "loss": 0.9947,
+      "step": 15009
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.6957259682147815e-05,
+      "loss": 0.4795,
+      "step": 15010
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.695468806254179e-05,
+      "loss": 0.7709,
+      "step": 15011
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.695211644293576e-05,
+      "loss": 0.8602,
+      "step": 15012
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.694954482332973e-05,
+      "loss": 0.6025,
+      "step": 15013
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.6946973203723707e-05,
+      "loss": 0.5483,
+      "step": 15014
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.6944401584117677e-05,
+      "loss": 0.6688,
+      "step": 15015
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.694182996451165e-05,
+      "loss": 0.6899,
+      "step": 15016
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.6939258344905623e-05,
+      "loss": 0.6066,
+      "step": 15017
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.6936686725299593e-05,
+      "loss": 0.7855,
+      "step": 15018
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.6934115105693566e-05,
+      "loss": 0.3078,
+      "step": 15019
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.693154348608754e-05,
+      "loss": 1.0656,
+      "step": 15020
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.6928971866481512e-05,
+      "loss": 0.7406,
+      "step": 15021
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.6926400246875482e-05,
+      "loss": 0.7085,
+      "step": 15022
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.6923828627269455e-05,
+      "loss": 0.7777,
+      "step": 15023
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.6921257007663428e-05,
+      "loss": 0.5928,
+      "step": 15024
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.6918685388057398e-05,
+      "loss": 0.6001,
+      "step": 15025
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.691611376845137e-05,
+      "loss": 0.8733,
+      "step": 15026
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.6913542148845344e-05,
+      "loss": 0.5543,
+      "step": 15027
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.6910970529239314e-05,
+      "loss": 0.9255,
+      "step": 15028
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.690839890963329e-05,
+      "loss": 0.9162,
+      "step": 15029
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.690582729002726e-05,
+      "loss": 0.6299,
+      "step": 15030
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.690325567042123e-05,
+      "loss": 0.7251,
+      "step": 15031
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.6900684050815206e-05,
+      "loss": 0.6976,
+      "step": 15032
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.6898112431209176e-05,
+      "loss": 0.4707,
+      "step": 15033
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.689554081160315e-05,
+      "loss": 0.3642,
+      "step": 15034
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.6892969191997122e-05,
+      "loss": 0.5645,
+      "step": 15035
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.689039757239109e-05,
+      "loss": 0.4269,
+      "step": 15036
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.6887825952785064e-05,
+      "loss": 0.5281,
+      "step": 15037
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.6885254333179038e-05,
+      "loss": 0.5824,
+      "step": 15038
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.6882682713573007e-05,
+      "loss": 0.7041,
+      "step": 15039
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.688011109396698e-05,
+      "loss": 0.7238,
+      "step": 15040
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.6877539474360953e-05,
+      "loss": 0.7154,
+      "step": 15041
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.6874967854754926e-05,
+      "loss": 0.7257,
+      "step": 15042
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.6872396235148896e-05,
+      "loss": 0.5491,
+      "step": 15043
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.686982461554287e-05,
+      "loss": 0.5749,
+      "step": 15044
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.6867252995936842e-05,
+      "loss": 0.7239,
+      "step": 15045
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.6864681376330812e-05,
+      "loss": 0.9267,
+      "step": 15046
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.6862109756724785e-05,
+      "loss": 0.9982,
+      "step": 15047
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.6859538137118758e-05,
+      "loss": 0.8365,
+      "step": 15048
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.685696651751273e-05,
+      "loss": 0.4262,
+      "step": 15049
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.6854394897906704e-05,
+      "loss": 0.5467,
+      "step": 15050
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.6851823278300674e-05,
+      "loss": 0.8924,
+      "step": 15051
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.6849251658694647e-05,
+      "loss": 0.594,
+      "step": 15052
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.684668003908862e-05,
+      "loss": 0.6802,
+      "step": 15053
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.684410841948259e-05,
+      "loss": 0.5481,
+      "step": 15054
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.6841536799876563e-05,
+      "loss": 1.1797,
+      "step": 15055
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.6838965180270536e-05,
+      "loss": 0.6939,
+      "step": 15056
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.6836393560664506e-05,
+      "loss": 0.5169,
+      "step": 15057
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.6833821941058482e-05,
+      "loss": 0.7192,
+      "step": 15058
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.6831250321452452e-05,
+      "loss": 0.8162,
+      "step": 15059
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.6828678701846425e-05,
+      "loss": 0.5428,
+      "step": 15060
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.6826107082240398e-05,
+      "loss": 0.6293,
+      "step": 15061
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.6823535462634368e-05,
+      "loss": 0.6181,
+      "step": 15062
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.682096384302834e-05,
+      "loss": 0.4892,
+      "step": 15063
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.6818392223422314e-05,
+      "loss": 0.8132,
+      "step": 15064
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.6815820603816283e-05,
+      "loss": 0.7843,
+      "step": 15065
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.6813248984210256e-05,
+      "loss": 0.6518,
+      "step": 15066
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.681067736460423e-05,
+      "loss": 0.6898,
+      "step": 15067
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.6808105744998203e-05,
+      "loss": 0.6726,
+      "step": 15068
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.6805534125392172e-05,
+      "loss": 0.9549,
+      "step": 15069
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.6802962505786145e-05,
+      "loss": 0.5496,
+      "step": 15070
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.680039088618012e-05,
+      "loss": 0.9843,
+      "step": 15071
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.6797819266574088e-05,
+      "loss": 0.4666,
+      "step": 15072
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.679524764696806e-05,
+      "loss": 0.7597,
+      "step": 15073
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.6792676027362034e-05,
+      "loss": 0.8033,
+      "step": 15074
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.6790104407756004e-05,
+      "loss": 0.6913,
+      "step": 15075
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.678753278814998e-05,
+      "loss": 0.4741,
+      "step": 15076
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.678496116854395e-05,
+      "loss": 0.4811,
+      "step": 15077
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.678238954893792e-05,
+      "loss": 0.5514,
+      "step": 15078
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.6779817929331896e-05,
+      "loss": 1.0542,
+      "step": 15079
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.6777246309725866e-05,
+      "loss": 0.5577,
+      "step": 15080
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.677467469011984e-05,
+      "loss": 0.7697,
+      "step": 15081
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.6772103070513812e-05,
+      "loss": 0.76,
+      "step": 15082
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.6769531450907782e-05,
+      "loss": 0.7719,
+      "step": 15083
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.6766959831301755e-05,
+      "loss": 0.6055,
+      "step": 15084
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.6764388211695728e-05,
+      "loss": 0.814,
+      "step": 15085
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.6761816592089698e-05,
+      "loss": 0.7647,
+      "step": 15086
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.675924497248367e-05,
+      "loss": 0.6512,
+      "step": 15087
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.6756673352877644e-05,
+      "loss": 0.7195,
+      "step": 15088
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.6754101733271617e-05,
+      "loss": 0.7311,
+      "step": 15089
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.6751530113665587e-05,
+      "loss": 0.876,
+      "step": 15090
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.674895849405956e-05,
+      "loss": 0.7551,
+      "step": 15091
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.6746386874453533e-05,
+      "loss": 0.4959,
+      "step": 15092
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.6743815254847502e-05,
+      "loss": 0.4835,
+      "step": 15093
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.6741243635241475e-05,
+      "loss": 0.6469,
+      "step": 15094
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.673867201563545e-05,
+      "loss": 0.7427,
+      "step": 15095
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.6736100396029418e-05,
+      "loss": 0.9331,
+      "step": 15096
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.6733528776423395e-05,
+      "loss": 1.1352,
+      "step": 15097
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.6730957156817364e-05,
+      "loss": 0.7927,
+      "step": 15098
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.6728385537211334e-05,
+      "loss": 1.1206,
+      "step": 15099
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.672581391760531e-05,
+      "loss": 0.9258,
+      "step": 15100
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.672324229799928e-05,
+      "loss": 0.5756,
+      "step": 15101
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.6720670678393253e-05,
+      "loss": 0.6497,
+      "step": 15102
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.6718099058787226e-05,
+      "loss": 0.591,
+      "step": 15103
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.6715527439181196e-05,
+      "loss": 0.6967,
+      "step": 15104
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.671295581957517e-05,
+      "loss": 0.8989,
+      "step": 15105
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.6710384199969142e-05,
+      "loss": 0.8168,
+      "step": 15106
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.6707812580363115e-05,
+      "loss": 0.4547,
+      "step": 15107
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.6705240960757085e-05,
+      "loss": 0.6722,
+      "step": 15108
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.6702669341151058e-05,
+      "loss": 0.6847,
+      "step": 15109
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.670009772154503e-05,
+      "loss": 0.9602,
+      "step": 15110
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.6697526101939e-05,
+      "loss": 0.6171,
+      "step": 15111
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.6694954482332974e-05,
+      "loss": 0.4848,
+      "step": 15112
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.6692382862726947e-05,
+      "loss": 0.6039,
+      "step": 15113
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.6689811243120917e-05,
+      "loss": 1.1409,
+      "step": 15114
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.6687239623514893e-05,
+      "loss": 0.6736,
+      "step": 15115
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.6684668003908863e-05,
+      "loss": 0.761,
+      "step": 15116
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.6682096384302832e-05,
+      "loss": 0.789,
+      "step": 15117
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.667952476469681e-05,
+      "loss": 0.6625,
+      "step": 15118
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.667695314509078e-05,
+      "loss": 0.7066,
+      "step": 15119
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.667438152548475e-05,
+      "loss": 0.7048,
+      "step": 15120
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.6671809905878725e-05,
+      "loss": 0.6644,
+      "step": 15121
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.6669238286272694e-05,
+      "loss": 0.8653,
+      "step": 15122
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.6666666666666667e-05,
+      "loss": 0.6462,
+      "step": 15123
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.666409504706064e-05,
+      "loss": 0.5285,
+      "step": 15124
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.666152342745461e-05,
+      "loss": 0.6691,
+      "step": 15125
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.6658951807848583e-05,
+      "loss": 0.6792,
+      "step": 15126
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.6656380188242556e-05,
+      "loss": 0.538,
+      "step": 15127
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.665380856863653e-05,
+      "loss": 0.8073,
+      "step": 15128
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.66512369490305e-05,
+      "loss": 0.5534,
+      "step": 15129
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.6648665329424472e-05,
+      "loss": 0.6545,
+      "step": 15130
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.6646093709818445e-05,
+      "loss": 0.7609,
+      "step": 15131
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.6643522090212415e-05,
+      "loss": 0.5519,
+      "step": 15132
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.6640950470606388e-05,
+      "loss": 0.7183,
+      "step": 15133
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.663837885100036e-05,
+      "loss": 0.5035,
+      "step": 15134
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.663580723139433e-05,
+      "loss": 0.8907,
+      "step": 15135
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.6633235611788307e-05,
+      "loss": 0.632,
+      "step": 15136
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.6630663992182277e-05,
+      "loss": 0.6506,
+      "step": 15137
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.6628092372576247e-05,
+      "loss": 0.3571,
+      "step": 15138
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.6625520752970223e-05,
+      "loss": 0.6889,
+      "step": 15139
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.6622949133364193e-05,
+      "loss": 1.0389,
+      "step": 15140
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.6620377513758166e-05,
+      "loss": 0.7877,
+      "step": 15141
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.661780589415214e-05,
+      "loss": 0.6173,
+      "step": 15142
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.661523427454611e-05,
+      "loss": 0.6174,
+      "step": 15143
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.661266265494008e-05,
+      "loss": 0.406,
+      "step": 15144
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.6610091035334055e-05,
+      "loss": 1.0076,
+      "step": 15145
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.6607519415728028e-05,
+      "loss": 0.8825,
+      "step": 15146
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.6604947796121998e-05,
+      "loss": 0.773,
+      "step": 15147
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.660237617651597e-05,
+      "loss": 0.5448,
+      "step": 15148
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.6599804556909944e-05,
+      "loss": 0.6476,
+      "step": 15149
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.6597232937303913e-05,
+      "loss": 0.7607,
+      "step": 15150
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.6594661317697886e-05,
+      "loss": 0.6416,
+      "step": 15151
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.659208969809186e-05,
+      "loss": 0.8436,
+      "step": 15152
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.658951807848583e-05,
+      "loss": 0.8244,
+      "step": 15153
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.6586946458879806e-05,
+      "loss": 0.9364,
+      "step": 15154
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.6584374839273775e-05,
+      "loss": 0.9349,
+      "step": 15155
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.6581803219667745e-05,
+      "loss": 0.7106,
+      "step": 15156
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.657923160006172e-05,
+      "loss": 0.4849,
+      "step": 15157
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.657665998045569e-05,
+      "loss": 0.4349,
+      "step": 15158
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.6574088360849664e-05,
+      "loss": 0.6674,
+      "step": 15159
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.6571516741243637e-05,
+      "loss": 0.5889,
+      "step": 15160
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.6568945121637607e-05,
+      "loss": 0.5646,
+      "step": 15161
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.656637350203158e-05,
+      "loss": 0.7818,
+      "step": 15162
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.6563801882425553e-05,
+      "loss": 0.7214,
+      "step": 15163
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.6561230262819523e-05,
+      "loss": 0.4542,
+      "step": 15164
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.6558658643213496e-05,
+      "loss": 0.5768,
+      "step": 15165
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.655608702360747e-05,
+      "loss": 0.6468,
+      "step": 15166
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.6553515404001442e-05,
+      "loss": 0.5187,
+      "step": 15167
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.6550943784395412e-05,
+      "loss": 0.7468,
+      "step": 15168
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.6548372164789385e-05,
+      "loss": 0.6334,
+      "step": 15169
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.6545800545183358e-05,
+      "loss": 0.8072,
+      "step": 15170
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.6543228925577328e-05,
+      "loss": 0.5764,
+      "step": 15171
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.65406573059713e-05,
+      "loss": 0.8205,
+      "step": 15172
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.6538085686365274e-05,
+      "loss": 0.6455,
+      "step": 15173
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.6535514066759243e-05,
+      "loss": 0.783,
+      "step": 15174
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.653294244715322e-05,
+      "loss": 0.6428,
+      "step": 15175
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.653037082754719e-05,
+      "loss": 0.6328,
+      "step": 15176
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.652779920794116e-05,
+      "loss": 0.7368,
+      "step": 15177
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.6525227588335136e-05,
+      "loss": 1.144,
+      "step": 15178
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.6522655968729105e-05,
+      "loss": 0.4268,
+      "step": 15179
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.652008434912308e-05,
+      "loss": 0.5064,
+      "step": 15180
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.651751272951705e-05,
+      "loss": 0.4342,
+      "step": 15181
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.651494110991102e-05,
+      "loss": 0.3907,
+      "step": 15182
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.6512369490304994e-05,
+      "loss": 0.8185,
+      "step": 15183
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.6509797870698967e-05,
+      "loss": 0.7433,
+      "step": 15184
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.650722625109294e-05,
+      "loss": 0.7868,
+      "step": 15185
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.650465463148691e-05,
+      "loss": 0.4259,
+      "step": 15186
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.6502083011880883e-05,
+      "loss": 0.9893,
+      "step": 15187
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.6499511392274856e-05,
+      "loss": 0.645,
+      "step": 15188
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.6496939772668826e-05,
+      "loss": 0.6837,
+      "step": 15189
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.64943681530628e-05,
+      "loss": 0.5655,
+      "step": 15190
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.6491796533456772e-05,
+      "loss": 0.7596,
+      "step": 15191
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.6489224913850742e-05,
+      "loss": 0.7459,
+      "step": 15192
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.6486653294244718e-05,
+      "loss": 0.5604,
+      "step": 15193
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.6484081674638688e-05,
+      "loss": 0.4815,
+      "step": 15194
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.6481510055032658e-05,
+      "loss": 0.4816,
+      "step": 15195
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.6478938435426634e-05,
+      "loss": 0.9143,
+      "step": 15196
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.6476366815820604e-05,
+      "loss": 0.8217,
+      "step": 15197
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.6473795196214577e-05,
+      "loss": 0.8651,
+      "step": 15198
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.647122357660855e-05,
+      "loss": 0.4383,
+      "step": 15199
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.646865195700252e-05,
+      "loss": 0.7699,
+      "step": 15200
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.6466080337396496e-05,
+      "loss": 0.6195,
+      "step": 15201
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.6463508717790466e-05,
+      "loss": 0.6378,
+      "step": 15202
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.6460937098184435e-05,
+      "loss": 0.6197,
+      "step": 15203
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.6458365478578412e-05,
+      "loss": 0.5941,
+      "step": 15204
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.645579385897238e-05,
+      "loss": 0.4815,
+      "step": 15205
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.6453222239366355e-05,
+      "loss": 0.4464,
+      "step": 15206
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.6450650619760328e-05,
+      "loss": 0.9185,
+      "step": 15207
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.6448079000154297e-05,
+      "loss": 0.5784,
+      "step": 15208
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.644550738054827e-05,
+      "loss": 0.8149,
+      "step": 15209
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.6442935760942244e-05,
+      "loss": 0.6418,
+      "step": 15210
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.6440364141336213e-05,
+      "loss": 0.4738,
+      "step": 15211
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.6437792521730186e-05,
+      "loss": 0.8632,
+      "step": 15212
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.643522090212416e-05,
+      "loss": 0.5862,
+      "step": 15213
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.6432649282518132e-05,
+      "loss": 0.7102,
+      "step": 15214
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.6430077662912102e-05,
+      "loss": 0.4902,
+      "step": 15215
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.6427506043306075e-05,
+      "loss": 0.5177,
+      "step": 15216
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.6424934423700048e-05,
+      "loss": 0.6924,
+      "step": 15217
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.6422362804094018e-05,
+      "loss": 0.9939,
+      "step": 15218
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.641979118448799e-05,
+      "loss": 0.914,
+      "step": 15219
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.6417219564881964e-05,
+      "loss": 0.5738,
+      "step": 15220
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.6414647945275934e-05,
+      "loss": 0.6693,
+      "step": 15221
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.641207632566991e-05,
+      "loss": 0.6216,
+      "step": 15222
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.640950470606388e-05,
+      "loss": 0.8556,
+      "step": 15223
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.6406933086457853e-05,
+      "loss": 0.5394,
+      "step": 15224
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.6404361466851826e-05,
+      "loss": 0.9454,
+      "step": 15225
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.6401789847245796e-05,
+      "loss": 0.774,
+      "step": 15226
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.639921822763977e-05,
+      "loss": 0.6979,
+      "step": 15227
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.6396646608033742e-05,
+      "loss": 0.581,
+      "step": 15228
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.639407498842771e-05,
+      "loss": 0.9865,
+      "step": 15229
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.6391503368821685e-05,
+      "loss": 0.9738,
+      "step": 15230
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.6388931749215658e-05,
+      "loss": 0.6426,
+      "step": 15231
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.638636012960963e-05,
+      "loss": 0.7168,
+      "step": 15232
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.63837885100036e-05,
+      "loss": 0.5805,
+      "step": 15233
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.6381216890397574e-05,
+      "loss": 0.8141,
+      "step": 15234
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.6378645270791547e-05,
+      "loss": 0.8572,
+      "step": 15235
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.6376073651185516e-05,
+      "loss": 0.8257,
+      "step": 15236
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.637350203157949e-05,
+      "loss": 0.9707,
+      "step": 15237
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.6370930411973462e-05,
+      "loss": 0.7962,
+      "step": 15238
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.6368358792367432e-05,
+      "loss": 0.5133,
+      "step": 15239
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.636578717276141e-05,
+      "loss": 0.7289,
+      "step": 15240
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.636321555315538e-05,
+      "loss": 0.911,
+      "step": 15241
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.6360643933549348e-05,
+      "loss": 1.01,
+      "step": 15242
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.6358072313943324e-05,
+      "loss": 0.4649,
+      "step": 15243
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.6355500694337294e-05,
+      "loss": 0.8136,
+      "step": 15244
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.6352929074731267e-05,
+      "loss": 0.6262,
+      "step": 15245
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.635035745512524e-05,
+      "loss": 0.8265,
+      "step": 15246
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.634778583551921e-05,
+      "loss": 0.6919,
+      "step": 15247
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.6345214215913183e-05,
+      "loss": 0.7677,
+      "step": 15248
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.6342642596307156e-05,
+      "loss": 0.5424,
+      "step": 15249
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.6340070976701126e-05,
+      "loss": 0.6632,
+      "step": 15250
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.63374993570951e-05,
+      "loss": 0.5287,
+      "step": 15251
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.6334927737489072e-05,
+      "loss": 0.903,
+      "step": 15252
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.6332356117883045e-05,
+      "loss": 0.5693,
+      "step": 15253
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.6329784498277015e-05,
+      "loss": 0.7418,
+      "step": 15254
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.6327212878670988e-05,
+      "loss": 0.5944,
+      "step": 15255
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.632464125906496e-05,
+      "loss": 0.4815,
+      "step": 15256
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.632206963945893e-05,
+      "loss": 0.9994,
+      "step": 15257
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.6319498019852904e-05,
+      "loss": 0.3982,
+      "step": 15258
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.6316926400246877e-05,
+      "loss": 0.8091,
+      "step": 15259
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.6314354780640846e-05,
+      "loss": 0.5163,
+      "step": 15260
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.6311783161034823e-05,
+      "loss": 1.2572,
+      "step": 15261
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.6309211541428793e-05,
+      "loss": 0.6982,
+      "step": 15262
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.6306639921822762e-05,
+      "loss": 0.8826,
+      "step": 15263
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.630406830221674e-05,
+      "loss": 0.9058,
+      "step": 15264
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.630149668261071e-05,
+      "loss": 0.8353,
+      "step": 15265
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.629892506300468e-05,
+      "loss": 0.4466,
+      "step": 15266
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.6296353443398655e-05,
+      "loss": 0.8295,
+      "step": 15267
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.6293781823792624e-05,
+      "loss": 0.861,
+      "step": 15268
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.6291210204186597e-05,
+      "loss": 0.7921,
+      "step": 15269
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.628863858458057e-05,
+      "loss": 0.3015,
+      "step": 15270
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.6286066964974543e-05,
+      "loss": 0.3788,
+      "step": 15271
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.6283495345368513e-05,
+      "loss": 0.6215,
+      "step": 15272
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.6280923725762486e-05,
+      "loss": 0.6475,
+      "step": 15273
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.627835210615646e-05,
+      "loss": 0.6657,
+      "step": 15274
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.627578048655043e-05,
+      "loss": 0.4948,
+      "step": 15275
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.6273208866944402e-05,
+      "loss": 0.7969,
+      "step": 15276
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.6270637247338375e-05,
+      "loss": 0.8553,
+      "step": 15277
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.6268065627732345e-05,
+      "loss": 0.9854,
+      "step": 15278
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.626549400812632e-05,
+      "loss": 0.433,
+      "step": 15279
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.626292238852029e-05,
+      "loss": 0.633,
+      "step": 15280
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.626035076891426e-05,
+      "loss": 0.5281,
+      "step": 15281
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.6257779149308237e-05,
+      "loss": 0.6346,
+      "step": 15282
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.6255207529702207e-05,
+      "loss": 0.7485,
+      "step": 15283
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.625263591009618e-05,
+      "loss": 0.7343,
+      "step": 15284
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.6250064290490153e-05,
+      "loss": 0.7353,
+      "step": 15285
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.6247492670884123e-05,
+      "loss": 0.5858,
+      "step": 15286
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.6244921051278096e-05,
+      "loss": 0.8697,
+      "step": 15287
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.624234943167207e-05,
+      "loss": 0.6284,
+      "step": 15288
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.623977781206604e-05,
+      "loss": 0.8866,
+      "step": 15289
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.623720619246001e-05,
+      "loss": 0.7409,
+      "step": 15290
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.6234634572853985e-05,
+      "loss": 0.7115,
+      "step": 15291
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.6232062953247958e-05,
+      "loss": 0.5861,
+      "step": 15292
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.6229491333641927e-05,
+      "loss": 0.3998,
+      "step": 15293
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.62269197140359e-05,
+      "loss": 0.2947,
+      "step": 15294
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.6224348094429873e-05,
+      "loss": 0.488,
+      "step": 15295
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.6221776474823843e-05,
+      "loss": 1.0746,
+      "step": 15296
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.6219204855217816e-05,
+      "loss": 0.9701,
+      "step": 15297
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.621663323561179e-05,
+      "loss": 0.2427,
+      "step": 15298
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.621406161600576e-05,
+      "loss": 0.3926,
+      "step": 15299
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.6211489996399735e-05,
+      "loss": 0.6658,
+      "step": 15300
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.6208918376793705e-05,
+      "loss": 0.7273,
+      "step": 15301
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.6206346757187675e-05,
+      "loss": 0.367,
+      "step": 15302
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.620377513758165e-05,
+      "loss": 0.7876,
+      "step": 15303
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.620120351797562e-05,
+      "loss": 0.6581,
+      "step": 15304
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.6198631898369594e-05,
+      "loss": 0.9757,
+      "step": 15305
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.6196060278763567e-05,
+      "loss": 0.7911,
+      "step": 15306
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.6193488659157537e-05,
+      "loss": 1.002,
+      "step": 15307
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.619091703955151e-05,
+      "loss": 0.7696,
+      "step": 15308
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.6188345419945483e-05,
+      "loss": 0.736,
+      "step": 15309
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.6185773800339456e-05,
+      "loss": 0.7226,
+      "step": 15310
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.6183202180733426e-05,
+      "loss": 0.4857,
+      "step": 15311
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.61806305611274e-05,
+      "loss": 1.0443,
+      "step": 15312
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.6178058941521372e-05,
+      "loss": 0.7326,
+      "step": 15313
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.617548732191534e-05,
+      "loss": 0.5945,
+      "step": 15314
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.6172915702309315e-05,
+      "loss": 0.847,
+      "step": 15315
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.6170344082703288e-05,
+      "loss": 0.6671,
+      "step": 15316
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.6167772463097257e-05,
+      "loss": 0.7285,
+      "step": 15317
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.6165200843491234e-05,
+      "loss": 0.6134,
+      "step": 15318
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.6162629223885204e-05,
+      "loss": 0.8889,
+      "step": 15319
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.6160057604279173e-05,
+      "loss": 0.8955,
+      "step": 15320
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.615748598467315e-05,
+      "loss": 0.7594,
+      "step": 15321
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.615491436506712e-05,
+      "loss": 0.9661,
+      "step": 15322
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.6152342745461092e-05,
+      "loss": 0.6602,
+      "step": 15323
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.6149771125855065e-05,
+      "loss": 0.701,
+      "step": 15324
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.6147199506249035e-05,
+      "loss": 0.9919,
+      "step": 15325
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.6144627886643008e-05,
+      "loss": 0.6168,
+      "step": 15326
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.614205626703698e-05,
+      "loss": 0.7654,
+      "step": 15327
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.613948464743095e-05,
+      "loss": 0.8879,
+      "step": 15328
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.6136913027824924e-05,
+      "loss": 0.6886,
+      "step": 15329
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.6134341408218897e-05,
+      "loss": 0.4424,
+      "step": 15330
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.613176978861287e-05,
+      "loss": 0.6817,
+      "step": 15331
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.612919816900684e-05,
+      "loss": 0.7819,
+      "step": 15332
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.6126626549400813e-05,
+      "loss": 0.4311,
+      "step": 15333
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.6124054929794786e-05,
+      "loss": 0.5936,
+      "step": 15334
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.6121483310188756e-05,
+      "loss": 0.632,
+      "step": 15335
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.611891169058273e-05,
+      "loss": 0.9889,
+      "step": 15336
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.6116340070976702e-05,
+      "loss": 0.865,
+      "step": 15337
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.611376845137067e-05,
+      "loss": 0.5749,
+      "step": 15338
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.6111196831764648e-05,
+      "loss": 0.3354,
+      "step": 15339
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.6108625212158618e-05,
+      "loss": 0.6722,
+      "step": 15340
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.6106053592552587e-05,
+      "loss": 0.5368,
+      "step": 15341
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.6103481972946564e-05,
+      "loss": 0.6786,
+      "step": 15342
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.6100910353340534e-05,
+      "loss": 0.5874,
+      "step": 15343
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.6098338733734507e-05,
+      "loss": 0.822,
+      "step": 15344
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.609576711412848e-05,
+      "loss": 0.5563,
+      "step": 15345
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.609319549452245e-05,
+      "loss": 0.7337,
+      "step": 15346
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.6090623874916422e-05,
+      "loss": 0.6263,
+      "step": 15347
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.6088052255310396e-05,
+      "loss": 0.7548,
+      "step": 15348
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.608548063570437e-05,
+      "loss": 0.5941,
+      "step": 15349
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.608290901609834e-05,
+      "loss": 0.7499,
+      "step": 15350
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.608033739649231e-05,
+      "loss": 0.6127,
+      "step": 15351
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.6077765776886284e-05,
+      "loss": 0.7617,
+      "step": 15352
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.6075194157280258e-05,
+      "loss": 0.5064,
+      "step": 15353
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.6072622537674227e-05,
+      "loss": 0.4497,
+      "step": 15354
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.60700509180682e-05,
+      "loss": 0.7235,
+      "step": 15355
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.6067479298462173e-05,
+      "loss": 0.826,
+      "step": 15356
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.6064907678856146e-05,
+      "loss": 0.6627,
+      "step": 15357
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.6062336059250116e-05,
+      "loss": 0.6644,
+      "step": 15358
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.605976443964409e-05,
+      "loss": 0.5423,
+      "step": 15359
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.6057192820038062e-05,
+      "loss": 0.7672,
+      "step": 15360
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.6054621200432032e-05,
+      "loss": 0.7548,
+      "step": 15361
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.6052049580826005e-05,
+      "loss": 0.7923,
+      "step": 15362
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.6049477961219978e-05,
+      "loss": 0.8882,
+      "step": 15363
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.6046906341613948e-05,
+      "loss": 0.6341,
+      "step": 15364
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.6044334722007924e-05,
+      "loss": 0.775,
+      "step": 15365
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.6041763102401894e-05,
+      "loss": 0.6648,
+      "step": 15366
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.6039191482795864e-05,
+      "loss": 0.8553,
+      "step": 15367
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.603661986318984e-05,
+      "loss": 0.596,
+      "step": 15368
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.603404824358381e-05,
+      "loss": 0.5532,
+      "step": 15369
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.6031476623977783e-05,
+      "loss": 0.8093,
+      "step": 15370
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.6028905004371756e-05,
+      "loss": 0.6922,
+      "step": 15371
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.6026333384765726e-05,
+      "loss": 0.7983,
+      "step": 15372
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.60237617651597e-05,
+      "loss": 0.4467,
+      "step": 15373
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.6021190145553672e-05,
+      "loss": 0.5544,
+      "step": 15374
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.601861852594764e-05,
+      "loss": 0.8453,
+      "step": 15375
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.6016046906341614e-05,
+      "loss": 0.673,
+      "step": 15376
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.6013475286735588e-05,
+      "loss": 0.4099,
+      "step": 15377
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.601090366712956e-05,
+      "loss": 0.5827,
+      "step": 15378
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.600833204752353e-05,
+      "loss": 0.9568,
+      "step": 15379
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.6005760427917503e-05,
+      "loss": 0.5694,
+      "step": 15380
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.6003188808311476e-05,
+      "loss": 0.8572,
+      "step": 15381
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.6000617188705446e-05,
+      "loss": 0.6668,
+      "step": 15382
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.599804556909942e-05,
+      "loss": 0.6822,
+      "step": 15383
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.5995473949493392e-05,
+      "loss": 0.7393,
+      "step": 15384
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.5992902329887362e-05,
+      "loss": 0.6821,
+      "step": 15385
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.599033071028134e-05,
+      "loss": 0.8622,
+      "step": 15386
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.5987759090675308e-05,
+      "loss": 0.8181,
+      "step": 15387
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.5985187471069278e-05,
+      "loss": 0.9543,
+      "step": 15388
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.5982615851463254e-05,
+      "loss": 0.5923,
+      "step": 15389
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.5980044231857224e-05,
+      "loss": 0.8026,
+      "step": 15390
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.5977472612251197e-05,
+      "loss": 0.8585,
+      "step": 15391
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.597490099264517e-05,
+      "loss": 0.481,
+      "step": 15392
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.597232937303914e-05,
+      "loss": 0.8298,
+      "step": 15393
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.5969757753433113e-05,
+      "loss": 0.4875,
+      "step": 15394
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.5967186133827086e-05,
+      "loss": 0.7174,
+      "step": 15395
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.596461451422106e-05,
+      "loss": 0.8548,
+      "step": 15396
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.596204289461503e-05,
+      "loss": 0.7699,
+      "step": 15397
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.5959471275009002e-05,
+      "loss": 0.5965,
+      "step": 15398
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.5956899655402975e-05,
+      "loss": 0.4887,
+      "step": 15399
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.5954328035796945e-05,
+      "loss": 0.7523,
+      "step": 15400
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.5951756416190918e-05,
+      "loss": 0.5095,
+      "step": 15401
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.594918479658489e-05,
+      "loss": 0.6562,
+      "step": 15402
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.594661317697886e-05,
+      "loss": 0.7546,
+      "step": 15403
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.5944041557372837e-05,
+      "loss": 0.9425,
+      "step": 15404
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.5941469937766807e-05,
+      "loss": 0.8464,
+      "step": 15405
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.5938898318160776e-05,
+      "loss": 1.1361,
+      "step": 15406
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.5936326698554753e-05,
+      "loss": 1.0268,
+      "step": 15407
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.5933755078948722e-05,
+      "loss": 0.7494,
+      "step": 15408
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.5931183459342695e-05,
+      "loss": 0.59,
+      "step": 15409
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.592861183973667e-05,
+      "loss": 0.5385,
+      "step": 15410
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.5926040220130638e-05,
+      "loss": 0.6768,
+      "step": 15411
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.592346860052461e-05,
+      "loss": 0.6192,
+      "step": 15412
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.5920896980918584e-05,
+      "loss": 0.4502,
+      "step": 15413
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.5918325361312554e-05,
+      "loss": 0.5939,
+      "step": 15414
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.5915753741706527e-05,
+      "loss": 0.5773,
+      "step": 15415
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.59131821221005e-05,
+      "loss": 0.9882,
+      "step": 15416
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.5910610502494473e-05,
+      "loss": 0.688,
+      "step": 15417
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.5908038882888443e-05,
+      "loss": 0.6446,
+      "step": 15418
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.5905467263282416e-05,
+      "loss": 0.6545,
+      "step": 15419
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.590289564367639e-05,
+      "loss": 0.7164,
+      "step": 15420
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.590032402407036e-05,
+      "loss": 0.5364,
+      "step": 15421
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.5897752404464332e-05,
+      "loss": 0.488,
+      "step": 15422
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.5895180784858305e-05,
+      "loss": 0.7182,
+      "step": 15423
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.5892609165252275e-05,
+      "loss": 0.9817,
+      "step": 15424
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.589003754564625e-05,
+      "loss": 0.5953,
+      "step": 15425
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.588746592604022e-05,
+      "loss": 0.7192,
+      "step": 15426
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.588489430643419e-05,
+      "loss": 0.8535,
+      "step": 15427
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.5882322686828167e-05,
+      "loss": 0.3661,
+      "step": 15428
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.5879751067222137e-05,
+      "loss": 0.6613,
+      "step": 15429
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.587717944761611e-05,
+      "loss": 0.7657,
+      "step": 15430
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.5874607828010083e-05,
+      "loss": 0.7827,
+      "step": 15431
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.5872036208404052e-05,
+      "loss": 0.6158,
+      "step": 15432
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.5869464588798025e-05,
+      "loss": 0.7444,
+      "step": 15433
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.5866892969192e-05,
+      "loss": 0.5171,
+      "step": 15434
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.586432134958597e-05,
+      "loss": 0.6369,
+      "step": 15435
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.586174972997994e-05,
+      "loss": 0.683,
+      "step": 15436
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.5859178110373914e-05,
+      "loss": 1.0338,
+      "step": 15437
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.5856606490767887e-05,
+      "loss": 0.6038,
+      "step": 15438
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.5854034871161857e-05,
+      "loss": 0.9495,
+      "step": 15439
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.585146325155583e-05,
+      "loss": 0.8003,
+      "step": 15440
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.5848891631949803e-05,
+      "loss": 0.3603,
+      "step": 15441
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.5846320012343773e-05,
+      "loss": 0.8944,
+      "step": 15442
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.584374839273775e-05,
+      "loss": 0.8179,
+      "step": 15443
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.584117677313172e-05,
+      "loss": 0.7107,
+      "step": 15444
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.583860515352569e-05,
+      "loss": 0.5002,
+      "step": 15445
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.5836033533919665e-05,
+      "loss": 0.6738,
+      "step": 15446
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.5833461914313635e-05,
+      "loss": 0.6315,
+      "step": 15447
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.5830890294707608e-05,
+      "loss": 0.791,
+      "step": 15448
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.582831867510158e-05,
+      "loss": 0.5481,
+      "step": 15449
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.582574705549555e-05,
+      "loss": 0.7602,
+      "step": 15450
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.5823175435889524e-05,
+      "loss": 0.8444,
+      "step": 15451
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.5820603816283497e-05,
+      "loss": 0.4438,
+      "step": 15452
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.5818032196677467e-05,
+      "loss": 0.7214,
+      "step": 15453
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.581546057707144e-05,
+      "loss": 0.617,
+      "step": 15454
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.5812888957465413e-05,
+      "loss": 0.6385,
+      "step": 15455
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.5810317337859386e-05,
+      "loss": 0.6288,
+      "step": 15456
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.5807745718253355e-05,
+      "loss": 0.82,
+      "step": 15457
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.580517409864733e-05,
+      "loss": 0.8038,
+      "step": 15458
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.58026024790413e-05,
+      "loss": 0.5888,
+      "step": 15459
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.580003085943527e-05,
+      "loss": 0.8008,
+      "step": 15460
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.5797459239829244e-05,
+      "loss": 0.8812,
+      "step": 15461
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.5794887620223217e-05,
+      "loss": 0.7898,
+      "step": 15462
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.5792316000617187e-05,
+      "loss": 0.5826,
+      "step": 15463
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.5789744381011164e-05,
+      "loss": 0.749,
+      "step": 15464
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.5787172761405133e-05,
+      "loss": 0.4946,
+      "step": 15465
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.5784601141799103e-05,
+      "loss": 0.6531,
+      "step": 15466
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.578202952219308e-05,
+      "loss": 0.6957,
+      "step": 15467
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.577945790258705e-05,
+      "loss": 0.4701,
+      "step": 15468
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.5776886282981022e-05,
+      "loss": 0.7051,
+      "step": 15469
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.5774314663374995e-05,
+      "loss": 0.5856,
+      "step": 15470
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.5771743043768965e-05,
+      "loss": 0.719,
+      "step": 15471
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.5769171424162938e-05,
+      "loss": 0.4416,
+      "step": 15472
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.576659980455691e-05,
+      "loss": 0.8599,
+      "step": 15473
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.5764028184950884e-05,
+      "loss": 0.5561,
+      "step": 15474
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.5761456565344854e-05,
+      "loss": 0.7633,
+      "step": 15475
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.5758884945738827e-05,
+      "loss": 0.5368,
+      "step": 15476
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.57563133261328e-05,
+      "loss": 0.5222,
+      "step": 15477
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.575374170652677e-05,
+      "loss": 0.6398,
+      "step": 15478
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.5751170086920743e-05,
+      "loss": 0.8037,
+      "step": 15479
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.5748598467314716e-05,
+      "loss": 0.7634,
+      "step": 15480
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.5746026847708686e-05,
+      "loss": 0.6138,
+      "step": 15481
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.5743455228102662e-05,
+      "loss": 0.6565,
+      "step": 15482
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.574088360849663e-05,
+      "loss": 0.6156,
+      "step": 15483
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.57383119888906e-05,
+      "loss": 0.7323,
+      "step": 15484
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.5735740369284578e-05,
+      "loss": 0.5884,
+      "step": 15485
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.5733168749678548e-05,
+      "loss": 0.8321,
+      "step": 15486
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.573059713007252e-05,
+      "loss": 0.7034,
+      "step": 15487
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.5728025510466494e-05,
+      "loss": 0.7934,
+      "step": 15488
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.5725453890860463e-05,
+      "loss": 1.2692,
+      "step": 15489
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.5722882271254436e-05,
+      "loss": 0.8136,
+      "step": 15490
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.572031065164841e-05,
+      "loss": 0.6452,
+      "step": 15491
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.571773903204238e-05,
+      "loss": 0.8194,
+      "step": 15492
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.5715167412436352e-05,
+      "loss": 0.8411,
+      "step": 15493
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.5712595792830325e-05,
+      "loss": 0.5753,
+      "step": 15494
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.57100241732243e-05,
+      "loss": 0.8005,
+      "step": 15495
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.5707452553618268e-05,
+      "loss": 0.6884,
+      "step": 15496
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.570488093401224e-05,
+      "loss": 0.5997,
+      "step": 15497
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.5702309314406214e-05,
+      "loss": 0.6478,
+      "step": 15498
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.5699737694800184e-05,
+      "loss": 0.6733,
+      "step": 15499
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.5697166075194157e-05,
+      "loss": 0.5661,
+      "step": 15500
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.569459445558813e-05,
+      "loss": 0.4519,
+      "step": 15501
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.5692022835982103e-05,
+      "loss": 0.8759,
+      "step": 15502
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.5689451216376076e-05,
+      "loss": 0.6228,
+      "step": 15503
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.5686879596770046e-05,
+      "loss": 0.8219,
+      "step": 15504
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.568430797716402e-05,
+      "loss": 0.6383,
+      "step": 15505
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.5681736357557992e-05,
+      "loss": 0.5003,
+      "step": 15506
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.5679164737951962e-05,
+      "loss": 0.6354,
+      "step": 15507
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.5676593118345935e-05,
+      "loss": 0.4866,
+      "step": 15508
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.5674021498739908e-05,
+      "loss": 0.4024,
+      "step": 15509
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.5671449879133878e-05,
+      "loss": 0.7569,
+      "step": 15510
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.5668878259527854e-05,
+      "loss": 0.4166,
+      "step": 15511
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.5666306639921824e-05,
+      "loss": 0.5557,
+      "step": 15512
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.5663735020315793e-05,
+      "loss": 0.8063,
+      "step": 15513
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.566116340070977e-05,
+      "loss": 0.819,
+      "step": 15514
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.565859178110374e-05,
+      "loss": 0.4937,
+      "step": 15515
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.5656020161497713e-05,
+      "loss": 0.5396,
+      "step": 15516
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.5653448541891686e-05,
+      "loss": 0.5127,
+      "step": 15517
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.5650876922285655e-05,
+      "loss": 0.8917,
+      "step": 15518
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.564830530267963e-05,
+      "loss": 0.7175,
+      "step": 15519
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.56457336830736e-05,
+      "loss": 0.5303,
+      "step": 15520
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.5643162063467575e-05,
+      "loss": 0.7587,
+      "step": 15521
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.5640590443861544e-05,
+      "loss": 0.6198,
+      "step": 15522
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.5638018824255517e-05,
+      "loss": 1.0588,
+      "step": 15523
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.563544720464949e-05,
+      "loss": 0.6671,
+      "step": 15524
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.563287558504346e-05,
+      "loss": 0.8732,
+      "step": 15525
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.5630303965437433e-05,
+      "loss": 0.7567,
+      "step": 15526
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.5627732345831406e-05,
+      "loss": 0.7969,
+      "step": 15527
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.5625160726225376e-05,
+      "loss": 0.5536,
+      "step": 15528
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.5622589106619352e-05,
+      "loss": 0.7462,
+      "step": 15529
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.5620017487013322e-05,
+      "loss": 0.5844,
+      "step": 15530
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.5617445867407292e-05,
+      "loss": 0.7505,
+      "step": 15531
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.5614874247801268e-05,
+      "loss": 0.7827,
+      "step": 15532
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.5612302628195238e-05,
+      "loss": 0.8035,
+      "step": 15533
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.560973100858921e-05,
+      "loss": 0.439,
+      "step": 15534
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.5607159388983184e-05,
+      "loss": 0.8864,
+      "step": 15535
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.5604587769377154e-05,
+      "loss": 0.7045,
+      "step": 15536
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.5602016149771127e-05,
+      "loss": 0.6905,
+      "step": 15537
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.55994445301651e-05,
+      "loss": 0.6007,
+      "step": 15538
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.559687291055907e-05,
+      "loss": 0.5648,
+      "step": 15539
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.5594301290953043e-05,
+      "loss": 0.924,
+      "step": 15540
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.5591729671347016e-05,
+      "loss": 0.5529,
+      "step": 15541
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.558915805174099e-05,
+      "loss": 0.8746,
+      "step": 15542
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.558658643213496e-05,
+      "loss": 0.6816,
+      "step": 15543
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.558401481252893e-05,
+      "loss": 0.6907,
+      "step": 15544
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.5581443192922905e-05,
+      "loss": 0.7259,
+      "step": 15545
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.5578871573316874e-05,
+      "loss": 0.4237,
+      "step": 15546
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.5576299953710847e-05,
+      "loss": 0.8784,
+      "step": 15547
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.557372833410482e-05,
+      "loss": 0.7513,
+      "step": 15548
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.557115671449879e-05,
+      "loss": 0.8302,
+      "step": 15549
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.5568585094892767e-05,
+      "loss": 0.9604,
+      "step": 15550
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.5566013475286736e-05,
+      "loss": 0.7677,
+      "step": 15551
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.5563441855680706e-05,
+      "loss": 0.7085,
+      "step": 15552
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.5560870236074682e-05,
+      "loss": 0.6754,
+      "step": 15553
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.5558298616468652e-05,
+      "loss": 0.7737,
+      "step": 15554
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.5555726996862625e-05,
+      "loss": 0.8403,
+      "step": 15555
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.5553155377256598e-05,
+      "loss": 0.7125,
+      "step": 15556
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.5550583757650568e-05,
+      "loss": 0.8395,
+      "step": 15557
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.554801213804454e-05,
+      "loss": 0.7486,
+      "step": 15558
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.5545440518438514e-05,
+      "loss": 0.5526,
+      "step": 15559
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.5542868898832487e-05,
+      "loss": 0.8012,
+      "step": 15560
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.5540297279226457e-05,
+      "loss": 0.7103,
+      "step": 15561
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.553772565962043e-05,
+      "loss": 0.8712,
+      "step": 15562
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.5535154040014403e-05,
+      "loss": 0.8481,
+      "step": 15563
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.5532582420408373e-05,
+      "loss": 0.6814,
+      "step": 15564
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.5530010800802346e-05,
+      "loss": 0.7659,
+      "step": 15565
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.552743918119632e-05,
+      "loss": 0.8466,
+      "step": 15566
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.552486756159029e-05,
+      "loss": 0.9455,
+      "step": 15567
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.5522295941984265e-05,
+      "loss": 0.765,
+      "step": 15568
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.5519724322378235e-05,
+      "loss": 0.87,
+      "step": 15569
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.5517152702772204e-05,
+      "loss": 0.5496,
+      "step": 15570
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.551458108316618e-05,
+      "loss": 0.5232,
+      "step": 15571
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.551200946356015e-05,
+      "loss": 0.5433,
+      "step": 15572
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.5509437843954124e-05,
+      "loss": 0.7853,
+      "step": 15573
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.5506866224348097e-05,
+      "loss": 0.4375,
+      "step": 15574
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.5504294604742066e-05,
+      "loss": 0.5946,
+      "step": 15575
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.550172298513604e-05,
+      "loss": 0.553,
+      "step": 15576
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.5499151365530013e-05,
+      "loss": 0.6465,
+      "step": 15577
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.5496579745923982e-05,
+      "loss": 0.6469,
+      "step": 15578
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.5494008126317955e-05,
+      "loss": 1.0799,
+      "step": 15579
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.549143650671193e-05,
+      "loss": 0.6018,
+      "step": 15580
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.54888648871059e-05,
+      "loss": 0.5625,
+      "step": 15581
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.548629326749987e-05,
+      "loss": 0.9719,
+      "step": 15582
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.5483721647893844e-05,
+      "loss": 0.6066,
+      "step": 15583
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.5481150028287817e-05,
+      "loss": 0.9128,
+      "step": 15584
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.5478578408681787e-05,
+      "loss": 0.6815,
+      "step": 15585
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.547600678907576e-05,
+      "loss": 0.4997,
+      "step": 15586
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.5473435169469733e-05,
+      "loss": 0.317,
+      "step": 15587
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.5470863549863703e-05,
+      "loss": 0.5463,
+      "step": 15588
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.546829193025768e-05,
+      "loss": 0.5844,
+      "step": 15589
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.546572031065165e-05,
+      "loss": 0.4398,
+      "step": 15590
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.546314869104562e-05,
+      "loss": 0.6117,
+      "step": 15591
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.5460577071439595e-05,
+      "loss": 0.8651,
+      "step": 15592
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.5458005451833565e-05,
+      "loss": 1.0009,
+      "step": 15593
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.5455433832227538e-05,
+      "loss": 0.532,
+      "step": 15594
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.545286221262151e-05,
+      "loss": 0.7916,
+      "step": 15595
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.545029059301548e-05,
+      "loss": 0.6598,
+      "step": 15596
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.5447718973409454e-05,
+      "loss": 0.5588,
+      "step": 15597
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.5445147353803427e-05,
+      "loss": 0.7986,
+      "step": 15598
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.54425757341974e-05,
+      "loss": 0.5549,
+      "step": 15599
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.544000411459137e-05,
+      "loss": 0.7839,
+      "step": 15600
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.5437432494985343e-05,
+      "loss": 0.9403,
+      "step": 15601
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.5434860875379316e-05,
+      "loss": 0.3701,
+      "step": 15602
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.5432289255773285e-05,
+      "loss": 0.9292,
+      "step": 15603
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.542971763616726e-05,
+      "loss": 0.6314,
+      "step": 15604
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.542714601656123e-05,
+      "loss": 0.6943,
+      "step": 15605
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.54245743969552e-05,
+      "loss": 0.3327,
+      "step": 15606
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.5422002777349178e-05,
+      "loss": 0.8112,
+      "step": 15607
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.5419431157743147e-05,
+      "loss": 0.8651,
+      "step": 15608
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.5416859538137117e-05,
+      "loss": 0.8054,
+      "step": 15609
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.5414287918531093e-05,
+      "loss": 0.6504,
+      "step": 15610
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.5411716298925063e-05,
+      "loss": 0.6834,
+      "step": 15611
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.5409144679319036e-05,
+      "loss": 0.7342,
+      "step": 15612
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.540657305971301e-05,
+      "loss": 0.5037,
+      "step": 15613
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.540400144010698e-05,
+      "loss": 0.6907,
+      "step": 15614
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.5401429820500952e-05,
+      "loss": 0.5438,
+      "step": 15615
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.5398858200894925e-05,
+      "loss": 0.8204,
+      "step": 15616
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.5396286581288895e-05,
+      "loss": 0.6533,
+      "step": 15617
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.5393714961682868e-05,
+      "loss": 0.7128,
+      "step": 15618
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.539114334207684e-05,
+      "loss": 0.8156,
+      "step": 15619
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.5388571722470814e-05,
+      "loss": 0.622,
+      "step": 15620
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.5386000102864784e-05,
+      "loss": 0.5983,
+      "step": 15621
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.5383428483258757e-05,
+      "loss": 0.7652,
+      "step": 15622
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.538085686365273e-05,
+      "loss": 0.7676,
+      "step": 15623
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.53782852440467e-05,
+      "loss": 0.7724,
+      "step": 15624
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.5375713624440673e-05,
+      "loss": 0.5797,
+      "step": 15625
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.5373142004834646e-05,
+      "loss": 0.9132,
+      "step": 15626
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.5370570385228615e-05,
+      "loss": 0.7151,
+      "step": 15627
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.5367998765622592e-05,
+      "loss": 0.6435,
+      "step": 15628
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.536542714601656e-05,
+      "loss": 0.9842,
+      "step": 15629
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.536285552641053e-05,
+      "loss": 0.5614,
+      "step": 15630
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.5360283906804508e-05,
+      "loss": 0.8547,
+      "step": 15631
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.5357712287198477e-05,
+      "loss": 0.5069,
+      "step": 15632
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.535514066759245e-05,
+      "loss": 0.4843,
+      "step": 15633
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.5352569047986423e-05,
+      "loss": 0.271,
+      "step": 15634
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.5349997428380393e-05,
+      "loss": 0.6383,
+      "step": 15635
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.5347425808774366e-05,
+      "loss": 0.6155,
+      "step": 15636
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.534485418916834e-05,
+      "loss": 0.6353,
+      "step": 15637
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.5342282569562312e-05,
+      "loss": 0.6587,
+      "step": 15638
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.5339710949956282e-05,
+      "loss": 0.8234,
+      "step": 15639
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.5337139330350255e-05,
+      "loss": 0.607,
+      "step": 15640
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.5334567710744228e-05,
+      "loss": 0.7949,
+      "step": 15641
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.5331996091138198e-05,
+      "loss": 0.7909,
+      "step": 15642
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.532942447153217e-05,
+      "loss": 0.8285,
+      "step": 15643
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.5326852851926144e-05,
+      "loss": 0.6264,
+      "step": 15644
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.5324281232320114e-05,
+      "loss": 0.7423,
+      "step": 15645
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.532170961271409e-05,
+      "loss": 0.5774,
+      "step": 15646
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.531913799310806e-05,
+      "loss": 0.3811,
+      "step": 15647
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.531656637350203e-05,
+      "loss": 0.769,
+      "step": 15648
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.5313994753896006e-05,
+      "loss": 0.7379,
+      "step": 15649
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.5311423134289976e-05,
+      "loss": 0.617,
+      "step": 15650
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.530885151468395e-05,
+      "loss": 0.3994,
+      "step": 15651
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.5306279895077922e-05,
+      "loss": 0.643,
+      "step": 15652
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.530370827547189e-05,
+      "loss": 0.7637,
+      "step": 15653
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.5301136655865868e-05,
+      "loss": 0.662,
+      "step": 15654
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.5298565036259838e-05,
+      "loss": 0.8155,
+      "step": 15655
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.5295993416653807e-05,
+      "loss": 0.4797,
+      "step": 15656
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.5293421797047784e-05,
+      "loss": 0.547,
+      "step": 15657
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.5290850177441754e-05,
+      "loss": 0.6685,
+      "step": 15658
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.5288278557835727e-05,
+      "loss": 0.5616,
+      "step": 15659
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.52857069382297e-05,
+      "loss": 0.6619,
+      "step": 15660
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.528313531862367e-05,
+      "loss": 0.7814,
+      "step": 15661
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.5280563699017642e-05,
+      "loss": 0.5058,
+      "step": 15662
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.5277992079411615e-05,
+      "loss": 0.9657,
+      "step": 15663
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.5275420459805585e-05,
+      "loss": 0.8542,
+      "step": 15664
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.5272848840199558e-05,
+      "loss": 0.8359,
+      "step": 15665
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.527027722059353e-05,
+      "loss": 0.7158,
+      "step": 15666
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.5267705600987504e-05,
+      "loss": 0.3176,
+      "step": 15667
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.5265133981381474e-05,
+      "loss": 0.7085,
+      "step": 15668
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.5262562361775447e-05,
+      "loss": 0.6673,
+      "step": 15669
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.525999074216942e-05,
+      "loss": 0.7686,
+      "step": 15670
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.525741912256339e-05,
+      "loss": 0.8959,
+      "step": 15671
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.5254847502957365e-05,
+      "loss": 1.0022,
+      "step": 15672
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.5252275883351336e-05,
+      "loss": 0.8606,
+      "step": 15673
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.5249704263745307e-05,
+      "loss": 0.7532,
+      "step": 15674
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.524713264413928e-05,
+      "loss": 0.6799,
+      "step": 15675
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.5244561024533252e-05,
+      "loss": 0.5532,
+      "step": 15676
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.5241989404927223e-05,
+      "loss": 0.8442,
+      "step": 15677
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.5239417785321198e-05,
+      "loss": 0.7587,
+      "step": 15678
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.5236846165715168e-05,
+      "loss": 0.7673,
+      "step": 15679
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.5234274546109139e-05,
+      "loss": 0.9143,
+      "step": 15680
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.5231702926503114e-05,
+      "loss": 1.1531,
+      "step": 15681
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.5229131306897085e-05,
+      "loss": 0.8393,
+      "step": 15682
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.5226559687291057e-05,
+      "loss": 0.8486,
+      "step": 15683
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.522398806768503e-05,
+      "loss": 0.6351,
+      "step": 15684
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.5221416448079001e-05,
+      "loss": 0.4445,
+      "step": 15685
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.5218844828472972e-05,
+      "loss": 0.9007,
+      "step": 15686
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.5216273208866946e-05,
+      "loss": 0.933,
+      "step": 15687
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.5213701589260917e-05,
+      "loss": 0.7082,
+      "step": 15688
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.5211129969654888e-05,
+      "loss": 0.5315,
+      "step": 15689
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.5208558350048863e-05,
+      "loss": 0.5698,
+      "step": 15690
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.5205986730442834e-05,
+      "loss": 0.5155,
+      "step": 15691
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.5203415110836806e-05,
+      "loss": 0.6175,
+      "step": 15692
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.5200843491230779e-05,
+      "loss": 0.6198,
+      "step": 15693
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.519827187162475e-05,
+      "loss": 0.8008,
+      "step": 15694
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.5195700252018722e-05,
+      "loss": 0.921,
+      "step": 15695
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.5193128632412695e-05,
+      "loss": 0.9881,
+      "step": 15696
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.5190557012806666e-05,
+      "loss": 0.7706,
+      "step": 15697
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.5187985393200637e-05,
+      "loss": 0.5568,
+      "step": 15698
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.5185413773594612e-05,
+      "loss": 0.7639,
+      "step": 15699
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.5182842153988584e-05,
+      "loss": 0.7178,
+      "step": 15700
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.5180270534382553e-05,
+      "loss": 0.7947,
+      "step": 15701
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.5177698914776528e-05,
+      "loss": 0.8598,
+      "step": 15702
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.51751272951705e-05,
+      "loss": 1.0077,
+      "step": 15703
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.5172555675564471e-05,
+      "loss": 0.6107,
+      "step": 15704
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.5169984055958444e-05,
+      "loss": 0.8607,
+      "step": 15705
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.5167412436352415e-05,
+      "loss": 0.575,
+      "step": 15706
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.5164840816746387e-05,
+      "loss": 0.6508,
+      "step": 15707
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.5162269197140361e-05,
+      "loss": 0.5159,
+      "step": 15708
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.5159697577534331e-05,
+      "loss": 0.6842,
+      "step": 15709
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.5157125957928303e-05,
+      "loss": 0.6995,
+      "step": 15710
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.5154554338322277e-05,
+      "loss": 0.6014,
+      "step": 15711
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.5151982718716249e-05,
+      "loss": 0.521,
+      "step": 15712
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.514941109911022e-05,
+      "loss": 0.7184,
+      "step": 15713
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.5146839479504193e-05,
+      "loss": 0.467,
+      "step": 15714
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.5144267859898164e-05,
+      "loss": 0.7172,
+      "step": 15715
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.5141696240292136e-05,
+      "loss": 0.5318,
+      "step": 15716
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.513912462068611e-05,
+      "loss": 0.7138,
+      "step": 15717
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.513655300108008e-05,
+      "loss": 0.7656,
+      "step": 15718
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.5133981381474052e-05,
+      "loss": 1.1249,
+      "step": 15719
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.5131409761868026e-05,
+      "loss": 0.8799,
+      "step": 15720
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.5128838142261998e-05,
+      "loss": 0.9671,
+      "step": 15721
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.512626652265597e-05,
+      "loss": 0.365,
+      "step": 15722
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.5123694903049942e-05,
+      "loss": 0.6523,
+      "step": 15723
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.5121123283443914e-05,
+      "loss": 0.8579,
+      "step": 15724
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.5118551663837885e-05,
+      "loss": 0.8581,
+      "step": 15725
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.5115980044231858e-05,
+      "loss": 0.5343,
+      "step": 15726
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.511340842462583e-05,
+      "loss": 0.9691,
+      "step": 15727
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.5110836805019801e-05,
+      "loss": 0.7317,
+      "step": 15728
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.5108265185413776e-05,
+      "loss": 0.5933,
+      "step": 15729
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.5105693565807747e-05,
+      "loss": 0.7986,
+      "step": 15730
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.5103121946201718e-05,
+      "loss": 0.5008,
+      "step": 15731
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.5100550326595691e-05,
+      "loss": 0.7084,
+      "step": 15732
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.5097978706989663e-05,
+      "loss": 0.8232,
+      "step": 15733
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.5095407087383634e-05,
+      "loss": 0.5951,
+      "step": 15734
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.5092835467777607e-05,
+      "loss": 0.4572,
+      "step": 15735
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.5090263848171579e-05,
+      "loss": 0.6417,
+      "step": 15736
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.508769222856555e-05,
+      "loss": 1.0237,
+      "step": 15737
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.5085120608959525e-05,
+      "loss": 0.6025,
+      "step": 15738
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.5082548989353496e-05,
+      "loss": 0.7963,
+      "step": 15739
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.5079977369747466e-05,
+      "loss": 0.7098,
+      "step": 15740
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.507740575014144e-05,
+      "loss": 0.6404,
+      "step": 15741
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.5074834130535412e-05,
+      "loss": 0.6335,
+      "step": 15742
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.5072262510929383e-05,
+      "loss": 0.5033,
+      "step": 15743
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.5069690891323357e-05,
+      "loss": 0.7321,
+      "step": 15744
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.5067119271717328e-05,
+      "loss": 0.8484,
+      "step": 15745
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.50645476521113e-05,
+      "loss": 1.2055,
+      "step": 15746
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.5061976032505274e-05,
+      "loss": 0.7052,
+      "step": 15747
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.5059404412899244e-05,
+      "loss": 0.5326,
+      "step": 15748
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.5056832793293215e-05,
+      "loss": 0.5409,
+      "step": 15749
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.505426117368719e-05,
+      "loss": 0.7599,
+      "step": 15750
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.5051689554081161e-05,
+      "loss": 0.8515,
+      "step": 15751
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.5049117934475133e-05,
+      "loss": 0.6745,
+      "step": 15752
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.5046546314869106e-05,
+      "loss": 0.4948,
+      "step": 15753
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.5043974695263077e-05,
+      "loss": 0.6395,
+      "step": 15754
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.5041403075657048e-05,
+      "loss": 0.6125,
+      "step": 15755
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.5038831456051023e-05,
+      "loss": 0.8781,
+      "step": 15756
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.5036259836444993e-05,
+      "loss": 0.6566,
+      "step": 15757
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.5033688216838964e-05,
+      "loss": 0.7087,
+      "step": 15758
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.5031116597232939e-05,
+      "loss": 0.7444,
+      "step": 15759
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.502854497762691e-05,
+      "loss": 0.6009,
+      "step": 15760
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.5025973358020882e-05,
+      "loss": 0.6941,
+      "step": 15761
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.5023401738414855e-05,
+      "loss": 0.5642,
+      "step": 15762
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.5020830118808826e-05,
+      "loss": 0.5757,
+      "step": 15763
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.5018258499202798e-05,
+      "loss": 0.7874,
+      "step": 15764
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.501568687959677e-05,
+      "loss": 0.7323,
+      "step": 15765
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.5013115259990742e-05,
+      "loss": 0.8276,
+      "step": 15766
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.5010543640384713e-05,
+      "loss": 0.7993,
+      "step": 15767
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.5007972020778688e-05,
+      "loss": 0.5813,
+      "step": 15768
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.500540040117266e-05,
+      "loss": 0.8065,
+      "step": 15769
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.5002828781566631e-05,
+      "loss": 0.8582,
+      "step": 15770
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.5000257161960604e-05,
+      "loss": 0.4706,
+      "step": 15771
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.4997685542354575e-05,
+      "loss": 0.7104,
+      "step": 15772
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.4995113922748547e-05,
+      "loss": 0.7456,
+      "step": 15773
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.499254230314252e-05,
+      "loss": 0.7461,
+      "step": 15774
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.4989970683536491e-05,
+      "loss": 0.5908,
+      "step": 15775
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.4987399063930463e-05,
+      "loss": 1.0472,
+      "step": 15776
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.4984827444324437e-05,
+      "loss": 0.5279,
+      "step": 15777
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.4982255824718409e-05,
+      "loss": 0.8464,
+      "step": 15778
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.4979684205112379e-05,
+      "loss": 0.5509,
+      "step": 15779
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.4977112585506353e-05,
+      "loss": 0.4748,
+      "step": 15780
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.4974540965900325e-05,
+      "loss": 0.9375,
+      "step": 15781
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.4971969346294296e-05,
+      "loss": 0.7591,
+      "step": 15782
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.4969397726688269e-05,
+      "loss": 0.6062,
+      "step": 15783
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.496682610708224e-05,
+      "loss": 0.8209,
+      "step": 15784
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.4964254487476212e-05,
+      "loss": 1.0084,
+      "step": 15785
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.4961682867870187e-05,
+      "loss": 0.5929,
+      "step": 15786
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.4959111248264156e-05,
+      "loss": 0.5571,
+      "step": 15787
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.4956539628658128e-05,
+      "loss": 0.5674,
+      "step": 15788
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.4953968009052102e-05,
+      "loss": 0.7172,
+      "step": 15789
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.4951396389446074e-05,
+      "loss": 0.6077,
+      "step": 15790
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.4948824769840045e-05,
+      "loss": 0.7921,
+      "step": 15791
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.4946253150234018e-05,
+      "loss": 0.4685,
+      "step": 15792
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.494368153062799e-05,
+      "loss": 0.8372,
+      "step": 15793
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.4941109911021961e-05,
+      "loss": 0.6352,
+      "step": 15794
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.4938538291415934e-05,
+      "loss": 0.5048,
+      "step": 15795
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.4935966671809906e-05,
+      "loss": 0.7063,
+      "step": 15796
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.4933395052203877e-05,
+      "loss": 0.6894,
+      "step": 15797
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.4930823432597852e-05,
+      "loss": 0.9176,
+      "step": 15798
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.4928251812991823e-05,
+      "loss": 0.8979,
+      "step": 15799
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.4925680193385794e-05,
+      "loss": 0.5839,
+      "step": 15800
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.4923108573779767e-05,
+      "loss": 0.4764,
+      "step": 15801
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.4920536954173739e-05,
+      "loss": 0.6414,
+      "step": 15802
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.4917965334567714e-05,
+      "loss": 0.687,
+      "step": 15803
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.4915393714961683e-05,
+      "loss": 0.4689,
+      "step": 15804
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.4912822095355655e-05,
+      "loss": 0.5893,
+      "step": 15805
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.491025047574963e-05,
+      "loss": 0.4431,
+      "step": 15806
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.4907678856143601e-05,
+      "loss": 0.4446,
+      "step": 15807
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.4905107236537572e-05,
+      "loss": 0.9306,
+      "step": 15808
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.4902535616931545e-05,
+      "loss": 0.7196,
+      "step": 15809
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.4899963997325517e-05,
+      "loss": 0.6793,
+      "step": 15810
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.4897392377719488e-05,
+      "loss": 0.6649,
+      "step": 15811
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.4894820758113461e-05,
+      "loss": 0.6931,
+      "step": 15812
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.4892249138507433e-05,
+      "loss": 0.893,
+      "step": 15813
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.4889677518901404e-05,
+      "loss": 0.7631,
+      "step": 15814
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.4887105899295379e-05,
+      "loss": 0.6318,
+      "step": 15815
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.488453427968935e-05,
+      "loss": 0.7684,
+      "step": 15816
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.4881962660083321e-05,
+      "loss": 0.8291,
+      "step": 15817
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.4879391040477294e-05,
+      "loss": 0.5012,
+      "step": 15818
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.4876819420871266e-05,
+      "loss": 0.9001,
+      "step": 15819
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.4874247801265237e-05,
+      "loss": 0.597,
+      "step": 15820
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.487167618165921e-05,
+      "loss": 0.4643,
+      "step": 15821
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.4869104562053182e-05,
+      "loss": 1.1815,
+      "step": 15822
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.4866532942447153e-05,
+      "loss": 0.5656,
+      "step": 15823
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.4863961322841128e-05,
+      "loss": 0.8452,
+      "step": 15824
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.48613897032351e-05,
+      "loss": 0.8808,
+      "step": 15825
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.4858818083629069e-05,
+      "loss": 0.8499,
+      "step": 15826
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.4856246464023044e-05,
+      "loss": 0.6488,
+      "step": 15827
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.4853674844417015e-05,
+      "loss": 0.3595,
+      "step": 15828
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.4851103224810986e-05,
+      "loss": 0.6911,
+      "step": 15829
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.484853160520496e-05,
+      "loss": 0.6808,
+      "step": 15830
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.4845959985598931e-05,
+      "loss": 0.6779,
+      "step": 15831
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.4843388365992902e-05,
+      "loss": 0.7745,
+      "step": 15832
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.4840816746386877e-05,
+      "loss": 0.586,
+      "step": 15833
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.4838245126780847e-05,
+      "loss": 0.4541,
+      "step": 15834
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.4835673507174818e-05,
+      "loss": 0.9729,
+      "step": 15835
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.4833101887568793e-05,
+      "loss": 0.6479,
+      "step": 15836
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.4830530267962764e-05,
+      "loss": 0.6585,
+      "step": 15837
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.4827958648356736e-05,
+      "loss": 0.5168,
+      "step": 15838
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.4825387028750709e-05,
+      "loss": 0.7445,
+      "step": 15839
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.482281540914468e-05,
+      "loss": 0.6621,
+      "step": 15840
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.4820243789538651e-05,
+      "loss": 0.6117,
+      "step": 15841
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.4817672169932626e-05,
+      "loss": 0.8921,
+      "step": 15842
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.4815100550326596e-05,
+      "loss": 0.7568,
+      "step": 15843
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.4812528930720567e-05,
+      "loss": 0.9375,
+      "step": 15844
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.4809957311114542e-05,
+      "loss": 0.9459,
+      "step": 15845
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.4807385691508513e-05,
+      "loss": 0.3268,
+      "step": 15846
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.4804814071902485e-05,
+      "loss": 0.717,
+      "step": 15847
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.4802242452296458e-05,
+      "loss": 0.7893,
+      "step": 15848
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.479967083269043e-05,
+      "loss": 0.9415,
+      "step": 15849
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.47970992130844e-05,
+      "loss": 0.7225,
+      "step": 15850
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.4794527593478374e-05,
+      "loss": 0.8796,
+      "step": 15851
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.4791955973872345e-05,
+      "loss": 1.0061,
+      "step": 15852
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.4789384354266316e-05,
+      "loss": 0.457,
+      "step": 15853
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.4786812734660291e-05,
+      "loss": 0.8665,
+      "step": 15854
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.4784241115054263e-05,
+      "loss": 0.7403,
+      "step": 15855
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.4781669495448234e-05,
+      "loss": 0.5969,
+      "step": 15856
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.4779097875842207e-05,
+      "loss": 0.7707,
+      "step": 15857
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.4776526256236178e-05,
+      "loss": 0.7169,
+      "step": 15858
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.477395463663015e-05,
+      "loss": 1.1401,
+      "step": 15859
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.4771383017024123e-05,
+      "loss": 0.9629,
+      "step": 15860
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.4768811397418094e-05,
+      "loss": 0.6741,
+      "step": 15861
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.4766239777812066e-05,
+      "loss": 0.6687,
+      "step": 15862
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.476366815820604e-05,
+      "loss": 0.8802,
+      "step": 15863
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.4761096538600012e-05,
+      "loss": 0.5997,
+      "step": 15864
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.4758524918993982e-05,
+      "loss": 0.3499,
+      "step": 15865
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.4755953299387956e-05,
+      "loss": 0.7366,
+      "step": 15866
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.4753381679781928e-05,
+      "loss": 0.6399,
+      "step": 15867
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.4750810060175899e-05,
+      "loss": 0.604,
+      "step": 15868
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.4748238440569872e-05,
+      "loss": 0.5199,
+      "step": 15869
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.4745666820963843e-05,
+      "loss": 0.6412,
+      "step": 15870
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.4743095201357815e-05,
+      "loss": 0.8162,
+      "step": 15871
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.474052358175179e-05,
+      "loss": 0.8271,
+      "step": 15872
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.473795196214576e-05,
+      "loss": 1.0155,
+      "step": 15873
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.473538034253973e-05,
+      "loss": 0.7472,
+      "step": 15874
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.4732808722933705e-05,
+      "loss": 0.6865,
+      "step": 15875
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.4730237103327677e-05,
+      "loss": 0.8485,
+      "step": 15876
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.4727665483721648e-05,
+      "loss": 0.6603,
+      "step": 15877
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.4725093864115621e-05,
+      "loss": 0.9852,
+      "step": 15878
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.4722522244509593e-05,
+      "loss": 0.612,
+      "step": 15879
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.4719950624903564e-05,
+      "loss": 0.4503,
+      "step": 15880
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.4717379005297539e-05,
+      "loss": 0.704,
+      "step": 15881
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.4714807385691509e-05,
+      "loss": 0.6671,
+      "step": 15882
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.471223576608548e-05,
+      "loss": 0.6865,
+      "step": 15883
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.4709664146479455e-05,
+      "loss": 0.6154,
+      "step": 15884
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.4707092526873426e-05,
+      "loss": 0.5857,
+      "step": 15885
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.4704520907267397e-05,
+      "loss": 0.5351,
+      "step": 15886
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.470194928766137e-05,
+      "loss": 0.8154,
+      "step": 15887
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.4699377668055342e-05,
+      "loss": 0.6375,
+      "step": 15888
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.4696806048449313e-05,
+      "loss": 0.6117,
+      "step": 15889
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.4694234428843286e-05,
+      "loss": 0.7704,
+      "step": 15890
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.4691662809237258e-05,
+      "loss": 0.8006,
+      "step": 15891
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.4689091189631229e-05,
+      "loss": 0.798,
+      "step": 15892
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.4686519570025204e-05,
+      "loss": 0.6752,
+      "step": 15893
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.4683947950419175e-05,
+      "loss": 0.5193,
+      "step": 15894
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.4681376330813147e-05,
+      "loss": 0.5156,
+      "step": 15895
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.467880471120712e-05,
+      "loss": 0.7246,
+      "step": 15896
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.4676233091601091e-05,
+      "loss": 0.8127,
+      "step": 15897
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.4673661471995062e-05,
+      "loss": 0.6071,
+      "step": 15898
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.4671089852389036e-05,
+      "loss": 0.4875,
+      "step": 15899
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.4668518232783007e-05,
+      "loss": 0.8119,
+      "step": 15900
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.4665946613176978e-05,
+      "loss": 0.7033,
+      "step": 15901
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.4663374993570953e-05,
+      "loss": 0.8234,
+      "step": 15902
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.4660803373964924e-05,
+      "loss": 0.7475,
+      "step": 15903
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.4658231754358894e-05,
+      "loss": 0.7544,
+      "step": 15904
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.4655660134752869e-05,
+      "loss": 0.8431,
+      "step": 15905
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.465308851514684e-05,
+      "loss": 0.6381,
+      "step": 15906
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.4650516895540812e-05,
+      "loss": 0.7875,
+      "step": 15907
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.4647945275934785e-05,
+      "loss": 0.4845,
+      "step": 15908
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.4645373656328756e-05,
+      "loss": 0.689,
+      "step": 15909
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.4642802036722727e-05,
+      "loss": 0.6134,
+      "step": 15910
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.4640230417116702e-05,
+      "loss": 0.5956,
+      "step": 15911
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.4637658797510672e-05,
+      "loss": 0.6697,
+      "step": 15912
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.4635087177904643e-05,
+      "loss": 0.6748,
+      "step": 15913
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.4632515558298618e-05,
+      "loss": 0.7001,
+      "step": 15914
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.462994393869259e-05,
+      "loss": 0.6157,
+      "step": 15915
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.462737231908656e-05,
+      "loss": 0.847,
+      "step": 15916
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.4624800699480534e-05,
+      "loss": 0.7023,
+      "step": 15917
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.4622229079874505e-05,
+      "loss": 0.7036,
+      "step": 15918
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.4619657460268477e-05,
+      "loss": 0.798,
+      "step": 15919
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.461708584066245e-05,
+      "loss": 0.6166,
+      "step": 15920
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.4614514221056421e-05,
+      "loss": 0.7503,
+      "step": 15921
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.4611942601450392e-05,
+      "loss": 0.5304,
+      "step": 15922
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.4609370981844367e-05,
+      "loss": 0.7977,
+      "step": 15923
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.4606799362238339e-05,
+      "loss": 0.7123,
+      "step": 15924
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.460422774263231e-05,
+      "loss": 0.5305,
+      "step": 15925
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.4601656123026283e-05,
+      "loss": 0.4762,
+      "step": 15926
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.4599084503420254e-05,
+      "loss": 0.7689,
+      "step": 15927
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.4596512883814226e-05,
+      "loss": 0.8638,
+      "step": 15928
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.4593941264208199e-05,
+      "loss": 0.649,
+      "step": 15929
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.459136964460217e-05,
+      "loss": 0.3685,
+      "step": 15930
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.4588798024996142e-05,
+      "loss": 0.6015,
+      "step": 15931
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.4586226405390116e-05,
+      "loss": 1.0704,
+      "step": 15932
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.4583654785784088e-05,
+      "loss": 0.5737,
+      "step": 15933
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.458108316617806e-05,
+      "loss": 0.5455,
+      "step": 15934
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.4578511546572032e-05,
+      "loss": 0.4231,
+      "step": 15935
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.4575939926966004e-05,
+      "loss": 0.4508,
+      "step": 15936
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.4573368307359975e-05,
+      "loss": 0.4152,
+      "step": 15937
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.4570796687753948e-05,
+      "loss": 0.5104,
+      "step": 15938
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.456822506814792e-05,
+      "loss": 0.8454,
+      "step": 15939
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.4565653448541891e-05,
+      "loss": 0.7102,
+      "step": 15940
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.4563081828935866e-05,
+      "loss": 0.6469,
+      "step": 15941
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.4560510209329837e-05,
+      "loss": 0.3593,
+      "step": 15942
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.4557938589723807e-05,
+      "loss": 0.612,
+      "step": 15943
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.4555366970117781e-05,
+      "loss": 0.336,
+      "step": 15944
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.4552795350511753e-05,
+      "loss": 0.9981,
+      "step": 15945
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.4550223730905724e-05,
+      "loss": 0.6197,
+      "step": 15946
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.4547652111299697e-05,
+      "loss": 0.736,
+      "step": 15947
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.4545080491693669e-05,
+      "loss": 1.0093,
+      "step": 15948
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.454250887208764e-05,
+      "loss": 0.8711,
+      "step": 15949
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.4539937252481615e-05,
+      "loss": 0.7173,
+      "step": 15950
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.4537365632875585e-05,
+      "loss": 0.9427,
+      "step": 15951
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.453479401326956e-05,
+      "loss": 0.765,
+      "step": 15952
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.453222239366353e-05,
+      "loss": 0.4725,
+      "step": 15953
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.4529650774057502e-05,
+      "loss": 0.5054,
+      "step": 15954
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.4527079154451475e-05,
+      "loss": 0.6954,
+      "step": 15955
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.4524507534845446e-05,
+      "loss": 0.6826,
+      "step": 15956
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.4521935915239418e-05,
+      "loss": 1.072,
+      "step": 15957
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.4519364295633393e-05,
+      "loss": 0.4962,
+      "step": 15958
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.4516792676027362e-05,
+      "loss": 0.4196,
+      "step": 15959
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.4514221056421334e-05,
+      "loss": 0.6056,
+      "step": 15960
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.4511649436815308e-05,
+      "loss": 0.615,
+      "step": 15961
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.450907781720928e-05,
+      "loss": 0.6349,
+      "step": 15962
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.4506506197603251e-05,
+      "loss": 0.6483,
+      "step": 15963
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.4503934577997224e-05,
+      "loss": 0.4551,
+      "step": 15964
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.4501362958391196e-05,
+      "loss": 0.2236,
+      "step": 15965
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.4498791338785167e-05,
+      "loss": 0.4189,
+      "step": 15966
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.4496219719179142e-05,
+      "loss": 0.3353,
+      "step": 15967
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.4493648099573112e-05,
+      "loss": 0.7965,
+      "step": 15968
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.4491076479967083e-05,
+      "loss": 0.9871,
+      "step": 15969
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.4488504860361058e-05,
+      "loss": 0.6718,
+      "step": 15970
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.4485933240755029e-05,
+      "loss": 0.6602,
+      "step": 15971
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.4483361621149e-05,
+      "loss": 0.5626,
+      "step": 15972
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.4480790001542973e-05,
+      "loss": 0.5515,
+      "step": 15973
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.4478218381936945e-05,
+      "loss": 0.5595,
+      "step": 15974
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.4475646762330916e-05,
+      "loss": 0.3374,
+      "step": 15975
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.447307514272489e-05,
+      "loss": 0.5866,
+      "step": 15976
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.447050352311886e-05,
+      "loss": 0.5177,
+      "step": 15977
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.4467931903512832e-05,
+      "loss": 0.8015,
+      "step": 15978
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.4465360283906807e-05,
+      "loss": 0.5491,
+      "step": 15979
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.4462788664300778e-05,
+      "loss": 0.612,
+      "step": 15980
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.446021704469475e-05,
+      "loss": 0.7879,
+      "step": 15981
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.4457645425088723e-05,
+      "loss": 1.005,
+      "step": 15982
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.4455073805482694e-05,
+      "loss": 0.6779,
+      "step": 15983
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.4452502185876665e-05,
+      "loss": 0.673,
+      "step": 15984
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.4449930566270639e-05,
+      "loss": 0.9176,
+      "step": 15985
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.444735894666461e-05,
+      "loss": 0.6129,
+      "step": 15986
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.4444787327058581e-05,
+      "loss": 0.6797,
+      "step": 15987
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.4442215707452556e-05,
+      "loss": 0.7776,
+      "step": 15988
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.4439644087846527e-05,
+      "loss": 0.7924,
+      "step": 15989
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.4437072468240497e-05,
+      "loss": 0.5417,
+      "step": 15990
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.4434500848634472e-05,
+      "loss": 0.8275,
+      "step": 15991
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.4431929229028443e-05,
+      "loss": 0.6065,
+      "step": 15992
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.4429357609422415e-05,
+      "loss": 0.4949,
+      "step": 15993
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.4426785989816388e-05,
+      "loss": 0.7067,
+      "step": 15994
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.4424214370210359e-05,
+      "loss": 0.7398,
+      "step": 15995
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.442164275060433e-05,
+      "loss": 0.4349,
+      "step": 15996
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.4419071130998305e-05,
+      "loss": 0.7393,
+      "step": 15997
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.4416499511392275e-05,
+      "loss": 0.5596,
+      "step": 15998
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.4413927891786246e-05,
+      "loss": 0.7792,
+      "step": 15999
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.4411356272180221e-05,
+      "loss": 0.448,
+      "step": 16000
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.4408784652574192e-05,
+      "loss": 0.5082,
+      "step": 16001
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.4406213032968164e-05,
+      "loss": 0.7505,
+      "step": 16002
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.4403641413362137e-05,
+      "loss": 0.3838,
+      "step": 16003
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.4401069793756108e-05,
+      "loss": 0.7463,
+      "step": 16004
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.439849817415008e-05,
+      "loss": 0.8231,
+      "step": 16005
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.4395926554544054e-05,
+      "loss": 0.6799,
+      "step": 16006
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.4393354934938024e-05,
+      "loss": 0.5281,
+      "step": 16007
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.4390783315331995e-05,
+      "loss": 0.345,
+      "step": 16008
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.438821169572597e-05,
+      "loss": 0.6169,
+      "step": 16009
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.4385640076119942e-05,
+      "loss": 0.5786,
+      "step": 16010
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.4383068456513913e-05,
+      "loss": 0.6717,
+      "step": 16011
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.4380496836907886e-05,
+      "loss": 0.7014,
+      "step": 16012
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.4377925217301857e-05,
+      "loss": 0.7582,
+      "step": 16013
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.4375353597695829e-05,
+      "loss": 0.9192,
+      "step": 16014
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.4372781978089802e-05,
+      "loss": 0.6746,
+      "step": 16015
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.4370210358483773e-05,
+      "loss": 0.6418,
+      "step": 16016
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.4367638738877745e-05,
+      "loss": 0.9588,
+      "step": 16017
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.436506711927172e-05,
+      "loss": 0.7347,
+      "step": 16018
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.436249549966569e-05,
+      "loss": 0.8105,
+      "step": 16019
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.4359923880059662e-05,
+      "loss": 0.7503,
+      "step": 16020
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.4357352260453635e-05,
+      "loss": 0.4937,
+      "step": 16021
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.4354780640847607e-05,
+      "loss": 0.3677,
+      "step": 16022
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.4352209021241578e-05,
+      "loss": 1.0128,
+      "step": 16023
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.4349637401635551e-05,
+      "loss": 0.4961,
+      "step": 16024
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.4347065782029522e-05,
+      "loss": 0.5685,
+      "step": 16025
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.4344494162423494e-05,
+      "loss": 0.7218,
+      "step": 16026
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.4341922542817469e-05,
+      "loss": 0.4862,
+      "step": 16027
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.433935092321144e-05,
+      "loss": 0.6746,
+      "step": 16028
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.433677930360541e-05,
+      "loss": 0.8025,
+      "step": 16029
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.4334207683999384e-05,
+      "loss": 1.1125,
+      "step": 16030
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.4331636064393356e-05,
+      "loss": 0.6719,
+      "step": 16031
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.4329064444787327e-05,
+      "loss": 0.8393,
+      "step": 16032
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.43264928251813e-05,
+      "loss": 1.0105,
+      "step": 16033
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.4323921205575272e-05,
+      "loss": 0.6275,
+      "step": 16034
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.4321349585969243e-05,
+      "loss": 0.6565,
+      "step": 16035
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.4318777966363218e-05,
+      "loss": 0.847,
+      "step": 16036
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.4316206346757188e-05,
+      "loss": 1.15,
+      "step": 16037
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.4313634727151159e-05,
+      "loss": 0.5199,
+      "step": 16038
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.4311063107545134e-05,
+      "loss": 0.83,
+      "step": 16039
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.4308491487939105e-05,
+      "loss": 0.7003,
+      "step": 16040
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.4305919868333076e-05,
+      "loss": 0.7875,
+      "step": 16041
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.430334824872705e-05,
+      "loss": 0.8475,
+      "step": 16042
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.4300776629121021e-05,
+      "loss": 0.6979,
+      "step": 16043
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.4298205009514992e-05,
+      "loss": 0.7457,
+      "step": 16044
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.4295633389908967e-05,
+      "loss": 0.6119,
+      "step": 16045
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.4293061770302937e-05,
+      "loss": 0.4418,
+      "step": 16046
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.4290490150696908e-05,
+      "loss": 1.0182,
+      "step": 16047
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.4287918531090883e-05,
+      "loss": 0.6979,
+      "step": 16048
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.4285346911484854e-05,
+      "loss": 0.7924,
+      "step": 16049
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.4282775291878826e-05,
+      "loss": 0.6022,
+      "step": 16050
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.4280203672272799e-05,
+      "loss": 0.4788,
+      "step": 16051
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.427763205266677e-05,
+      "loss": 0.8594,
+      "step": 16052
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.4275060433060741e-05,
+      "loss": 0.7224,
+      "step": 16053
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.4272488813454715e-05,
+      "loss": 0.8065,
+      "step": 16054
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.4269917193848686e-05,
+      "loss": 0.7323,
+      "step": 16055
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.4267345574242657e-05,
+      "loss": 0.791,
+      "step": 16056
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.4264773954636632e-05,
+      "loss": 0.6995,
+      "step": 16057
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.4262202335030603e-05,
+      "loss": 0.832,
+      "step": 16058
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.4259630715424575e-05,
+      "loss": 0.553,
+      "step": 16059
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.4257059095818548e-05,
+      "loss": 0.8558,
+      "step": 16060
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.425448747621252e-05,
+      "loss": 0.5669,
+      "step": 16061
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.425191585660649e-05,
+      "loss": 0.8064,
+      "step": 16062
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.4249344237000464e-05,
+      "loss": 0.5768,
+      "step": 16063
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.4246772617394435e-05,
+      "loss": 0.6343,
+      "step": 16064
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.4244200997788406e-05,
+      "loss": 0.5374,
+      "step": 16065
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.4241629378182381e-05,
+      "loss": 0.6478,
+      "step": 16066
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.4239057758576353e-05,
+      "loss": 0.5754,
+      "step": 16067
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.4236486138970322e-05,
+      "loss": 0.4867,
+      "step": 16068
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.4233914519364297e-05,
+      "loss": 0.6979,
+      "step": 16069
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.4231342899758268e-05,
+      "loss": 0.6882,
+      "step": 16070
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.422877128015224e-05,
+      "loss": 0.9101,
+      "step": 16071
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.4226199660546213e-05,
+      "loss": 0.8431,
+      "step": 16072
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.4223628040940184e-05,
+      "loss": 0.5781,
+      "step": 16073
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.4221056421334156e-05,
+      "loss": 0.7106,
+      "step": 16074
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.421848480172813e-05,
+      "loss": 0.6579,
+      "step": 16075
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.42159131821221e-05,
+      "loss": 0.9282,
+      "step": 16076
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.4213341562516071e-05,
+      "loss": 0.4411,
+      "step": 16077
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.4210769942910046e-05,
+      "loss": 0.7063,
+      "step": 16078
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.4208198323304018e-05,
+      "loss": 0.8032,
+      "step": 16079
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.4205626703697989e-05,
+      "loss": 0.8093,
+      "step": 16080
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.4203055084091962e-05,
+      "loss": 0.8804,
+      "step": 16081
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.4200483464485933e-05,
+      "loss": 0.5887,
+      "step": 16082
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.4197911844879905e-05,
+      "loss": 0.6852,
+      "step": 16083
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.4195340225273878e-05,
+      "loss": 0.6436,
+      "step": 16084
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.419276860566785e-05,
+      "loss": 0.7606,
+      "step": 16085
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.419019698606182e-05,
+      "loss": 0.7853,
+      "step": 16086
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.4187625366455795e-05,
+      "loss": 0.4252,
+      "step": 16087
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.4185053746849767e-05,
+      "loss": 0.6649,
+      "step": 16088
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.4182482127243738e-05,
+      "loss": 0.6376,
+      "step": 16089
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.4179910507637711e-05,
+      "loss": 0.7209,
+      "step": 16090
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.4177338888031683e-05,
+      "loss": 0.5754,
+      "step": 16091
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.4174767268425654e-05,
+      "loss": 0.5628,
+      "step": 16092
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.4172195648819627e-05,
+      "loss": 0.8106,
+      "step": 16093
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.4169624029213598e-05,
+      "loss": 0.6865,
+      "step": 16094
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.416705240960757e-05,
+      "loss": 0.8599,
+      "step": 16095
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.4164480790001545e-05,
+      "loss": 0.894,
+      "step": 16096
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.4161909170395516e-05,
+      "loss": 0.9767,
+      "step": 16097
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.4159337550789486e-05,
+      "loss": 0.6123,
+      "step": 16098
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.415676593118346e-05,
+      "loss": 0.5757,
+      "step": 16099
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.4154194311577432e-05,
+      "loss": 1.123,
+      "step": 16100
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.4151622691971403e-05,
+      "loss": 0.3913,
+      "step": 16101
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.4149051072365376e-05,
+      "loss": 0.7649,
+      "step": 16102
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.4146479452759348e-05,
+      "loss": 1.0119,
+      "step": 16103
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.4143907833153322e-05,
+      "loss": 0.5875,
+      "step": 16104
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.4141336213547294e-05,
+      "loss": 0.8603,
+      "step": 16105
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.4138764593941265e-05,
+      "loss": 0.5033,
+      "step": 16106
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.4136192974335238e-05,
+      "loss": 0.9537,
+      "step": 16107
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.413362135472921e-05,
+      "loss": 0.7525,
+      "step": 16108
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.4131049735123181e-05,
+      "loss": 0.5378,
+      "step": 16109
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.4128478115517154e-05,
+      "loss": 0.5195,
+      "step": 16110
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.4125906495911125e-05,
+      "loss": 0.7851,
+      "step": 16111
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.4123334876305097e-05,
+      "loss": 0.7697,
+      "step": 16112
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.4120763256699072e-05,
+      "loss": 0.4538,
+      "step": 16113
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.4118191637093043e-05,
+      "loss": 0.5593,
+      "step": 16114
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.4115620017487013e-05,
+      "loss": 0.8668,
+      "step": 16115
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.4113048397880987e-05,
+      "loss": 0.4116,
+      "step": 16116
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.4110476778274959e-05,
+      "loss": 0.9321,
+      "step": 16117
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.410790515866893e-05,
+      "loss": 0.477,
+      "step": 16118
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.4105333539062903e-05,
+      "loss": 0.8586,
+      "step": 16119
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.4102761919456875e-05,
+      "loss": 0.6829,
+      "step": 16120
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.4100190299850846e-05,
+      "loss": 0.4438,
+      "step": 16121
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.409761868024482e-05,
+      "loss": 0.5666,
+      "step": 16122
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.409504706063879e-05,
+      "loss": 0.707,
+      "step": 16123
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.4092475441032762e-05,
+      "loss": 0.5186,
+      "step": 16124
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.4089903821426737e-05,
+      "loss": 0.9365,
+      "step": 16125
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.4087332201820708e-05,
+      "loss": 1.2377,
+      "step": 16126
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.408476058221468e-05,
+      "loss": 0.556,
+      "step": 16127
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.4082188962608652e-05,
+      "loss": 0.8377,
+      "step": 16128
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.4079617343002624e-05,
+      "loss": 0.5648,
+      "step": 16129
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.4077045723396595e-05,
+      "loss": 0.7056,
+      "step": 16130
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.407447410379057e-05,
+      "loss": 0.6314,
+      "step": 16131
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.407190248418454e-05,
+      "loss": 0.4484,
+      "step": 16132
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.4069330864578511e-05,
+      "loss": 0.6214,
+      "step": 16133
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.4066759244972486e-05,
+      "loss": 0.7559,
+      "step": 16134
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.4064187625366457e-05,
+      "loss": 0.96,
+      "step": 16135
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.4061616005760429e-05,
+      "loss": 0.8581,
+      "step": 16136
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.4059044386154402e-05,
+      "loss": 0.8043,
+      "step": 16137
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.4056472766548373e-05,
+      "loss": 0.6151,
+      "step": 16138
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.4053901146942344e-05,
+      "loss": 0.6054,
+      "step": 16139
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.4051329527336318e-05,
+      "loss": 0.861,
+      "step": 16140
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.4048757907730289e-05,
+      "loss": 0.4404,
+      "step": 16141
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.404618628812426e-05,
+      "loss": 0.7871,
+      "step": 16142
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.4043614668518235e-05,
+      "loss": 0.9383,
+      "step": 16143
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.4041043048912206e-05,
+      "loss": 0.5257,
+      "step": 16144
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.4038471429306178e-05,
+      "loss": 0.6564,
+      "step": 16145
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.4035899809700151e-05,
+      "loss": 0.6842,
+      "step": 16146
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.4033328190094122e-05,
+      "loss": 0.6925,
+      "step": 16147
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.4030756570488094e-05,
+      "loss": 0.5626,
+      "step": 16148
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.4028184950882067e-05,
+      "loss": 0.699,
+      "step": 16149
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.4025613331276038e-05,
+      "loss": 0.7236,
+      "step": 16150
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.402304171167001e-05,
+      "loss": 1.045,
+      "step": 16151
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.4020470092063984e-05,
+      "loss": 0.5756,
+      "step": 16152
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.4017898472457956e-05,
+      "loss": 1.0601,
+      "step": 16153
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.4015326852851925e-05,
+      "loss": 0.6933,
+      "step": 16154
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.40127552332459e-05,
+      "loss": 0.455,
+      "step": 16155
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.4010183613639871e-05,
+      "loss": 0.7956,
+      "step": 16156
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.4007611994033843e-05,
+      "loss": 0.7172,
+      "step": 16157
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.4005040374427816e-05,
+      "loss": 0.6978,
+      "step": 16158
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.4002468754821787e-05,
+      "loss": 0.5623,
+      "step": 16159
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.3999897135215759e-05,
+      "loss": 0.7991,
+      "step": 16160
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.3997325515609733e-05,
+      "loss": 0.7711,
+      "step": 16161
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.3994753896003703e-05,
+      "loss": 0.9254,
+      "step": 16162
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.3992182276397674e-05,
+      "loss": 0.4584,
+      "step": 16163
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.398961065679165e-05,
+      "loss": 0.5108,
+      "step": 16164
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.398703903718562e-05,
+      "loss": 0.6167,
+      "step": 16165
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.3984467417579592e-05,
+      "loss": 0.8028,
+      "step": 16166
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.3981895797973565e-05,
+      "loss": 0.6786,
+      "step": 16167
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.3979324178367536e-05,
+      "loss": 0.7339,
+      "step": 16168
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.3976752558761508e-05,
+      "loss": 0.8578,
+      "step": 16169
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.3974180939155483e-05,
+      "loss": 0.4427,
+      "step": 16170
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.3971609319549452e-05,
+      "loss": 0.5942,
+      "step": 16171
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.3969037699943424e-05,
+      "loss": 0.8603,
+      "step": 16172
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.3966466080337398e-05,
+      "loss": 0.439,
+      "step": 16173
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.396389446073137e-05,
+      "loss": 0.5124,
+      "step": 16174
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.3961322841125341e-05,
+      "loss": 0.7366,
+      "step": 16175
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.3958751221519314e-05,
+      "loss": 0.6823,
+      "step": 16176
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.3956179601913286e-05,
+      "loss": 0.7457,
+      "step": 16177
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.3953607982307257e-05,
+      "loss": 0.6006,
+      "step": 16178
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.395103636270123e-05,
+      "loss": 0.7677,
+      "step": 16179
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.3948464743095201e-05,
+      "loss": 0.5771,
+      "step": 16180
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.3945893123489173e-05,
+      "loss": 0.5069,
+      "step": 16181
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.3943321503883148e-05,
+      "loss": 0.4743,
+      "step": 16182
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.3940749884277119e-05,
+      "loss": 0.8199,
+      "step": 16183
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.393817826467109e-05,
+      "loss": 0.5384,
+      "step": 16184
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.3935606645065063e-05,
+      "loss": 0.7479,
+      "step": 16185
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.3933035025459035e-05,
+      "loss": 0.6944,
+      "step": 16186
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.3930463405853006e-05,
+      "loss": 0.7439,
+      "step": 16187
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.392789178624698e-05,
+      "loss": 0.6411,
+      "step": 16188
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.392532016664095e-05,
+      "loss": 0.9911,
+      "step": 16189
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.3922748547034922e-05,
+      "loss": 0.607,
+      "step": 16190
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.3920176927428897e-05,
+      "loss": 0.6132,
+      "step": 16191
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.3917605307822868e-05,
+      "loss": 0.2376,
+      "step": 16192
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.3915033688216838e-05,
+      "loss": 0.4748,
+      "step": 16193
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.3912462068610813e-05,
+      "loss": 0.7111,
+      "step": 16194
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.3909890449004784e-05,
+      "loss": 0.573,
+      "step": 16195
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.3907318829398755e-05,
+      "loss": 0.563,
+      "step": 16196
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.3904747209792728e-05,
+      "loss": 0.6579,
+      "step": 16197
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.39021755901867e-05,
+      "loss": 0.7524,
+      "step": 16198
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.3899603970580671e-05,
+      "loss": 0.4762,
+      "step": 16199
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.3897032350974646e-05,
+      "loss": 0.7803,
+      "step": 16200
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.3894460731368616e-05,
+      "loss": 0.8126,
+      "step": 16201
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.3891889111762587e-05,
+      "loss": 0.4364,
+      "step": 16202
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.3889317492156562e-05,
+      "loss": 0.5652,
+      "step": 16203
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.3886745872550533e-05,
+      "loss": 0.7209,
+      "step": 16204
+    },
+    {
+      "epoch": 3.0,
+      "eval_accuracy": 0.7120788710881225,
+      "eval_loss": 0.8852449655532837,
+      "eval_runtime": 40.5459,
+      "eval_samples_per_second": 1420.932,
+      "eval_steps_per_second": 44.419,
+      "step": 16204
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.3884174252944505e-05,
+      "loss": 0.64,
+      "step": 16205
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.3881602633338478e-05,
+      "loss": 0.6548,
+      "step": 16206
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.3879031013732449e-05,
+      "loss": 0.8608,
+      "step": 16207
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.387645939412642e-05,
+      "loss": 0.5368,
+      "step": 16208
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.3873887774520394e-05,
+      "loss": 0.6909,
+      "step": 16209
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.3871316154914365e-05,
+      "loss": 0.468,
+      "step": 16210
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.3868744535308336e-05,
+      "loss": 0.4124,
+      "step": 16211
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.3866172915702311e-05,
+      "loss": 0.5281,
+      "step": 16212
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.3863601296096282e-05,
+      "loss": 0.6964,
+      "step": 16213
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.3861029676490254e-05,
+      "loss": 0.6819,
+      "step": 16214
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.3858458056884227e-05,
+      "loss": 0.4977,
+      "step": 16215
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.3855886437278198e-05,
+      "loss": 0.5155,
+      "step": 16216
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.385331481767217e-05,
+      "loss": 0.7274,
+      "step": 16217
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.3850743198066143e-05,
+      "loss": 0.3813,
+      "step": 16218
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.3848171578460114e-05,
+      "loss": 0.6184,
+      "step": 16219
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.3845599958854085e-05,
+      "loss": 0.7081,
+      "step": 16220
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.384302833924806e-05,
+      "loss": 0.6132,
+      "step": 16221
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.3840456719642032e-05,
+      "loss": 0.295,
+      "step": 16222
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.3837885100036003e-05,
+      "loss": 0.8474,
+      "step": 16223
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.3835313480429976e-05,
+      "loss": 0.7475,
+      "step": 16224
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.3832741860823947e-05,
+      "loss": 0.3988,
+      "step": 16225
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.3830170241217919e-05,
+      "loss": 0.4577,
+      "step": 16226
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.3827598621611892e-05,
+      "loss": 0.3312,
+      "step": 16227
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.3825027002005863e-05,
+      "loss": 0.5551,
+      "step": 16228
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.3822455382399835e-05,
+      "loss": 0.4107,
+      "step": 16229
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.381988376279381e-05,
+      "loss": 0.7793,
+      "step": 16230
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.381731214318778e-05,
+      "loss": 0.8047,
+      "step": 16231
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.381474052358175e-05,
+      "loss": 0.6416,
+      "step": 16232
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.3812168903975725e-05,
+      "loss": 0.689,
+      "step": 16233
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.3809597284369697e-05,
+      "loss": 0.6505,
+      "step": 16234
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.3807025664763668e-05,
+      "loss": 0.305,
+      "step": 16235
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.3804454045157641e-05,
+      "loss": 0.7354,
+      "step": 16236
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.3801882425551612e-05,
+      "loss": 0.9658,
+      "step": 16237
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.3799310805945584e-05,
+      "loss": 0.6112,
+      "step": 16238
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.3796739186339559e-05,
+      "loss": 0.7227,
+      "step": 16239
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.3794167566733528e-05,
+      "loss": 0.4225,
+      "step": 16240
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.37915959471275e-05,
+      "loss": 0.5507,
+      "step": 16241
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.3789024327521474e-05,
+      "loss": 0.6306,
+      "step": 16242
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.3786452707915446e-05,
+      "loss": 0.7295,
+      "step": 16243
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.3783881088309417e-05,
+      "loss": 0.843,
+      "step": 16244
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.378130946870339e-05,
+      "loss": 1.0724,
+      "step": 16245
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.3778737849097362e-05,
+      "loss": 0.4113,
+      "step": 16246
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.3776166229491333e-05,
+      "loss": 0.6361,
+      "step": 16247
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.3773594609885306e-05,
+      "loss": 0.427,
+      "step": 16248
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.3771022990279277e-05,
+      "loss": 0.4452,
+      "step": 16249
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.3768451370673249e-05,
+      "loss": 0.7445,
+      "step": 16250
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.3765879751067224e-05,
+      "loss": 0.9203,
+      "step": 16251
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.3763308131461195e-05,
+      "loss": 0.597,
+      "step": 16252
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.3760736511855166e-05,
+      "loss": 0.625,
+      "step": 16253
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.375816489224914e-05,
+      "loss": 0.5354,
+      "step": 16254
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.375559327264311e-05,
+      "loss": 0.497,
+      "step": 16255
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.3753021653037086e-05,
+      "loss": 0.5122,
+      "step": 16256
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.3750450033431055e-05,
+      "loss": 0.7044,
+      "step": 16257
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.3747878413825027e-05,
+      "loss": 0.466,
+      "step": 16258
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.3745306794219001e-05,
+      "loss": 0.5225,
+      "step": 16259
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.3742735174612973e-05,
+      "loss": 0.4263,
+      "step": 16260
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.3740163555006944e-05,
+      "loss": 0.5606,
+      "step": 16261
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.3737591935400917e-05,
+      "loss": 0.5938,
+      "step": 16262
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.3735020315794889e-05,
+      "loss": 0.7381,
+      "step": 16263
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.373244869618886e-05,
+      "loss": 1.1143,
+      "step": 16264
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.3729877076582833e-05,
+      "loss": 0.6605,
+      "step": 16265
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.3727305456976804e-05,
+      "loss": 0.8136,
+      "step": 16266
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.3724733837370776e-05,
+      "loss": 0.4369,
+      "step": 16267
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.372216221776475e-05,
+      "loss": 0.7481,
+      "step": 16268
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.3719590598158722e-05,
+      "loss": 0.813,
+      "step": 16269
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.3717018978552693e-05,
+      "loss": 0.5063,
+      "step": 16270
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.3714447358946666e-05,
+      "loss": 0.8533,
+      "step": 16271
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.3711875739340638e-05,
+      "loss": 0.4015,
+      "step": 16272
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.370930411973461e-05,
+      "loss": 0.8074,
+      "step": 16273
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.3706732500128582e-05,
+      "loss": 0.6023,
+      "step": 16274
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.3704160880522554e-05,
+      "loss": 0.7145,
+      "step": 16275
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.3701589260916525e-05,
+      "loss": 0.6971,
+      "step": 16276
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.36990176413105e-05,
+      "loss": 0.3472,
+      "step": 16277
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.3696446021704471e-05,
+      "loss": 0.5048,
+      "step": 16278
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.3693874402098441e-05,
+      "loss": 0.7646,
+      "step": 16279
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.3691302782492416e-05,
+      "loss": 0.5221,
+      "step": 16280
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.3688731162886387e-05,
+      "loss": 0.6246,
+      "step": 16281
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.3686159543280358e-05,
+      "loss": 0.4432,
+      "step": 16282
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.3683587923674331e-05,
+      "loss": 0.4273,
+      "step": 16283
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.3681016304068303e-05,
+      "loss": 0.506,
+      "step": 16284
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.3678444684462274e-05,
+      "loss": 0.8673,
+      "step": 16285
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.3675873064856249e-05,
+      "loss": 0.6827,
+      "step": 16286
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.3673301445250219e-05,
+      "loss": 0.6277,
+      "step": 16287
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.367072982564419e-05,
+      "loss": 0.4346,
+      "step": 16288
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.3668158206038165e-05,
+      "loss": 0.3708,
+      "step": 16289
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.3665586586432136e-05,
+      "loss": 0.4202,
+      "step": 16290
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.3663014966826108e-05,
+      "loss": 0.5654,
+      "step": 16291
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.366044334722008e-05,
+      "loss": 0.6488,
+      "step": 16292
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.3657871727614052e-05,
+      "loss": 0.6008,
+      "step": 16293
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.3655300108008023e-05,
+      "loss": 0.5975,
+      "step": 16294
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.3652728488401998e-05,
+      "loss": 0.9557,
+      "step": 16295
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.3650156868795968e-05,
+      "loss": 0.6539,
+      "step": 16296
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.364758524918994e-05,
+      "loss": 0.7383,
+      "step": 16297
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.3645013629583914e-05,
+      "loss": 0.5361,
+      "step": 16298
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.3642442009977885e-05,
+      "loss": 0.4409,
+      "step": 16299
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.3639870390371857e-05,
+      "loss": 0.554,
+      "step": 16300
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.363729877076583e-05,
+      "loss": 0.6484,
+      "step": 16301
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.3634727151159801e-05,
+      "loss": 0.4103,
+      "step": 16302
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.3632155531553773e-05,
+      "loss": 0.7728,
+      "step": 16303
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.3629583911947746e-05,
+      "loss": 0.6692,
+      "step": 16304
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.3627012292341717e-05,
+      "loss": 0.6866,
+      "step": 16305
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.3624440672735688e-05,
+      "loss": 0.7904,
+      "step": 16306
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.3621869053129663e-05,
+      "loss": 0.4338,
+      "step": 16307
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.3619297433523635e-05,
+      "loss": 0.6003,
+      "step": 16308
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.3616725813917606e-05,
+      "loss": 0.3607,
+      "step": 16309
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.3614154194311579e-05,
+      "loss": 0.8651,
+      "step": 16310
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.361158257470555e-05,
+      "loss": 0.731,
+      "step": 16311
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.3609010955099522e-05,
+      "loss": 0.6756,
+      "step": 16312
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.3606439335493495e-05,
+      "loss": 0.5542,
+      "step": 16313
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.3603867715887466e-05,
+      "loss": 0.8595,
+      "step": 16314
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.3601296096281438e-05,
+      "loss": 0.6041,
+      "step": 16315
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.3598724476675412e-05,
+      "loss": 0.5662,
+      "step": 16316
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.3596152857069384e-05,
+      "loss": 0.6312,
+      "step": 16317
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.3593581237463353e-05,
+      "loss": 0.557,
+      "step": 16318
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.3591009617857328e-05,
+      "loss": 0.7617,
+      "step": 16319
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.35884379982513e-05,
+      "loss": 0.8046,
+      "step": 16320
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.3585866378645271e-05,
+      "loss": 0.5997,
+      "step": 16321
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.3583294759039244e-05,
+      "loss": 0.3951,
+      "step": 16322
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.3580723139433215e-05,
+      "loss": 0.5215,
+      "step": 16323
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.3578151519827187e-05,
+      "loss": 0.6395,
+      "step": 16324
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.3575579900221162e-05,
+      "loss": 0.6283,
+      "step": 16325
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.3573008280615131e-05,
+      "loss": 0.5434,
+      "step": 16326
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.3570436661009103e-05,
+      "loss": 0.4962,
+      "step": 16327
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.3567865041403077e-05,
+      "loss": 0.6177,
+      "step": 16328
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.3565293421797049e-05,
+      "loss": 0.7309,
+      "step": 16329
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.356272180219102e-05,
+      "loss": 0.8718,
+      "step": 16330
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.3560150182584993e-05,
+      "loss": 0.7808,
+      "step": 16331
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.3557578562978965e-05,
+      "loss": 0.9136,
+      "step": 16332
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.3555006943372936e-05,
+      "loss": 0.5886,
+      "step": 16333
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.355243532376691e-05,
+      "loss": 0.6731,
+      "step": 16334
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.354986370416088e-05,
+      "loss": 0.8454,
+      "step": 16335
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.3547292084554852e-05,
+      "loss": 0.5142,
+      "step": 16336
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.3544720464948827e-05,
+      "loss": 1.0538,
+      "step": 16337
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.3542148845342798e-05,
+      "loss": 0.3662,
+      "step": 16338
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.353957722573677e-05,
+      "loss": 0.6429,
+      "step": 16339
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.3537005606130742e-05,
+      "loss": 0.3565,
+      "step": 16340
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.3534433986524714e-05,
+      "loss": 0.4709,
+      "step": 16341
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.3531862366918685e-05,
+      "loss": 0.5855,
+      "step": 16342
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.3529290747312658e-05,
+      "loss": 0.4881,
+      "step": 16343
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.352671912770663e-05,
+      "loss": 0.6417,
+      "step": 16344
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.3524147508100601e-05,
+      "loss": 0.5464,
+      "step": 16345
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.3521575888494576e-05,
+      "loss": 0.8349,
+      "step": 16346
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.3519004268888547e-05,
+      "loss": 0.3923,
+      "step": 16347
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.3516432649282519e-05,
+      "loss": 0.634,
+      "step": 16348
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.3513861029676492e-05,
+      "loss": 0.5289,
+      "step": 16349
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.3511289410070463e-05,
+      "loss": 0.9734,
+      "step": 16350
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.3508717790464434e-05,
+      "loss": 0.5586,
+      "step": 16351
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.3506146170858407e-05,
+      "loss": 0.5628,
+      "step": 16352
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.3503574551252379e-05,
+      "loss": 0.9823,
+      "step": 16353
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.350100293164635e-05,
+      "loss": 0.5637,
+      "step": 16354
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.3498431312040325e-05,
+      "loss": 0.6987,
+      "step": 16355
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.3495859692434296e-05,
+      "loss": 0.6454,
+      "step": 16356
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.3493288072828266e-05,
+      "loss": 0.3988,
+      "step": 16357
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.349071645322224e-05,
+      "loss": 0.7461,
+      "step": 16358
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.3488144833616212e-05,
+      "loss": 0.5999,
+      "step": 16359
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.3485573214010184e-05,
+      "loss": 0.6183,
+      "step": 16360
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.3483001594404157e-05,
+      "loss": 0.4861,
+      "step": 16361
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.3480429974798128e-05,
+      "loss": 0.8247,
+      "step": 16362
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.34778583551921e-05,
+      "loss": 0.7436,
+      "step": 16363
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.3475286735586074e-05,
+      "loss": 0.807,
+      "step": 16364
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.3472715115980044e-05,
+      "loss": 0.5927,
+      "step": 16365
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.3470143496374015e-05,
+      "loss": 0.8559,
+      "step": 16366
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.346757187676799e-05,
+      "loss": 0.5832,
+      "step": 16367
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.3465000257161961e-05,
+      "loss": 0.819,
+      "step": 16368
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.3462428637555933e-05,
+      "loss": 0.6881,
+      "step": 16369
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.3459857017949906e-05,
+      "loss": 0.6215,
+      "step": 16370
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.3457285398343877e-05,
+      "loss": 0.6221,
+      "step": 16371
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.3454713778737849e-05,
+      "loss": 0.5269,
+      "step": 16372
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.3452142159131822e-05,
+      "loss": 0.5588,
+      "step": 16373
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.3449570539525793e-05,
+      "loss": 0.371,
+      "step": 16374
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.3446998919919764e-05,
+      "loss": 0.3062,
+      "step": 16375
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.344442730031374e-05,
+      "loss": 0.4722,
+      "step": 16376
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.344185568070771e-05,
+      "loss": 0.3607,
+      "step": 16377
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.3439284061101682e-05,
+      "loss": 0.8985,
+      "step": 16378
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.3436712441495655e-05,
+      "loss": 0.6272,
+      "step": 16379
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.3434140821889626e-05,
+      "loss": 0.5278,
+      "step": 16380
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.3431569202283598e-05,
+      "loss": 0.9783,
+      "step": 16381
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.3428997582677571e-05,
+      "loss": 0.6143,
+      "step": 16382
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.3426425963071542e-05,
+      "loss": 0.5919,
+      "step": 16383
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.3423854343465514e-05,
+      "loss": 0.6554,
+      "step": 16384
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.3421282723859488e-05,
+      "loss": 0.4491,
+      "step": 16385
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.341871110425346e-05,
+      "loss": 0.3565,
+      "step": 16386
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.341613948464743e-05,
+      "loss": 0.4104,
+      "step": 16387
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.3413567865041404e-05,
+      "loss": 0.6125,
+      "step": 16388
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.3410996245435376e-05,
+      "loss": 0.5479,
+      "step": 16389
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.3408424625829347e-05,
+      "loss": 0.6382,
+      "step": 16390
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.340585300622332e-05,
+      "loss": 0.8709,
+      "step": 16391
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.3403281386617291e-05,
+      "loss": 0.4705,
+      "step": 16392
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.3400709767011263e-05,
+      "loss": 0.5846,
+      "step": 16393
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 1.3398138147405238e-05,
+      "loss": 0.389,
+      "step": 16394
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 1.3395566527799209e-05,
+      "loss": 0.5561,
+      "step": 16395
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 1.3392994908193179e-05,
+      "loss": 0.633,
+      "step": 16396
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 1.3390423288587153e-05,
+      "loss": 0.4204,
+      "step": 16397
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 1.3387851668981125e-05,
+      "loss": 0.7255,
+      "step": 16398
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 1.3385280049375096e-05,
+      "loss": 0.4171,
+      "step": 16399
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 1.338270842976907e-05,
+      "loss": 0.5923,
+      "step": 16400
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 1.338013681016304e-05,
+      "loss": 0.513,
+      "step": 16401
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 1.3377565190557012e-05,
+      "loss": 0.7391,
+      "step": 16402
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 1.3374993570950987e-05,
+      "loss": 0.4881,
+      "step": 16403
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 1.3372421951344956e-05,
+      "loss": 0.4627,
+      "step": 16404
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 1.3369850331738931e-05,
+      "loss": 0.4415,
+      "step": 16405
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 1.3367278712132903e-05,
+      "loss": 0.5905,
+      "step": 16406
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 1.3364707092526874e-05,
+      "loss": 0.6939,
+      "step": 16407
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 1.3362135472920847e-05,
+      "loss": 0.5005,
+      "step": 16408
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 1.3359563853314818e-05,
+      "loss": 0.3628,
+      "step": 16409
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 1.335699223370879e-05,
+      "loss": 0.3813,
+      "step": 16410
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 1.3354420614102765e-05,
+      "loss": 0.7117,
+      "step": 16411
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 1.3351848994496734e-05,
+      "loss": 0.7206,
+      "step": 16412
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 1.3349277374890706e-05,
+      "loss": 0.5941,
+      "step": 16413
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 1.334670575528468e-05,
+      "loss": 0.5479,
+      "step": 16414
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 1.3344134135678652e-05,
+      "loss": 0.5186,
+      "step": 16415
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 1.3341562516072623e-05,
+      "loss": 0.5745,
+      "step": 16416
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 1.3338990896466596e-05,
+      "loss": 0.5269,
+      "step": 16417
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 1.3336419276860568e-05,
+      "loss": 0.6116,
+      "step": 16418
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 1.3333847657254539e-05,
+      "loss": 0.5696,
+      "step": 16419
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 1.3331276037648514e-05,
+      "loss": 0.3213,
+      "step": 16420
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 1.3328704418042483e-05,
+      "loss": 0.4819,
+      "step": 16421
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 1.3326132798436455e-05,
+      "loss": 0.6716,
+      "step": 16422
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 1.332356117883043e-05,
+      "loss": 0.4293,
+      "step": 16423
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 1.3320989559224401e-05,
+      "loss": 0.3799,
+      "step": 16424
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 1.3318417939618372e-05,
+      "loss": 0.8745,
+      "step": 16425
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 1.3315846320012345e-05,
+      "loss": 0.6309,
+      "step": 16426
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 1.3313274700406317e-05,
+      "loss": 0.7406,
+      "step": 16427
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 1.3310703080800288e-05,
+      "loss": 0.5376,
+      "step": 16428
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 1.3308131461194261e-05,
+      "loss": 0.3492,
+      "step": 16429
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 1.3305559841588233e-05,
+      "loss": 0.6578,
+      "step": 16430
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 1.3302988221982204e-05,
+      "loss": 0.3821,
+      "step": 16431
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 1.3300416602376179e-05,
+      "loss": 1.0182,
+      "step": 16432
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 1.329784498277015e-05,
+      "loss": 0.6908,
+      "step": 16433
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 1.3295273363164122e-05,
+      "loss": 0.3166,
+      "step": 16434
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 1.3292701743558095e-05,
+      "loss": 0.3162,
+      "step": 16435
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 1.3290130123952066e-05,
+      "loss": 0.497,
+      "step": 16436
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 1.3287558504346037e-05,
+      "loss": 0.6543,
+      "step": 16437
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 1.328498688474001e-05,
+      "loss": 0.593,
+      "step": 16438
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 1.3282415265133982e-05,
+      "loss": 0.4949,
+      "step": 16439
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 1.3279843645527953e-05,
+      "loss": 0.7048,
+      "step": 16440
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 1.3277272025921928e-05,
+      "loss": 0.7223,
+      "step": 16441
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 1.32747004063159e-05,
+      "loss": 0.3551,
+      "step": 16442
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 1.3272128786709869e-05,
+      "loss": 0.3564,
+      "step": 16443
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 1.3269557167103844e-05,
+      "loss": 0.769,
+      "step": 16444
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 1.3266985547497815e-05,
+      "loss": 0.6403,
+      "step": 16445
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 1.3264413927891787e-05,
+      "loss": 0.55,
+      "step": 16446
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 1.326184230828576e-05,
+      "loss": 0.7197,
+      "step": 16447
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 1.3259270688679731e-05,
+      "loss": 0.4721,
+      "step": 16448
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 1.3256699069073702e-05,
+      "loss": 0.5409,
+      "step": 16449
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 1.3254127449467677e-05,
+      "loss": 1.1209,
+      "step": 16450
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 1.3251555829861647e-05,
+      "loss": 0.5431,
+      "step": 16451
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 1.3248984210255618e-05,
+      "loss": 0.9072,
+      "step": 16452
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 1.3246412590649593e-05,
+      "loss": 0.343,
+      "step": 16453
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 1.3243840971043564e-05,
+      "loss": 0.7257,
+      "step": 16454
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 1.3241269351437536e-05,
+      "loss": 0.735,
+      "step": 16455
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 1.3238697731831509e-05,
+      "loss": 0.6053,
+      "step": 16456
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 1.323612611222548e-05,
+      "loss": 0.6505,
+      "step": 16457
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 1.3233554492619452e-05,
+      "loss": 0.6035,
+      "step": 16458
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 1.3230982873013426e-05,
+      "loss": 0.6556,
+      "step": 16459
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 1.3228411253407396e-05,
+      "loss": 0.6165,
+      "step": 16460
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 1.3225839633801367e-05,
+      "loss": 0.5069,
+      "step": 16461
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 1.3223268014195342e-05,
+      "loss": 0.5915,
+      "step": 16462
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 1.3220696394589314e-05,
+      "loss": 0.6719,
+      "step": 16463
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 1.3218124774983285e-05,
+      "loss": 0.3345,
+      "step": 16464
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 1.3215553155377258e-05,
+      "loss": 0.4165,
+      "step": 16465
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 1.321298153577123e-05,
+      "loss": 0.8346,
+      "step": 16466
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 1.32104099161652e-05,
+      "loss": 0.4267,
+      "step": 16467
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 1.3207838296559174e-05,
+      "loss": 0.4736,
+      "step": 16468
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 1.3205266676953145e-05,
+      "loss": 0.4668,
+      "step": 16469
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 1.3202695057347117e-05,
+      "loss": 0.432,
+      "step": 16470
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 1.3200123437741091e-05,
+      "loss": 0.6507,
+      "step": 16471
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 1.3197551818135063e-05,
+      "loss": 0.8055,
+      "step": 16472
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 1.3194980198529034e-05,
+      "loss": 0.6346,
+      "step": 16473
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 1.3192408578923007e-05,
+      "loss": 0.5301,
+      "step": 16474
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 1.3189836959316979e-05,
+      "loss": 0.5123,
+      "step": 16475
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 1.318726533971095e-05,
+      "loss": 0.5116,
+      "step": 16476
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 1.3184693720104923e-05,
+      "loss": 0.4876,
+      "step": 16477
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 1.3182122100498894e-05,
+      "loss": 0.7646,
+      "step": 16478
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 1.3179550480892866e-05,
+      "loss": 0.7448,
+      "step": 16479
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 1.317697886128684e-05,
+      "loss": 0.6569,
+      "step": 16480
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 1.3174407241680812e-05,
+      "loss": 0.3881,
+      "step": 16481
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 1.3171835622074782e-05,
+      "loss": 0.4228,
+      "step": 16482
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 1.3169264002468756e-05,
+      "loss": 0.5183,
+      "step": 16483
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 1.3166692382862728e-05,
+      "loss": 0.7849,
+      "step": 16484
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 1.31641207632567e-05,
+      "loss": 0.3215,
+      "step": 16485
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 1.3161549143650672e-05,
+      "loss": 0.5951,
+      "step": 16486
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 1.3158977524044644e-05,
+      "loss": 0.7523,
+      "step": 16487
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 1.3156405904438615e-05,
+      "loss": 0.459,
+      "step": 16488
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 1.315383428483259e-05,
+      "loss": 0.5145,
+      "step": 16489
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 1.315126266522656e-05,
+      "loss": 0.4063,
+      "step": 16490
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 1.3148691045620531e-05,
+      "loss": 0.5927,
+      "step": 16491
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 1.3146119426014506e-05,
+      "loss": 0.6025,
+      "step": 16492
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 1.3143547806408477e-05,
+      "loss": 1.1262,
+      "step": 16493
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 1.3140976186802448e-05,
+      "loss": 0.7918,
+      "step": 16494
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 1.3138404567196421e-05,
+      "loss": 0.5878,
+      "step": 16495
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 1.3135832947590393e-05,
+      "loss": 0.5826,
+      "step": 16496
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 1.3133261327984364e-05,
+      "loss": 0.5988,
+      "step": 16497
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 1.3130689708378337e-05,
+      "loss": 0.7566,
+      "step": 16498
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 1.3128118088772309e-05,
+      "loss": 0.3761,
+      "step": 16499
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 1.312554646916628e-05,
+      "loss": 0.6875,
+      "step": 16500
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 1.3122974849560255e-05,
+      "loss": 0.2919,
+      "step": 16501
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 1.3120403229954226e-05,
+      "loss": 0.4263,
+      "step": 16502
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 1.3117831610348198e-05,
+      "loss": 0.6802,
+      "step": 16503
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 1.311525999074217e-05,
+      "loss": 0.9743,
+      "step": 16504
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 1.3112688371136142e-05,
+      "loss": 0.3365,
+      "step": 16505
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 1.3110116751530113e-05,
+      "loss": 0.8476,
+      "step": 16506
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 1.3107545131924086e-05,
+      "loss": 0.8269,
+      "step": 16507
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 1.3104973512318058e-05,
+      "loss": 0.6112,
+      "step": 16508
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 1.310240189271203e-05,
+      "loss": 0.7662,
+      "step": 16509
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 1.3099830273106004e-05,
+      "loss": 0.5632,
+      "step": 16510
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 1.3097258653499975e-05,
+      "loss": 0.5058,
+      "step": 16511
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 1.3094687033893947e-05,
+      "loss": 0.5199,
+      "step": 16512
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 1.309211541428792e-05,
+      "loss": 0.5961,
+      "step": 16513
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 1.3089543794681891e-05,
+      "loss": 0.749,
+      "step": 16514
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 1.3086972175075863e-05,
+      "loss": 0.5099,
+      "step": 16515
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 1.3084400555469836e-05,
+      "loss": 0.5637,
+      "step": 16516
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 1.3081828935863807e-05,
+      "loss": 0.3893,
+      "step": 16517
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 1.3079257316257778e-05,
+      "loss": 0.759,
+      "step": 16518
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 1.3076685696651753e-05,
+      "loss": 0.6676,
+      "step": 16519
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 1.3074114077045725e-05,
+      "loss": 0.607,
+      "step": 16520
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 1.3071542457439694e-05,
+      "loss": 0.454,
+      "step": 16521
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 1.3068970837833669e-05,
+      "loss": 0.8592,
+      "step": 16522
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 1.306639921822764e-05,
+      "loss": 0.3533,
+      "step": 16523
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 1.3063827598621612e-05,
+      "loss": 0.6321,
+      "step": 16524
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 1.3061255979015585e-05,
+      "loss": 0.6127,
+      "step": 16525
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 1.3058684359409556e-05,
+      "loss": 0.7429,
+      "step": 16526
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 1.3056112739803528e-05,
+      "loss": 0.4413,
+      "step": 16527
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 1.3053541120197502e-05,
+      "loss": 0.617,
+      "step": 16528
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 1.3050969500591472e-05,
+      "loss": 0.5896,
+      "step": 16529
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 1.3048397880985443e-05,
+      "loss": 0.467,
+      "step": 16530
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 1.3045826261379418e-05,
+      "loss": 0.8026,
+      "step": 16531
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 1.304325464177339e-05,
+      "loss": 0.6652,
+      "step": 16532
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 1.3040683022167361e-05,
+      "loss": 0.6896,
+      "step": 16533
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 1.3038111402561334e-05,
+      "loss": 0.4758,
+      "step": 16534
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 1.3035539782955305e-05,
+      "loss": 0.3352,
+      "step": 16535
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 1.3032968163349277e-05,
+      "loss": 1.0131,
+      "step": 16536
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 1.303039654374325e-05,
+      "loss": 0.4792,
+      "step": 16537
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 1.3027824924137221e-05,
+      "loss": 0.3914,
+      "step": 16538
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 1.3025253304531193e-05,
+      "loss": 0.7659,
+      "step": 16539
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 1.3022681684925167e-05,
+      "loss": 0.6262,
+      "step": 16540
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 1.3020110065319139e-05,
+      "loss": 0.6794,
+      "step": 16541
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 1.301753844571311e-05,
+      "loss": 0.493,
+      "step": 16542
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 1.3014966826107083e-05,
+      "loss": 0.4442,
+      "step": 16543
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 1.3012395206501055e-05,
+      "loss": 0.4992,
+      "step": 16544
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 1.3009823586895026e-05,
+      "loss": 0.4594,
+      "step": 16545
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 1.3007251967288999e-05,
+      "loss": 0.474,
+      "step": 16546
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 1.300468034768297e-05,
+      "loss": 0.6231,
+      "step": 16547
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 1.3002108728076942e-05,
+      "loss": 0.8099,
+      "step": 16548
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 1.2999537108470917e-05,
+      "loss": 0.4525,
+      "step": 16549
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 1.2996965488864888e-05,
+      "loss": 0.5592,
+      "step": 16550
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 1.2994393869258858e-05,
+      "loss": 0.4573,
+      "step": 16551
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 1.2991822249652832e-05,
+      "loss": 0.3984,
+      "step": 16552
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 1.2989250630046804e-05,
+      "loss": 0.4936,
+      "step": 16553
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 1.2986679010440775e-05,
+      "loss": 0.6407,
+      "step": 16554
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 1.2984107390834748e-05,
+      "loss": 0.6353,
+      "step": 16555
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 1.298153577122872e-05,
+      "loss": 0.8219,
+      "step": 16556
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 1.2978964151622694e-05,
+      "loss": 0.6802,
+      "step": 16557
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 1.2976392532016666e-05,
+      "loss": 0.5652,
+      "step": 16558
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 1.2973820912410637e-05,
+      "loss": 0.7015,
+      "step": 16559
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 1.297124929280461e-05,
+      "loss": 0.3725,
+      "step": 16560
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 1.2968677673198582e-05,
+      "loss": 0.5119,
+      "step": 16561
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 1.2966106053592553e-05,
+      "loss": 0.6766,
+      "step": 16562
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 1.2963534433986526e-05,
+      "loss": 0.4319,
+      "step": 16563
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 1.2960962814380497e-05,
+      "loss": 0.3987,
+      "step": 16564
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 1.2958391194774469e-05,
+      "loss": 0.5615,
+      "step": 16565
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 1.2955819575168444e-05,
+      "loss": 1.0284,
+      "step": 16566
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 1.2953247955562415e-05,
+      "loss": 0.88,
+      "step": 16567
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 1.2950676335956385e-05,
+      "loss": 0.477,
+      "step": 16568
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 1.294810471635036e-05,
+      "loss": 0.537,
+      "step": 16569
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 1.294553309674433e-05,
+      "loss": 0.362,
+      "step": 16570
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 1.2942961477138302e-05,
+      "loss": 0.6655,
+      "step": 16571
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 1.2940389857532275e-05,
+      "loss": 0.6702,
+      "step": 16572
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 1.2937818237926247e-05,
+      "loss": 0.5422,
+      "step": 16573
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 1.2935246618320218e-05,
+      "loss": 0.7132,
+      "step": 16574
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 1.2932674998714193e-05,
+      "loss": 0.5011,
+      "step": 16575
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 1.2930103379108162e-05,
+      "loss": 0.3666,
+      "step": 16576
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 1.2927531759502134e-05,
+      "loss": 0.626,
+      "step": 16577
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 1.2924960139896109e-05,
+      "loss": 0.8441,
+      "step": 16578
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 1.292238852029008e-05,
+      "loss": 0.4404,
+      "step": 16579
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 1.2919816900684051e-05,
+      "loss": 0.4412,
+      "step": 16580
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 1.2917245281078024e-05,
+      "loss": 0.4765,
+      "step": 16581
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 1.2914673661471996e-05,
+      "loss": 0.5804,
+      "step": 16582
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 1.2912102041865967e-05,
+      "loss": 0.5236,
+      "step": 16583
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 1.2909530422259942e-05,
+      "loss": 0.8865,
+      "step": 16584
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 1.2906958802653912e-05,
+      "loss": 0.8729,
+      "step": 16585
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 1.2904387183047883e-05,
+      "loss": 0.4139,
+      "step": 16586
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 1.2901815563441858e-05,
+      "loss": 0.4977,
+      "step": 16587
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 1.289924394383583e-05,
+      "loss": 0.6053,
+      "step": 16588
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 1.28966723242298e-05,
+      "loss": 1.0336,
+      "step": 16589
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 1.2894100704623774e-05,
+      "loss": 0.6394,
+      "step": 16590
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 1.2891529085017745e-05,
+      "loss": 0.563,
+      "step": 16591
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 1.2888957465411716e-05,
+      "loss": 0.4074,
+      "step": 16592
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 1.288638584580569e-05,
+      "loss": 0.3957,
+      "step": 16593
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 1.2883814226199661e-05,
+      "loss": 0.421,
+      "step": 16594
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 1.2881242606593632e-05,
+      "loss": 0.644,
+      "step": 16595
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 1.2878670986987607e-05,
+      "loss": 0.6694,
+      "step": 16596
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 1.2876099367381578e-05,
+      "loss": 0.5286,
+      "step": 16597
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 1.287352774777555e-05,
+      "loss": 0.5565,
+      "step": 16598
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 1.2870956128169523e-05,
+      "loss": 0.558,
+      "step": 16599
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 1.2868384508563494e-05,
+      "loss": 0.5375,
+      "step": 16600
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 1.2865812888957466e-05,
+      "loss": 0.3004,
+      "step": 16601
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 1.2863241269351439e-05,
+      "loss": 0.3612,
+      "step": 16602
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 1.286066964974541e-05,
+      "loss": 0.5449,
+      "step": 16603
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 1.2858098030139381e-05,
+      "loss": 0.6791,
+      "step": 16604
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 1.2855526410533356e-05,
+      "loss": 0.7833,
+      "step": 16605
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 1.2852954790927328e-05,
+      "loss": 0.7013,
+      "step": 16606
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 1.2850383171321297e-05,
+      "loss": 0.4802,
+      "step": 16607
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 1.2847811551715272e-05,
+      "loss": 0.5044,
+      "step": 16608
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 1.2845239932109243e-05,
+      "loss": 0.6957,
+      "step": 16609
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 1.2842668312503215e-05,
+      "loss": 1.4766,
+      "step": 16610
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 1.2840096692897188e-05,
+      "loss": 0.8325,
+      "step": 16611
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 1.283752507329116e-05,
+      "loss": 0.6243,
+      "step": 16612
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 1.283495345368513e-05,
+      "loss": 0.5364,
+      "step": 16613
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 1.2832381834079105e-05,
+      "loss": 0.7091,
+      "step": 16614
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 1.2829810214473075e-05,
+      "loss": 0.4031,
+      "step": 16615
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 1.2827238594867046e-05,
+      "loss": 0.4481,
+      "step": 16616
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 1.2824666975261021e-05,
+      "loss": 0.5139,
+      "step": 16617
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 1.2822095355654993e-05,
+      "loss": 0.629,
+      "step": 16618
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 1.2819523736048964e-05,
+      "loss": 0.7351,
+      "step": 16619
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 1.2816952116442937e-05,
+      "loss": 0.4835,
+      "step": 16620
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 1.2814380496836908e-05,
+      "loss": 0.671,
+      "step": 16621
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 1.281180887723088e-05,
+      "loss": 0.6689,
+      "step": 16622
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 1.2809237257624853e-05,
+      "loss": 0.8081,
+      "step": 16623
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 1.2806665638018824e-05,
+      "loss": 0.6986,
+      "step": 16624
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 1.2804094018412796e-05,
+      "loss": 0.5498,
+      "step": 16625
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 1.280152239880677e-05,
+      "loss": 0.6369,
+      "step": 16626
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 1.2798950779200742e-05,
+      "loss": 0.3823,
+      "step": 16627
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 1.2796379159594713e-05,
+      "loss": 0.8273,
+      "step": 16628
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 1.2793807539988686e-05,
+      "loss": 0.5053,
+      "step": 16629
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 1.2791235920382658e-05,
+      "loss": 0.5658,
+      "step": 16630
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 1.2788664300776629e-05,
+      "loss": 0.6192,
+      "step": 16631
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 1.2786092681170602e-05,
+      "loss": 0.5305,
+      "step": 16632
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 1.2783521061564573e-05,
+      "loss": 0.6062,
+      "step": 16633
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 1.2780949441958545e-05,
+      "loss": 0.7714,
+      "step": 16634
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 1.277837782235252e-05,
+      "loss": 0.4589,
+      "step": 16635
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 1.2775806202746491e-05,
+      "loss": 0.9039,
+      "step": 16636
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 1.2773234583140462e-05,
+      "loss": 0.7678,
+      "step": 16637
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 1.2770662963534435e-05,
+      "loss": 0.6417,
+      "step": 16638
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 1.2768091343928407e-05,
+      "loss": 0.4726,
+      "step": 16639
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 1.2765519724322378e-05,
+      "loss": 0.8083,
+      "step": 16640
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 1.2762948104716351e-05,
+      "loss": 0.6007,
+      "step": 16641
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 1.2760376485110323e-05,
+      "loss": 0.3644,
+      "step": 16642
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 1.2757804865504294e-05,
+      "loss": 0.5645,
+      "step": 16643
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 1.2755233245898269e-05,
+      "loss": 0.4623,
+      "step": 16644
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 1.275266162629224e-05,
+      "loss": 0.4214,
+      "step": 16645
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 1.275009000668621e-05,
+      "loss": 0.6312,
+      "step": 16646
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 1.2747518387080185e-05,
+      "loss": 0.5187,
+      "step": 16647
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 1.2744946767474156e-05,
+      "loss": 0.6539,
+      "step": 16648
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 1.2742375147868127e-05,
+      "loss": 1.012,
+      "step": 16649
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 1.27398035282621e-05,
+      "loss": 0.7183,
+      "step": 16650
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 1.2737231908656072e-05,
+      "loss": 0.5536,
+      "step": 16651
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 1.2734660289050043e-05,
+      "loss": 0.7732,
+      "step": 16652
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 1.2732088669444018e-05,
+      "loss": 0.4062,
+      "step": 16653
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 1.2729517049837988e-05,
+      "loss": 0.407,
+      "step": 16654
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 1.2726945430231959e-05,
+      "loss": 0.4254,
+      "step": 16655
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 1.2724373810625934e-05,
+      "loss": 0.6524,
+      "step": 16656
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 1.2721802191019905e-05,
+      "loss": 0.6234,
+      "step": 16657
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 1.2719230571413877e-05,
+      "loss": 0.4879,
+      "step": 16658
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 1.271665895180785e-05,
+      "loss": 0.4598,
+      "step": 16659
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 1.2714087332201821e-05,
+      "loss": 0.5417,
+      "step": 16660
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 1.2711515712595792e-05,
+      "loss": 0.3428,
+      "step": 16661
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 1.2708944092989765e-05,
+      "loss": 0.7057,
+      "step": 16662
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 1.2706372473383737e-05,
+      "loss": 0.5543,
+      "step": 16663
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 1.2703800853777708e-05,
+      "loss": 0.4506,
+      "step": 16664
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 1.2701229234171683e-05,
+      "loss": 0.7076,
+      "step": 16665
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 1.2698657614565654e-05,
+      "loss": 0.7501,
+      "step": 16666
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 1.2696085994959626e-05,
+      "loss": 0.2701,
+      "step": 16667
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 1.2693514375353599e-05,
+      "loss": 0.6231,
+      "step": 16668
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 1.269094275574757e-05,
+      "loss": 0.4083,
+      "step": 16669
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 1.2688371136141542e-05,
+      "loss": 0.458,
+      "step": 16670
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 1.2685799516535515e-05,
+      "loss": 0.6899,
+      "step": 16671
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 1.2683227896929486e-05,
+      "loss": 0.3277,
+      "step": 16672
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 1.2680656277323457e-05,
+      "loss": 0.6106,
+      "step": 16673
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 1.2678084657717432e-05,
+      "loss": 0.589,
+      "step": 16674
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 1.2675513038111404e-05,
+      "loss": 0.533,
+      "step": 16675
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 1.2672941418505373e-05,
+      "loss": 0.8506,
+      "step": 16676
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 1.2670369798899348e-05,
+      "loss": 0.4306,
+      "step": 16677
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 1.266779817929332e-05,
+      "loss": 0.2507,
+      "step": 16678
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 1.266522655968729e-05,
+      "loss": 0.5046,
+      "step": 16679
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 1.2662654940081264e-05,
+      "loss": 0.4587,
+      "step": 16680
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 1.2660083320475235e-05,
+      "loss": 0.6557,
+      "step": 16681
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 1.2657511700869207e-05,
+      "loss": 0.6434,
+      "step": 16682
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 1.2654940081263181e-05,
+      "loss": 0.4691,
+      "step": 16683
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 1.2652368461657153e-05,
+      "loss": 0.6558,
+      "step": 16684
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 1.2649796842051122e-05,
+      "loss": 1.0618,
+      "step": 16685
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 1.2647225222445097e-05,
+      "loss": 0.4105,
+      "step": 16686
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 1.2644653602839069e-05,
+      "loss": 0.4066,
+      "step": 16687
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 1.264208198323304e-05,
+      "loss": 0.9087,
+      "step": 16688
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 1.2639510363627013e-05,
+      "loss": 0.7937,
+      "step": 16689
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 1.2636938744020984e-05,
+      "loss": 0.7838,
+      "step": 16690
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 1.2634367124414956e-05,
+      "loss": 0.6548,
+      "step": 16691
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 1.263179550480893e-05,
+      "loss": 0.616,
+      "step": 16692
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 1.26292238852029e-05,
+      "loss": 0.5898,
+      "step": 16693
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 1.2626652265596872e-05,
+      "loss": 0.8256,
+      "step": 16694
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 1.2624080645990846e-05,
+      "loss": 0.546,
+      "step": 16695
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 1.2621509026384818e-05,
+      "loss": 0.3062,
+      "step": 16696
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 1.2618937406778789e-05,
+      "loss": 0.6275,
+      "step": 16697
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 1.2616365787172762e-05,
+      "loss": 0.6884,
+      "step": 16698
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 1.2613794167566734e-05,
+      "loss": 0.348,
+      "step": 16699
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 1.2611222547960705e-05,
+      "loss": 0.823,
+      "step": 16700
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 1.2608650928354678e-05,
+      "loss": 0.6654,
+      "step": 16701
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 1.260607930874865e-05,
+      "loss": 0.5926,
+      "step": 16702
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 1.260350768914262e-05,
+      "loss": 0.6507,
+      "step": 16703
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 1.2600936069536596e-05,
+      "loss": 0.6036,
+      "step": 16704
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 1.2598364449930567e-05,
+      "loss": 0.4743,
+      "step": 16705
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 1.259579283032454e-05,
+      "loss": 0.718,
+      "step": 16706
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 1.2593221210718511e-05,
+      "loss": 0.5946,
+      "step": 16707
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 1.2590649591112483e-05,
+      "loss": 0.3169,
+      "step": 16708
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 1.2588077971506458e-05,
+      "loss": 0.7599,
+      "step": 16709
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 1.2585506351900427e-05,
+      "loss": 0.5359,
+      "step": 16710
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 1.2582934732294399e-05,
+      "loss": 0.7755,
+      "step": 16711
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 1.2580363112688373e-05,
+      "loss": 0.6435,
+      "step": 16712
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 1.2577791493082345e-05,
+      "loss": 0.8226,
+      "step": 16713
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 1.2575219873476316e-05,
+      "loss": 0.6316,
+      "step": 16714
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 1.257264825387029e-05,
+      "loss": 0.7726,
+      "step": 16715
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 1.257007663426426e-05,
+      "loss": 0.3766,
+      "step": 16716
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 1.2567505014658232e-05,
+      "loss": 0.5601,
+      "step": 16717
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 1.2564933395052205e-05,
+      "loss": 0.5783,
+      "step": 16718
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 1.2562361775446176e-05,
+      "loss": 0.5256,
+      "step": 16719
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 1.2559790155840148e-05,
+      "loss": 0.6142,
+      "step": 16720
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 1.2557218536234123e-05,
+      "loss": 0.4238,
+      "step": 16721
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 1.2554646916628094e-05,
+      "loss": 0.4275,
+      "step": 16722
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 1.2552075297022065e-05,
+      "loss": 0.638,
+      "step": 16723
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 1.2549503677416038e-05,
+      "loss": 0.7065,
+      "step": 16724
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 1.254693205781001e-05,
+      "loss": 0.6664,
+      "step": 16725
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 1.2544360438203981e-05,
+      "loss": 0.3895,
+      "step": 16726
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 1.2541788818597954e-05,
+      "loss": 0.512,
+      "step": 16727
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 1.2539217198991926e-05,
+      "loss": 0.5066,
+      "step": 16728
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 1.2536645579385897e-05,
+      "loss": 0.455,
+      "step": 16729
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 1.2534073959779872e-05,
+      "loss": 0.5555,
+      "step": 16730
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 1.2531502340173843e-05,
+      "loss": 0.3129,
+      "step": 16731
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 1.2528930720567813e-05,
+      "loss": 0.5947,
+      "step": 16732
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 1.2526359100961788e-05,
+      "loss": 0.6924,
+      "step": 16733
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 1.2523787481355759e-05,
+      "loss": 0.3695,
+      "step": 16734
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 1.252121586174973e-05,
+      "loss": 0.3949,
+      "step": 16735
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 1.2518644242143703e-05,
+      "loss": 0.3805,
+      "step": 16736
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 1.2516072622537675e-05,
+      "loss": 0.6515,
+      "step": 16737
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 1.2513501002931646e-05,
+      "loss": 0.5643,
+      "step": 16738
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 1.2510929383325621e-05,
+      "loss": 0.504,
+      "step": 16739
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 1.250835776371959e-05,
+      "loss": 0.7258,
+      "step": 16740
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 1.2505786144113562e-05,
+      "loss": 0.7386,
+      "step": 16741
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 1.2503214524507537e-05,
+      "loss": 0.4643,
+      "step": 16742
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 1.2500642904901508e-05,
+      "loss": 0.6856,
+      "step": 16743
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 1.249807128529548e-05,
+      "loss": 0.5399,
+      "step": 16744
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 1.2495499665689451e-05,
+      "loss": 0.6784,
+      "step": 16745
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 1.2492928046083424e-05,
+      "loss": 0.5797,
+      "step": 16746
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 1.2490356426477397e-05,
+      "loss": 0.7325,
+      "step": 16747
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 1.2487784806871368e-05,
+      "loss": 0.7649,
+      "step": 16748
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 1.248521318726534e-05,
+      "loss": 0.5098,
+      "step": 16749
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 1.2482641567659313e-05,
+      "loss": 0.9299,
+      "step": 16750
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 1.2480069948053284e-05,
+      "loss": 0.4631,
+      "step": 16751
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 1.2477498328447257e-05,
+      "loss": 0.5372,
+      "step": 16752
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 1.2474926708841229e-05,
+      "loss": 0.4569,
+      "step": 16753
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 1.24723550892352e-05,
+      "loss": 0.7956,
+      "step": 16754
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 1.2469783469629173e-05,
+      "loss": 0.5762,
+      "step": 16755
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 1.2467211850023146e-05,
+      "loss": 0.5526,
+      "step": 16756
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 1.2464640230417118e-05,
+      "loss": 0.6512,
+      "step": 16757
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 1.2462068610811089e-05,
+      "loss": 0.4877,
+      "step": 16758
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 1.2459496991205062e-05,
+      "loss": 0.4647,
+      "step": 16759
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 1.2456925371599033e-05,
+      "loss": 0.4683,
+      "step": 16760
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 1.2454353751993007e-05,
+      "loss": 0.9631,
+      "step": 16761
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 1.2451782132386978e-05,
+      "loss": 0.6105,
+      "step": 16762
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 1.244921051278095e-05,
+      "loss": 0.5878,
+      "step": 16763
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 1.2446638893174922e-05,
+      "loss": 0.3957,
+      "step": 16764
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 1.2444067273568895e-05,
+      "loss": 0.5381,
+      "step": 16765
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 1.2441495653962867e-05,
+      "loss": 0.8019,
+      "step": 16766
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 1.2438924034356838e-05,
+      "loss": 0.6082,
+      "step": 16767
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 1.2436352414750811e-05,
+      "loss": 0.472,
+      "step": 16768
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 1.2433780795144783e-05,
+      "loss": 0.5433,
+      "step": 16769
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 1.2431209175538756e-05,
+      "loss": 0.4558,
+      "step": 16770
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 1.2428637555932727e-05,
+      "loss": 0.5675,
+      "step": 16771
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 1.2426065936326699e-05,
+      "loss": 0.5109,
+      "step": 16772
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 1.2423494316720672e-05,
+      "loss": 0.6674,
+      "step": 16773
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 1.2420922697114645e-05,
+      "loss": 0.5758,
+      "step": 16774
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 1.2418351077508614e-05,
+      "loss": 0.4912,
+      "step": 16775
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 1.2415779457902587e-05,
+      "loss": 0.4894,
+      "step": 16776
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 1.241320783829656e-05,
+      "loss": 0.5037,
+      "step": 16777
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 1.2410636218690532e-05,
+      "loss": 0.8568,
+      "step": 16778
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 1.2408064599084503e-05,
+      "loss": 0.6815,
+      "step": 16779
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 1.2405492979478476e-05,
+      "loss": 0.6472,
+      "step": 16780
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 1.2402921359872448e-05,
+      "loss": 0.5681,
+      "step": 16781
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 1.240034974026642e-05,
+      "loss": 0.6206,
+      "step": 16782
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 1.2397778120660392e-05,
+      "loss": 0.261,
+      "step": 16783
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 1.2395206501054364e-05,
+      "loss": 0.7841,
+      "step": 16784
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 1.2392634881448337e-05,
+      "loss": 0.3922,
+      "step": 16785
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 1.239006326184231e-05,
+      "loss": 0.5339,
+      "step": 16786
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 1.2387491642236281e-05,
+      "loss": 0.8543,
+      "step": 16787
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 1.2384920022630252e-05,
+      "loss": 0.6533,
+      "step": 16788
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 1.2382348403024226e-05,
+      "loss": 0.2966,
+      "step": 16789
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 1.2379776783418197e-05,
+      "loss": 0.7087,
+      "step": 16790
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 1.237720516381217e-05,
+      "loss": 0.2914,
+      "step": 16791
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 1.2374633544206141e-05,
+      "loss": 0.5665,
+      "step": 16792
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 1.2372061924600113e-05,
+      "loss": 0.7058,
+      "step": 16793
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 1.2369490304994086e-05,
+      "loss": 0.7922,
+      "step": 16794
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 1.2366918685388059e-05,
+      "loss": 0.9093,
+      "step": 16795
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 1.236434706578203e-05,
+      "loss": 0.5802,
+      "step": 16796
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 1.2361775446176002e-05,
+      "loss": 0.6883,
+      "step": 16797
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 1.2359203826569975e-05,
+      "loss": 0.3521,
+      "step": 16798
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 1.2356632206963946e-05,
+      "loss": 0.9301,
+      "step": 16799
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 1.2354060587357919e-05,
+      "loss": 0.2308,
+      "step": 16800
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 1.235148896775189e-05,
+      "loss": 0.8028,
+      "step": 16801
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 1.2348917348145862e-05,
+      "loss": 0.7518,
+      "step": 16802
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 1.2346345728539835e-05,
+      "loss": 0.7084,
+      "step": 16803
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 1.2343774108933808e-05,
+      "loss": 0.859,
+      "step": 16804
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 1.234120248932778e-05,
+      "loss": 0.8443,
+      "step": 16805
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 1.233863086972175e-05,
+      "loss": 0.4853,
+      "step": 16806
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 1.2336059250115724e-05,
+      "loss": 0.6749,
+      "step": 16807
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 1.2333487630509695e-05,
+      "loss": 0.3896,
+      "step": 16808
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 1.2330916010903668e-05,
+      "loss": 0.6093,
+      "step": 16809
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 1.232834439129764e-05,
+      "loss": 0.5624,
+      "step": 16810
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 1.2325772771691611e-05,
+      "loss": 0.547,
+      "step": 16811
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 1.2323201152085584e-05,
+      "loss": 0.4755,
+      "step": 16812
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 1.2320629532479557e-05,
+      "loss": 0.6553,
+      "step": 16813
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 1.2318057912873527e-05,
+      "loss": 0.6305,
+      "step": 16814
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 1.23154862932675e-05,
+      "loss": 0.2364,
+      "step": 16815
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 1.2312914673661473e-05,
+      "loss": 0.5485,
+      "step": 16816
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 1.2310343054055444e-05,
+      "loss": 0.2406,
+      "step": 16817
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 1.2307771434449416e-05,
+      "loss": 0.7171,
+      "step": 16818
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 1.2305199814843389e-05,
+      "loss": 0.354,
+      "step": 16819
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 1.2302628195237362e-05,
+      "loss": 0.5713,
+      "step": 16820
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 1.2300056575631333e-05,
+      "loss": 0.5106,
+      "step": 16821
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 1.2297484956025305e-05,
+      "loss": 0.74,
+      "step": 16822
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 1.2294913336419278e-05,
+      "loss": 0.6123,
+      "step": 16823
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 1.229234171681325e-05,
+      "loss": 0.3967,
+      "step": 16824
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 1.2289770097207222e-05,
+      "loss": 0.4942,
+      "step": 16825
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 1.2287198477601194e-05,
+      "loss": 0.6416,
+      "step": 16826
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 1.2284626857995165e-05,
+      "loss": 0.4204,
+      "step": 16827
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 1.2282055238389138e-05,
+      "loss": 0.547,
+      "step": 16828
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 1.2279483618783111e-05,
+      "loss": 0.4993,
+      "step": 16829
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 1.2276911999177083e-05,
+      "loss": 0.4861,
+      "step": 16830
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 1.2274340379571054e-05,
+      "loss": 0.4636,
+      "step": 16831
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 1.2271768759965027e-05,
+      "loss": 0.7419,
+      "step": 16832
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 1.2269197140358998e-05,
+      "loss": 0.4006,
+      "step": 16833
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 1.2266625520752971e-05,
+      "loss": 0.5036,
+      "step": 16834
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 1.2264053901146943e-05,
+      "loss": 0.6093,
+      "step": 16835
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 1.2261482281540914e-05,
+      "loss": 0.4372,
+      "step": 16836
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 1.2258910661934887e-05,
+      "loss": 0.6946,
+      "step": 16837
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 1.225633904232886e-05,
+      "loss": 0.6591,
+      "step": 16838
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 1.2253767422722832e-05,
+      "loss": 0.7176,
+      "step": 16839
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 1.2251195803116803e-05,
+      "loss": 0.8355,
+      "step": 16840
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 1.2248624183510776e-05,
+      "loss": 0.5617,
+      "step": 16841
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 1.2246052563904748e-05,
+      "loss": 0.4902,
+      "step": 16842
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 1.224348094429872e-05,
+      "loss": 0.705,
+      "step": 16843
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 1.2240909324692692e-05,
+      "loss": 0.5897,
+      "step": 16844
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 1.2238337705086663e-05,
+      "loss": 0.5662,
+      "step": 16845
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 1.2235766085480636e-05,
+      "loss": 0.506,
+      "step": 16846
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 1.223319446587461e-05,
+      "loss": 0.4554,
+      "step": 16847
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 1.2230622846268581e-05,
+      "loss": 0.5772,
+      "step": 16848
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 1.2228051226662552e-05,
+      "loss": 0.6723,
+      "step": 16849
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 1.2225479607056525e-05,
+      "loss": 0.6301,
+      "step": 16850
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 1.2222907987450497e-05,
+      "loss": 0.7057,
+      "step": 16851
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 1.222033636784447e-05,
+      "loss": 0.5213,
+      "step": 16852
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 1.2217764748238441e-05,
+      "loss": 0.7819,
+      "step": 16853
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 1.2215193128632413e-05,
+      "loss": 0.5545,
+      "step": 16854
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 1.2212621509026386e-05,
+      "loss": 0.7142,
+      "step": 16855
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 1.2210049889420359e-05,
+      "loss": 0.5973,
+      "step": 16856
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 1.2207478269814328e-05,
+      "loss": 0.4988,
+      "step": 16857
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 1.2204906650208302e-05,
+      "loss": 0.3948,
+      "step": 16858
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 1.2202335030602275e-05,
+      "loss": 0.8425,
+      "step": 16859
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 1.2199763410996246e-05,
+      "loss": 0.8987,
+      "step": 16860
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 1.2197191791390217e-05,
+      "loss": 0.5225,
+      "step": 16861
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 1.219462017178419e-05,
+      "loss": 0.5958,
+      "step": 16862
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 1.2192048552178162e-05,
+      "loss": 0.5399,
+      "step": 16863
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 1.2189476932572135e-05,
+      "loss": 0.6012,
+      "step": 16864
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 1.2186905312966106e-05,
+      "loss": 0.5678,
+      "step": 16865
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 1.2184333693360078e-05,
+      "loss": 0.2564,
+      "step": 16866
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 1.218176207375405e-05,
+      "loss": 0.4661,
+      "step": 16867
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 1.2179190454148024e-05,
+      "loss": 0.7109,
+      "step": 16868
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 1.2176618834541995e-05,
+      "loss": 0.6129,
+      "step": 16869
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 1.2174047214935967e-05,
+      "loss": 0.5713,
+      "step": 16870
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 1.217147559532994e-05,
+      "loss": 0.3991,
+      "step": 16871
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 1.2168903975723911e-05,
+      "loss": 0.5142,
+      "step": 16872
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 1.2166332356117884e-05,
+      "loss": 0.7886,
+      "step": 16873
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 1.2163760736511855e-05,
+      "loss": 0.7331,
+      "step": 16874
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 1.2161189116905827e-05,
+      "loss": 0.5816,
+      "step": 16875
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 1.21586174972998e-05,
+      "loss": 0.7366,
+      "step": 16876
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 1.2156045877693773e-05,
+      "loss": 0.716,
+      "step": 16877
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 1.2153474258087744e-05,
+      "loss": 0.3336,
+      "step": 16878
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 1.2150902638481716e-05,
+      "loss": 0.4925,
+      "step": 16879
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 1.2148331018875689e-05,
+      "loss": 0.7731,
+      "step": 16880
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 1.214575939926966e-05,
+      "loss": 0.4408,
+      "step": 16881
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 1.2143187779663633e-05,
+      "loss": 0.4349,
+      "step": 16882
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 1.2140616160057605e-05,
+      "loss": 0.7025,
+      "step": 16883
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 1.2138044540451576e-05,
+      "loss": 0.6117,
+      "step": 16884
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 1.2135472920845549e-05,
+      "loss": 0.4208,
+      "step": 16885
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 1.2132901301239522e-05,
+      "loss": 0.5548,
+      "step": 16886
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 1.2130329681633494e-05,
+      "loss": 0.7709,
+      "step": 16887
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 1.2127758062027465e-05,
+      "loss": 0.5472,
+      "step": 16888
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 1.2125186442421438e-05,
+      "loss": 0.3263,
+      "step": 16889
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 1.212261482281541e-05,
+      "loss": 0.7227,
+      "step": 16890
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 1.2120043203209382e-05,
+      "loss": 0.4628,
+      "step": 16891
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 1.2117471583603354e-05,
+      "loss": 0.6573,
+      "step": 16892
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 1.2114899963997325e-05,
+      "loss": 0.5803,
+      "step": 16893
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 1.2112328344391298e-05,
+      "loss": 0.7586,
+      "step": 16894
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 1.2109756724785271e-05,
+      "loss": 0.7156,
+      "step": 16895
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 1.2107185105179243e-05,
+      "loss": 0.4408,
+      "step": 16896
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 1.2104613485573214e-05,
+      "loss": 0.8194,
+      "step": 16897
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 1.2102041865967187e-05,
+      "loss": 0.4598,
+      "step": 16898
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 1.209947024636116e-05,
+      "loss": 0.6998,
+      "step": 16899
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 1.209689862675513e-05,
+      "loss": 0.6565,
+      "step": 16900
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 1.2094327007149103e-05,
+      "loss": 0.5381,
+      "step": 16901
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 1.2091755387543076e-05,
+      "loss": 0.4795,
+      "step": 16902
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 1.2089183767937047e-05,
+      "loss": 0.6697,
+      "step": 16903
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 1.2086612148331019e-05,
+      "loss": 0.4844,
+      "step": 16904
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 1.2084040528724992e-05,
+      "loss": 0.4785,
+      "step": 16905
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 1.2081468909118963e-05,
+      "loss": 0.4913,
+      "step": 16906
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 1.2078897289512936e-05,
+      "loss": 0.5317,
+      "step": 16907
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 1.2076325669906908e-05,
+      "loss": 0.7422,
+      "step": 16908
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 1.2073754050300879e-05,
+      "loss": 0.4731,
+      "step": 16909
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 1.2071182430694852e-05,
+      "loss": 0.6761,
+      "step": 16910
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 1.2068610811088825e-05,
+      "loss": 0.9468,
+      "step": 16911
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 1.2066039191482797e-05,
+      "loss": 0.7574,
+      "step": 16912
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 1.2063467571876768e-05,
+      "loss": 0.7833,
+      "step": 16913
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 1.2060895952270741e-05,
+      "loss": 0.6268,
+      "step": 16914
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 1.2058324332664712e-05,
+      "loss": 0.6702,
+      "step": 16915
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 1.2055752713058686e-05,
+      "loss": 0.5254,
+      "step": 16916
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 1.2053181093452657e-05,
+      "loss": 0.4514,
+      "step": 16917
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 1.2050609473846628e-05,
+      "loss": 0.5475,
+      "step": 16918
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 1.2048037854240601e-05,
+      "loss": 0.4556,
+      "step": 16919
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 1.2045466234634574e-05,
+      "loss": 0.4854,
+      "step": 16920
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 1.2042894615028546e-05,
+      "loss": 0.8089,
+      "step": 16921
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 1.2040322995422517e-05,
+      "loss": 0.6179,
+      "step": 16922
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 1.203775137581649e-05,
+      "loss": 0.5025,
+      "step": 16923
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 1.2035179756210462e-05,
+      "loss": 0.5702,
+      "step": 16924
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 1.2032608136604435e-05,
+      "loss": 0.7461,
+      "step": 16925
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 1.2030036516998406e-05,
+      "loss": 0.3978,
+      "step": 16926
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 1.2027464897392378e-05,
+      "loss": 0.6089,
+      "step": 16927
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 1.202489327778635e-05,
+      "loss": 0.49,
+      "step": 16928
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 1.2022321658180324e-05,
+      "loss": 0.6795,
+      "step": 16929
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 1.2019750038574295e-05,
+      "loss": 0.5021,
+      "step": 16930
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 1.2017178418968266e-05,
+      "loss": 0.6911,
+      "step": 16931
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 1.201460679936224e-05,
+      "loss": 0.4389,
+      "step": 16932
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 1.2012035179756211e-05,
+      "loss": 0.4459,
+      "step": 16933
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 1.2009463560150184e-05,
+      "loss": 0.5999,
+      "step": 16934
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 1.2006891940544155e-05,
+      "loss": 0.6027,
+      "step": 16935
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 1.2004320320938127e-05,
+      "loss": 0.5585,
+      "step": 16936
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 1.20017487013321e-05,
+      "loss": 0.7009,
+      "step": 16937
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 1.1999177081726073e-05,
+      "loss": 0.5084,
+      "step": 16938
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 1.1996605462120043e-05,
+      "loss": 0.6953,
+      "step": 16939
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 1.1994033842514016e-05,
+      "loss": 0.5832,
+      "step": 16940
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 1.1991462222907989e-05,
+      "loss": 0.8866,
+      "step": 16941
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 1.198889060330196e-05,
+      "loss": 0.6887,
+      "step": 16942
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 1.1986318983695931e-05,
+      "loss": 0.7094,
+      "step": 16943
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 1.1983747364089905e-05,
+      "loss": 0.6561,
+      "step": 16944
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 1.1981175744483876e-05,
+      "loss": 0.4234,
+      "step": 16945
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 1.1978604124877849e-05,
+      "loss": 0.7299,
+      "step": 16946
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 1.197603250527182e-05,
+      "loss": 0.9321,
+      "step": 16947
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 1.1973460885665792e-05,
+      "loss": 0.598,
+      "step": 16948
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 1.1970889266059765e-05,
+      "loss": 0.5061,
+      "step": 16949
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 1.1968317646453738e-05,
+      "loss": 0.6773,
+      "step": 16950
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 1.196574602684771e-05,
+      "loss": 0.5815,
+      "step": 16951
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 1.196317440724168e-05,
+      "loss": 0.5356,
+      "step": 16952
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 1.1960602787635654e-05,
+      "loss": 0.5913,
+      "step": 16953
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 1.1958031168029625e-05,
+      "loss": 0.5905,
+      "step": 16954
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 1.1955459548423598e-05,
+      "loss": 0.3934,
+      "step": 16955
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 1.195288792881757e-05,
+      "loss": 0.4854,
+      "step": 16956
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 1.1950316309211541e-05,
+      "loss": 0.6971,
+      "step": 16957
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 1.1947744689605514e-05,
+      "loss": 0.5059,
+      "step": 16958
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 1.1945173069999487e-05,
+      "loss": 0.5514,
+      "step": 16959
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 1.1942601450393458e-05,
+      "loss": 0.6407,
+      "step": 16960
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 1.194002983078743e-05,
+      "loss": 0.4046,
+      "step": 16961
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 1.1937458211181403e-05,
+      "loss": 0.644,
+      "step": 16962
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 1.1934886591575374e-05,
+      "loss": 0.3619,
+      "step": 16963
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 1.1932314971969347e-05,
+      "loss": 0.8143,
+      "step": 16964
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 1.1929743352363319e-05,
+      "loss": 0.7085,
+      "step": 16965
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 1.192717173275729e-05,
+      "loss": 0.9056,
+      "step": 16966
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 1.1924600113151263e-05,
+      "loss": 0.5546,
+      "step": 16967
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 1.1922028493545236e-05,
+      "loss": 0.734,
+      "step": 16968
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 1.1919456873939208e-05,
+      "loss": 0.4764,
+      "step": 16969
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 1.1916885254333179e-05,
+      "loss": 0.9864,
+      "step": 16970
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 1.1914313634727152e-05,
+      "loss": 0.5496,
+      "step": 16971
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 1.1911742015121125e-05,
+      "loss": 0.7669,
+      "step": 16972
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 1.1909170395515097e-05,
+      "loss": 0.553,
+      "step": 16973
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 1.1906598775909068e-05,
+      "loss": 0.3578,
+      "step": 16974
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 1.1904027156303041e-05,
+      "loss": 0.6141,
+      "step": 16975
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 1.1901455536697012e-05,
+      "loss": 0.5012,
+      "step": 16976
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 1.1898883917090985e-05,
+      "loss": 0.3578,
+      "step": 16977
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 1.1896312297484957e-05,
+      "loss": 0.6386,
+      "step": 16978
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 1.1893740677878928e-05,
+      "loss": 0.558,
+      "step": 16979
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 1.1891169058272901e-05,
+      "loss": 0.5712,
+      "step": 16980
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 1.1888597438666874e-05,
+      "loss": 0.297,
+      "step": 16981
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 1.1886025819060844e-05,
+      "loss": 0.5854,
+      "step": 16982
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 1.1883454199454817e-05,
+      "loss": 0.5673,
+      "step": 16983
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 1.188088257984879e-05,
+      "loss": 0.6065,
+      "step": 16984
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 1.1878310960242762e-05,
+      "loss": 0.4518,
+      "step": 16985
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 1.1875739340636733e-05,
+      "loss": 0.4427,
+      "step": 16986
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 1.1873167721030706e-05,
+      "loss": 0.4296,
+      "step": 16987
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 1.1870596101424677e-05,
+      "loss": 0.7346,
+      "step": 16988
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 1.186802448181865e-05,
+      "loss": 0.6638,
+      "step": 16989
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 1.1865452862212622e-05,
+      "loss": 0.6356,
+      "step": 16990
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 1.1862881242606593e-05,
+      "loss": 0.7709,
+      "step": 16991
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 1.1860309623000566e-05,
+      "loss": 0.7324,
+      "step": 16992
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 1.185773800339454e-05,
+      "loss": 0.3253,
+      "step": 16993
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 1.185516638378851e-05,
+      "loss": 0.4894,
+      "step": 16994
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 1.1852594764182482e-05,
+      "loss": 0.5941,
+      "step": 16995
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 1.1850023144576455e-05,
+      "loss": 0.524,
+      "step": 16996
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 1.1847451524970427e-05,
+      "loss": 0.634,
+      "step": 16997
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 1.18448799053644e-05,
+      "loss": 0.5689,
+      "step": 16998
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 1.1842308285758371e-05,
+      "loss": 0.6528,
+      "step": 16999
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 1.1839736666152342e-05,
+      "loss": 0.3835,
+      "step": 17000
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 1.1837165046546315e-05,
+      "loss": 0.5877,
+      "step": 17001
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 1.1834593426940289e-05,
+      "loss": 0.3578,
+      "step": 17002
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 1.183202180733426e-05,
+      "loss": 0.7195,
+      "step": 17003
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 1.1829450187728231e-05,
+      "loss": 0.4766,
+      "step": 17004
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 1.1826878568122204e-05,
+      "loss": 0.6602,
+      "step": 17005
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 1.1824306948516176e-05,
+      "loss": 0.3003,
+      "step": 17006
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 1.1821735328910149e-05,
+      "loss": 0.603,
+      "step": 17007
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 1.181916370930412e-05,
+      "loss": 0.417,
+      "step": 17008
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 1.1816592089698092e-05,
+      "loss": 0.5607,
+      "step": 17009
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 1.1814020470092065e-05,
+      "loss": 0.6143,
+      "step": 17010
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 1.1811448850486038e-05,
+      "loss": 0.8861,
+      "step": 17011
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 1.1808877230880009e-05,
+      "loss": 0.7693,
+      "step": 17012
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 1.180630561127398e-05,
+      "loss": 0.7319,
+      "step": 17013
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 1.1803733991667954e-05,
+      "loss": 0.6695,
+      "step": 17014
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 1.1801162372061925e-05,
+      "loss": 0.4305,
+      "step": 17015
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 1.1798590752455898e-05,
+      "loss": 0.4625,
+      "step": 17016
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 1.179601913284987e-05,
+      "loss": 0.6958,
+      "step": 17017
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 1.179344751324384e-05,
+      "loss": 0.6615,
+      "step": 17018
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 1.1790875893637814e-05,
+      "loss": 0.7347,
+      "step": 17019
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 1.1788304274031787e-05,
+      "loss": 0.6355,
+      "step": 17020
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 1.1785732654425757e-05,
+      "loss": 0.5266,
+      "step": 17021
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 1.178316103481973e-05,
+      "loss": 0.6488,
+      "step": 17022
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 1.1780589415213703e-05,
+      "loss": 0.5817,
+      "step": 17023
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 1.1778017795607674e-05,
+      "loss": 0.3001,
+      "step": 17024
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 1.1775446176001646e-05,
+      "loss": 0.3532,
+      "step": 17025
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 1.1772874556395619e-05,
+      "loss": 0.5702,
+      "step": 17026
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 1.177030293678959e-05,
+      "loss": 0.4775,
+      "step": 17027
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 1.1767731317183563e-05,
+      "loss": 0.6695,
+      "step": 17028
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 1.1765159697577534e-05,
+      "loss": 0.7109,
+      "step": 17029
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 1.1762588077971506e-05,
+      "loss": 0.5816,
+      "step": 17030
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 1.1760016458365479e-05,
+      "loss": 0.6032,
+      "step": 17031
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 1.1757444838759452e-05,
+      "loss": 0.4594,
+      "step": 17032
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 1.1754873219153423e-05,
+      "loss": 0.6773,
+      "step": 17033
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 1.1752301599547395e-05,
+      "loss": 0.5588,
+      "step": 17034
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 1.1749729979941368e-05,
+      "loss": 0.4203,
+      "step": 17035
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 1.174715836033534e-05,
+      "loss": 0.3519,
+      "step": 17036
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 1.1744586740729312e-05,
+      "loss": 0.2943,
+      "step": 17037
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 1.1742015121123284e-05,
+      "loss": 0.3568,
+      "step": 17038
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 1.1739443501517255e-05,
+      "loss": 0.5384,
+      "step": 17039
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 1.1736871881911228e-05,
+      "loss": 0.7351,
+      "step": 17040
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 1.1734300262305201e-05,
+      "loss": 0.5227,
+      "step": 17041
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 1.1731728642699173e-05,
+      "loss": 0.5646,
+      "step": 17042
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 1.1729157023093144e-05,
+      "loss": 0.5347,
+      "step": 17043
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 1.1726585403487117e-05,
+      "loss": 0.3318,
+      "step": 17044
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 1.172401378388109e-05,
+      "loss": 0.4608,
+      "step": 17045
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 1.1721442164275061e-05,
+      "loss": 0.5724,
+      "step": 17046
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 1.1718870544669033e-05,
+      "loss": 0.6029,
+      "step": 17047
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 1.1716298925063006e-05,
+      "loss": 0.7774,
+      "step": 17048
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 1.1713727305456977e-05,
+      "loss": 0.5583,
+      "step": 17049
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 1.171115568585095e-05,
+      "loss": 0.7116,
+      "step": 17050
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 1.1708584066244922e-05,
+      "loss": 0.4858,
+      "step": 17051
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 1.1706012446638893e-05,
+      "loss": 0.7361,
+      "step": 17052
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 1.1703440827032866e-05,
+      "loss": 0.481,
+      "step": 17053
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 1.170086920742684e-05,
+      "loss": 0.6313,
+      "step": 17054
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 1.169829758782081e-05,
+      "loss": 0.5904,
+      "step": 17055
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 1.1695725968214782e-05,
+      "loss": 0.837,
+      "step": 17056
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 1.1693154348608755e-05,
+      "loss": 0.6781,
+      "step": 17057
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 1.1690582729002726e-05,
+      "loss": 0.4733,
+      "step": 17058
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 1.16880111093967e-05,
+      "loss": 0.6586,
+      "step": 17059
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 1.1685439489790671e-05,
+      "loss": 0.3698,
+      "step": 17060
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 1.1682867870184642e-05,
+      "loss": 0.7335,
+      "step": 17061
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 1.1680296250578615e-05,
+      "loss": 0.6498,
+      "step": 17062
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 1.1677724630972588e-05,
+      "loss": 1.0305,
+      "step": 17063
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 1.1675153011366558e-05,
+      "loss": 0.5342,
+      "step": 17064
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 1.1672581391760531e-05,
+      "loss": 0.3923,
+      "step": 17065
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 1.1670009772154504e-05,
+      "loss": 0.367,
+      "step": 17066
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 1.1667438152548476e-05,
+      "loss": 0.7486,
+      "step": 17067
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 1.1664866532942447e-05,
+      "loss": 0.5752,
+      "step": 17068
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 1.166229491333642e-05,
+      "loss": 0.6067,
+      "step": 17069
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 1.1659723293730391e-05,
+      "loss": 0.3518,
+      "step": 17070
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 1.1657151674124365e-05,
+      "loss": 0.4584,
+      "step": 17071
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 1.1654580054518336e-05,
+      "loss": 0.6388,
+      "step": 17072
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 1.1652008434912307e-05,
+      "loss": 0.5809,
+      "step": 17073
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 1.164943681530628e-05,
+      "loss": 0.7137,
+      "step": 17074
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 1.1646865195700253e-05,
+      "loss": 0.4547,
+      "step": 17075
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 1.1644293576094225e-05,
+      "loss": 0.3519,
+      "step": 17076
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 1.1641721956488196e-05,
+      "loss": 0.4835,
+      "step": 17077
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 1.163915033688217e-05,
+      "loss": 0.7143,
+      "step": 17078
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 1.163657871727614e-05,
+      "loss": 0.6738,
+      "step": 17079
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 1.1634007097670114e-05,
+      "loss": 0.4356,
+      "step": 17080
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 1.1631435478064085e-05,
+      "loss": 0.6839,
+      "step": 17081
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 1.1628863858458056e-05,
+      "loss": 0.6225,
+      "step": 17082
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 1.162629223885203e-05,
+      "loss": 0.6046,
+      "step": 17083
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 1.1623720619246003e-05,
+      "loss": 0.599,
+      "step": 17084
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 1.1621148999639974e-05,
+      "loss": 0.8744,
+      "step": 17085
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 1.1618577380033945e-05,
+      "loss": 0.603,
+      "step": 17086
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 1.1616005760427918e-05,
+      "loss": 0.7209,
+      "step": 17087
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 1.161343414082189e-05,
+      "loss": 0.9083,
+      "step": 17088
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 1.1610862521215863e-05,
+      "loss": 0.5638,
+      "step": 17089
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 1.1608290901609834e-05,
+      "loss": 0.5523,
+      "step": 17090
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 1.1605719282003806e-05,
+      "loss": 0.392,
+      "step": 17091
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 1.1603147662397779e-05,
+      "loss": 0.7311,
+      "step": 17092
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 1.1600576042791752e-05,
+      "loss": 0.5158,
+      "step": 17093
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 1.1598004423185723e-05,
+      "loss": 0.4057,
+      "step": 17094
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 1.1595432803579695e-05,
+      "loss": 0.4444,
+      "step": 17095
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 1.1592861183973668e-05,
+      "loss": 0.6018,
+      "step": 17096
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 1.1590289564367639e-05,
+      "loss": 0.3895,
+      "step": 17097
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 1.1587717944761612e-05,
+      "loss": 0.7306,
+      "step": 17098
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 1.1585146325155583e-05,
+      "loss": 0.6478,
+      "step": 17099
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 1.1582574705549555e-05,
+      "loss": 0.4622,
+      "step": 17100
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 1.1580003085943528e-05,
+      "loss": 0.5828,
+      "step": 17101
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 1.1577431466337501e-05,
+      "loss": 0.7379,
+      "step": 17102
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 1.157485984673147e-05,
+      "loss": 0.3798,
+      "step": 17103
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 1.1572288227125444e-05,
+      "loss": 0.4587,
+      "step": 17104
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 1.1569716607519417e-05,
+      "loss": 1.1492,
+      "step": 17105
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 1.1567144987913388e-05,
+      "loss": 0.7657,
+      "step": 17106
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 1.156457336830736e-05,
+      "loss": 0.6581,
+      "step": 17107
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 1.1562001748701333e-05,
+      "loss": 0.7236,
+      "step": 17108
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 1.1559430129095304e-05,
+      "loss": 0.4828,
+      "step": 17109
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 1.1556858509489277e-05,
+      "loss": 0.347,
+      "step": 17110
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 1.1554286889883249e-05,
+      "loss": 0.3771,
+      "step": 17111
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 1.155171527027722e-05,
+      "loss": 0.6933,
+      "step": 17112
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 1.1549143650671193e-05,
+      "loss": 0.8423,
+      "step": 17113
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 1.1546572031065166e-05,
+      "loss": 0.6066,
+      "step": 17114
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 1.1544000411459137e-05,
+      "loss": 0.6928,
+      "step": 17115
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 1.1541428791853109e-05,
+      "loss": 0.8451,
+      "step": 17116
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 1.1538857172247082e-05,
+      "loss": 0.5642,
+      "step": 17117
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 1.1536285552641053e-05,
+      "loss": 0.5183,
+      "step": 17118
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 1.1533713933035026e-05,
+      "loss": 0.7084,
+      "step": 17119
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 1.1531142313428998e-05,
+      "loss": 0.8544,
+      "step": 17120
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 1.152857069382297e-05,
+      "loss": 0.6072,
+      "step": 17121
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 1.1525999074216942e-05,
+      "loss": 0.4184,
+      "step": 17122
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 1.1523427454610915e-05,
+      "loss": 0.5808,
+      "step": 17123
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 1.1520855835004887e-05,
+      "loss": 0.6127,
+      "step": 17124
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 1.1518284215398858e-05,
+      "loss": 0.5362,
+      "step": 17125
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 1.1515712595792831e-05,
+      "loss": 0.3473,
+      "step": 17126
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 1.1513140976186804e-05,
+      "loss": 0.6205,
+      "step": 17127
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 1.1510569356580776e-05,
+      "loss": 0.6069,
+      "step": 17128
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 1.1507997736974747e-05,
+      "loss": 0.5593,
+      "step": 17129
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 1.150542611736872e-05,
+      "loss": 0.5202,
+      "step": 17130
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 1.1502854497762691e-05,
+      "loss": 0.7968,
+      "step": 17131
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 1.1500282878156664e-05,
+      "loss": 0.5,
+      "step": 17132
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 1.1497711258550636e-05,
+      "loss": 0.7615,
+      "step": 17133
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 1.1495139638944607e-05,
+      "loss": 0.3925,
+      "step": 17134
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 1.149256801933858e-05,
+      "loss": 0.6003,
+      "step": 17135
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 1.1489996399732553e-05,
+      "loss": 0.5193,
+      "step": 17136
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 1.1487424780126525e-05,
+      "loss": 0.6985,
+      "step": 17137
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 1.1484853160520496e-05,
+      "loss": 0.2514,
+      "step": 17138
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 1.148228154091447e-05,
+      "loss": 0.4014,
+      "step": 17139
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 1.147970992130844e-05,
+      "loss": 0.5487,
+      "step": 17140
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 1.1477138301702414e-05,
+      "loss": 0.6569,
+      "step": 17141
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 1.1474566682096385e-05,
+      "loss": 0.7692,
+      "step": 17142
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 1.1471995062490356e-05,
+      "loss": 0.8929,
+      "step": 17143
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 1.146942344288433e-05,
+      "loss": 0.6246,
+      "step": 17144
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 1.1466851823278303e-05,
+      "loss": 0.3413,
+      "step": 17145
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 1.1464280203672272e-05,
+      "loss": 0.7461,
+      "step": 17146
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 1.1461708584066245e-05,
+      "loss": 0.4413,
+      "step": 17147
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 1.1459136964460218e-05,
+      "loss": 0.7743,
+      "step": 17148
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 1.145656534485419e-05,
+      "loss": 0.8306,
+      "step": 17149
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 1.1453993725248161e-05,
+      "loss": 0.5419,
+      "step": 17150
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 1.1451422105642134e-05,
+      "loss": 1.039,
+      "step": 17151
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 1.1448850486036106e-05,
+      "loss": 0.617,
+      "step": 17152
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 1.1446278866430079e-05,
+      "loss": 0.3435,
+      "step": 17153
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 1.144370724682405e-05,
+      "loss": 0.7596,
+      "step": 17154
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 1.1441135627218021e-05,
+      "loss": 0.4768,
+      "step": 17155
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 1.1438564007611994e-05,
+      "loss": 0.5735,
+      "step": 17156
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 1.1435992388005968e-05,
+      "loss": 0.8745,
+      "step": 17157
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 1.1433420768399939e-05,
+      "loss": 0.4289,
+      "step": 17158
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 1.143084914879391e-05,
+      "loss": 0.7639,
+      "step": 17159
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 1.1428277529187883e-05,
+      "loss": 0.3334,
+      "step": 17160
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 1.1425705909581855e-05,
+      "loss": 0.7578,
+      "step": 17161
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 1.1423134289975828e-05,
+      "loss": 0.6494,
+      "step": 17162
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 1.14205626703698e-05,
+      "loss": 0.6532,
+      "step": 17163
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 1.141799105076377e-05,
+      "loss": 0.7548,
+      "step": 17164
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 1.1415419431157744e-05,
+      "loss": 0.6904,
+      "step": 17165
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 1.1412847811551717e-05,
+      "loss": 0.3724,
+      "step": 17166
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 1.1410276191945688e-05,
+      "loss": 0.3848,
+      "step": 17167
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 1.140770457233966e-05,
+      "loss": 0.8266,
+      "step": 17168
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 1.1405132952733633e-05,
+      "loss": 0.4571,
+      "step": 17169
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 1.1402561333127604e-05,
+      "loss": 0.7044,
+      "step": 17170
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 1.1399989713521577e-05,
+      "loss": 0.814,
+      "step": 17171
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 1.1397418093915548e-05,
+      "loss": 0.6868,
+      "step": 17172
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 1.139484647430952e-05,
+      "loss": 0.6327,
+      "step": 17173
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 1.1392274854703493e-05,
+      "loss": 0.2307,
+      "step": 17174
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 1.1389703235097466e-05,
+      "loss": 0.6253,
+      "step": 17175
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 1.1387131615491437e-05,
+      "loss": 0.4891,
+      "step": 17176
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 1.1384559995885409e-05,
+      "loss": 0.6783,
+      "step": 17177
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 1.1381988376279382e-05,
+      "loss": 0.6712,
+      "step": 17178
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 1.1379416756673353e-05,
+      "loss": 0.634,
+      "step": 17179
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 1.1376845137067326e-05,
+      "loss": 0.7676,
+      "step": 17180
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 1.1374273517461298e-05,
+      "loss": 0.6563,
+      "step": 17181
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 1.1371701897855269e-05,
+      "loss": 0.639,
+      "step": 17182
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 1.1369130278249242e-05,
+      "loss": 0.485,
+      "step": 17183
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 1.1366558658643215e-05,
+      "loss": 0.4511,
+      "step": 17184
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 1.1363987039037185e-05,
+      "loss": 0.5476,
+      "step": 17185
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 1.1361415419431158e-05,
+      "loss": 0.4924,
+      "step": 17186
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 1.1358843799825131e-05,
+      "loss": 0.7483,
+      "step": 17187
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 1.1356272180219102e-05,
+      "loss": 0.3612,
+      "step": 17188
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 1.1353700560613074e-05,
+      "loss": 0.3712,
+      "step": 17189
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 1.1351128941007047e-05,
+      "loss": 0.5697,
+      "step": 17190
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 1.1348557321401018e-05,
+      "loss": 0.6394,
+      "step": 17191
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 1.1345985701794991e-05,
+      "loss": 0.4512,
+      "step": 17192
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 1.1343414082188963e-05,
+      "loss": 0.4729,
+      "step": 17193
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 1.1340842462582934e-05,
+      "loss": 0.854,
+      "step": 17194
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 1.1338270842976907e-05,
+      "loss": 0.6208,
+      "step": 17195
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 1.133569922337088e-05,
+      "loss": 0.7781,
+      "step": 17196
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 1.1333127603764852e-05,
+      "loss": 0.6894,
+      "step": 17197
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 1.1330555984158823e-05,
+      "loss": 0.4737,
+      "step": 17198
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 1.1327984364552796e-05,
+      "loss": 0.4794,
+      "step": 17199
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 1.1325412744946769e-05,
+      "loss": 0.7619,
+      "step": 17200
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 1.132284112534074e-05,
+      "loss": 0.7224,
+      "step": 17201
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 1.1320269505734712e-05,
+      "loss": 0.6195,
+      "step": 17202
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 1.1317697886128685e-05,
+      "loss": 0.6923,
+      "step": 17203
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 1.1315126266522656e-05,
+      "loss": 0.3941,
+      "step": 17204
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 1.131255464691663e-05,
+      "loss": 0.4556,
+      "step": 17205
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 1.13099830273106e-05,
+      "loss": 0.6603,
+      "step": 17206
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 1.1307411407704572e-05,
+      "loss": 0.7764,
+      "step": 17207
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 1.1304839788098545e-05,
+      "loss": 0.6097,
+      "step": 17208
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 1.1302268168492518e-05,
+      "loss": 0.4405,
+      "step": 17209
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 1.129969654888649e-05,
+      "loss": 0.539,
+      "step": 17210
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 1.1297124929280461e-05,
+      "loss": 0.8221,
+      "step": 17211
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 1.1294553309674434e-05,
+      "loss": 0.2855,
+      "step": 17212
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 1.1291981690068405e-05,
+      "loss": 0.5731,
+      "step": 17213
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 1.1289410070462379e-05,
+      "loss": 0.8697,
+      "step": 17214
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 1.128683845085635e-05,
+      "loss": 0.727,
+      "step": 17215
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 1.1284266831250321e-05,
+      "loss": 0.9403,
+      "step": 17216
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 1.1281695211644294e-05,
+      "loss": 0.5769,
+      "step": 17217
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 1.1279123592038267e-05,
+      "loss": 0.2386,
+      "step": 17218
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 1.1276551972432239e-05,
+      "loss": 0.7665,
+      "step": 17219
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 1.127398035282621e-05,
+      "loss": 0.6223,
+      "step": 17220
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 1.1271408733220183e-05,
+      "loss": 0.5444,
+      "step": 17221
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 1.1268837113614155e-05,
+      "loss": 0.7694,
+      "step": 17222
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 1.1266265494008128e-05,
+      "loss": 0.5702,
+      "step": 17223
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 1.1263693874402099e-05,
+      "loss": 0.6334,
+      "step": 17224
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 1.126112225479607e-05,
+      "loss": 0.4822,
+      "step": 17225
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 1.1258550635190044e-05,
+      "loss": 0.4653,
+      "step": 17226
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 1.1255979015584017e-05,
+      "loss": 0.549,
+      "step": 17227
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 1.1253407395977986e-05,
+      "loss": 0.4813,
+      "step": 17228
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 1.125083577637196e-05,
+      "loss": 0.2414,
+      "step": 17229
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 1.1248264156765932e-05,
+      "loss": 0.5136,
+      "step": 17230
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 1.1245692537159904e-05,
+      "loss": 0.552,
+      "step": 17231
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 1.1243120917553875e-05,
+      "loss": 0.7787,
+      "step": 17232
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 1.1240549297947848e-05,
+      "loss": 0.6519,
+      "step": 17233
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 1.123797767834182e-05,
+      "loss": 0.5811,
+      "step": 17234
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 1.1235406058735793e-05,
+      "loss": 0.7169,
+      "step": 17235
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 1.1232834439129764e-05,
+      "loss": 0.4924,
+      "step": 17236
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 1.1230262819523735e-05,
+      "loss": 0.6411,
+      "step": 17237
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 1.1227691199917709e-05,
+      "loss": 0.5518,
+      "step": 17238
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 1.1225119580311682e-05,
+      "loss": 0.4997,
+      "step": 17239
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 1.1222547960705653e-05,
+      "loss": 1.0369,
+      "step": 17240
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 1.1219976341099624e-05,
+      "loss": 0.6406,
+      "step": 17241
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 1.1217404721493597e-05,
+      "loss": 0.3786,
+      "step": 17242
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 1.1214833101887569e-05,
+      "loss": 0.8016,
+      "step": 17243
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 1.1212261482281542e-05,
+      "loss": 0.931,
+      "step": 17244
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 1.1209689862675513e-05,
+      "loss": 0.4925,
+      "step": 17245
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 1.1207118243069485e-05,
+      "loss": 0.7397,
+      "step": 17246
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 1.1204546623463458e-05,
+      "loss": 0.5123,
+      "step": 17247
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 1.120197500385743e-05,
+      "loss": 1.2011,
+      "step": 17248
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 1.1199403384251402e-05,
+      "loss": 0.7682,
+      "step": 17249
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 1.1196831764645374e-05,
+      "loss": 0.6849,
+      "step": 17250
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 1.1194260145039347e-05,
+      "loss": 0.4138,
+      "step": 17251
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 1.1191688525433318e-05,
+      "loss": 0.5084,
+      "step": 17252
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 1.1189116905827291e-05,
+      "loss": 0.5271,
+      "step": 17253
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 1.1186545286221262e-05,
+      "loss": 0.4484,
+      "step": 17254
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 1.1183973666615234e-05,
+      "loss": 0.417,
+      "step": 17255
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 1.1181402047009207e-05,
+      "loss": 0.6468,
+      "step": 17256
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 1.117883042740318e-05,
+      "loss": 0.3367,
+      "step": 17257
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 1.1176258807797151e-05,
+      "loss": 0.2038,
+      "step": 17258
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 1.1173687188191123e-05,
+      "loss": 0.6859,
+      "step": 17259
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 1.1171115568585096e-05,
+      "loss": 0.7125,
+      "step": 17260
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 1.1168543948979067e-05,
+      "loss": 0.5068,
+      "step": 17261
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 1.116597232937304e-05,
+      "loss": 0.3883,
+      "step": 17262
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 1.1163400709767012e-05,
+      "loss": 0.5485,
+      "step": 17263
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 1.1160829090160983e-05,
+      "loss": 0.5949,
+      "step": 17264
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 1.1158257470554956e-05,
+      "loss": 0.5153,
+      "step": 17265
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 1.115568585094893e-05,
+      "loss": 0.6948,
+      "step": 17266
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 1.1153114231342899e-05,
+      "loss": 0.8554,
+      "step": 17267
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 1.1150542611736872e-05,
+      "loss": 0.4756,
+      "step": 17268
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 1.1147970992130845e-05,
+      "loss": 0.8179,
+      "step": 17269
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 1.1145399372524816e-05,
+      "loss": 0.5133,
+      "step": 17270
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 1.1142827752918788e-05,
+      "loss": 0.664,
+      "step": 17271
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 1.1140256133312761e-05,
+      "loss": 0.8512,
+      "step": 17272
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 1.1137684513706734e-05,
+      "loss": 0.6941,
+      "step": 17273
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 1.1135112894100705e-05,
+      "loss": 0.5319,
+      "step": 17274
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 1.1132541274494677e-05,
+      "loss": 0.4821,
+      "step": 17275
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 1.112996965488865e-05,
+      "loss": 0.858,
+      "step": 17276
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 1.1127398035282621e-05,
+      "loss": 0.7002,
+      "step": 17277
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 1.1124826415676594e-05,
+      "loss": 0.3992,
+      "step": 17278
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 1.1122254796070566e-05,
+      "loss": 0.5767,
+      "step": 17279
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 1.1119683176464537e-05,
+      "loss": 0.4217,
+      "step": 17280
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 1.111711155685851e-05,
+      "loss": 0.5928,
+      "step": 17281
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 1.1114539937252483e-05,
+      "loss": 0.4823,
+      "step": 17282
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 1.1111968317646455e-05,
+      "loss": 0.5511,
+      "step": 17283
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 1.1109396698040426e-05,
+      "loss": 0.5331,
+      "step": 17284
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 1.1106825078434399e-05,
+      "loss": 0.683,
+      "step": 17285
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 1.110425345882837e-05,
+      "loss": 0.7286,
+      "step": 17286
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 1.1101681839222343e-05,
+      "loss": 0.5067,
+      "step": 17287
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 1.1099110219616315e-05,
+      "loss": 0.412,
+      "step": 17288
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 1.1096538600010286e-05,
+      "loss": 0.5081,
+      "step": 17289
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 1.109396698040426e-05,
+      "loss": 0.3795,
+      "step": 17290
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 1.1091395360798232e-05,
+      "loss": 0.8238,
+      "step": 17291
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 1.1088823741192204e-05,
+      "loss": 0.6067,
+      "step": 17292
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 1.1086252121586175e-05,
+      "loss": 0.5245,
+      "step": 17293
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 1.1083680501980148e-05,
+      "loss": 0.6876,
+      "step": 17294
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 1.108110888237412e-05,
+      "loss": 0.6853,
+      "step": 17295
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 1.1078537262768093e-05,
+      "loss": 0.6218,
+      "step": 17296
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 1.1075965643162064e-05,
+      "loss": 0.548,
+      "step": 17297
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 1.1073394023556035e-05,
+      "loss": 0.614,
+      "step": 17298
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 1.1070822403950008e-05,
+      "loss": 0.4315,
+      "step": 17299
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 1.1068250784343982e-05,
+      "loss": 0.3678,
+      "step": 17300
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 1.1065679164737953e-05,
+      "loss": 0.4738,
+      "step": 17301
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 1.1063107545131924e-05,
+      "loss": 0.6776,
+      "step": 17302
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 1.1060535925525897e-05,
+      "loss": 0.5263,
+      "step": 17303
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 1.1057964305919869e-05,
+      "loss": 0.649,
+      "step": 17304
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 1.1055392686313842e-05,
+      "loss": 0.6052,
+      "step": 17305
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 1.1052821066707813e-05,
+      "loss": 0.7594,
+      "step": 17306
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 1.1050249447101785e-05,
+      "loss": 0.7885,
+      "step": 17307
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 1.1047677827495758e-05,
+      "loss": 0.5646,
+      "step": 17308
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 1.104510620788973e-05,
+      "loss": 0.3538,
+      "step": 17309
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 1.10425345882837e-05,
+      "loss": 0.5395,
+      "step": 17310
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 1.1039962968677673e-05,
+      "loss": 0.6419,
+      "step": 17311
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 1.1037391349071647e-05,
+      "loss": 0.7356,
+      "step": 17312
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 1.1034819729465618e-05,
+      "loss": 0.4352,
+      "step": 17313
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 1.103224810985959e-05,
+      "loss": 0.6804,
+      "step": 17314
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 1.1029676490253562e-05,
+      "loss": 0.6981,
+      "step": 17315
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 1.1027104870647534e-05,
+      "loss": 0.3828,
+      "step": 17316
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 1.1024533251041507e-05,
+      "loss": 0.4233,
+      "step": 17317
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 1.1021961631435478e-05,
+      "loss": 0.719,
+      "step": 17318
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 1.101939001182945e-05,
+      "loss": 0.6537,
+      "step": 17319
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 1.1016818392223423e-05,
+      "loss": 0.4296,
+      "step": 17320
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 1.1014246772617396e-05,
+      "loss": 0.508,
+      "step": 17321
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 1.1011675153011367e-05,
+      "loss": 0.5502,
+      "step": 17322
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 1.1009103533405338e-05,
+      "loss": 0.7123,
+      "step": 17323
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 1.1006531913799312e-05,
+      "loss": 0.5317,
+      "step": 17324
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 1.1003960294193283e-05,
+      "loss": 0.7903,
+      "step": 17325
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 1.1001388674587256e-05,
+      "loss": 0.6634,
+      "step": 17326
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 1.0998817054981227e-05,
+      "loss": 0.7774,
+      "step": 17327
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 1.0996245435375199e-05,
+      "loss": 0.4959,
+      "step": 17328
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 1.0993673815769172e-05,
+      "loss": 0.7265,
+      "step": 17329
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 1.0991102196163145e-05,
+      "loss": 0.5308,
+      "step": 17330
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 1.0988530576557116e-05,
+      "loss": 0.3112,
+      "step": 17331
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 1.0985958956951088e-05,
+      "loss": 0.7961,
+      "step": 17332
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 1.098338733734506e-05,
+      "loss": 0.3096,
+      "step": 17333
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 1.0980815717739032e-05,
+      "loss": 0.4896,
+      "step": 17334
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 1.0978244098133005e-05,
+      "loss": 0.4953,
+      "step": 17335
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 1.0975672478526977e-05,
+      "loss": 0.6339,
+      "step": 17336
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 1.0973100858920948e-05,
+      "loss": 0.4875,
+      "step": 17337
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 1.0970529239314921e-05,
+      "loss": 0.688,
+      "step": 17338
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 1.0967957619708894e-05,
+      "loss": 0.593,
+      "step": 17339
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 1.0965386000102865e-05,
+      "loss": 0.3793,
+      "step": 17340
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 1.0962814380496837e-05,
+      "loss": 0.7689,
+      "step": 17341
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 1.096024276089081e-05,
+      "loss": 0.3354,
+      "step": 17342
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 1.0957671141284781e-05,
+      "loss": 0.6395,
+      "step": 17343
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 1.0955099521678754e-05,
+      "loss": 0.4611,
+      "step": 17344
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 1.0952527902072726e-05,
+      "loss": 0.5036,
+      "step": 17345
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 1.0949956282466697e-05,
+      "loss": 0.6862,
+      "step": 17346
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 1.094738466286067e-05,
+      "loss": 0.4541,
+      "step": 17347
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 1.0944813043254643e-05,
+      "loss": 0.4956,
+      "step": 17348
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 1.0942241423648615e-05,
+      "loss": 0.6053,
+      "step": 17349
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 1.0939669804042586e-05,
+      "loss": 0.7005,
+      "step": 17350
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 1.0937098184436559e-05,
+      "loss": 0.4665,
+      "step": 17351
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 1.0934526564830532e-05,
+      "loss": 0.5415,
+      "step": 17352
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 1.0931954945224502e-05,
+      "loss": 0.7753,
+      "step": 17353
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 1.0929383325618475e-05,
+      "loss": 0.76,
+      "step": 17354
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 1.0926811706012448e-05,
+      "loss": 0.5264,
+      "step": 17355
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 1.092424008640642e-05,
+      "loss": 0.6353,
+      "step": 17356
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 1.092166846680039e-05,
+      "loss": 0.7132,
+      "step": 17357
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 1.0919096847194364e-05,
+      "loss": 0.585,
+      "step": 17358
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 1.0916525227588335e-05,
+      "loss": 0.5973,
+      "step": 17359
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 1.0913953607982308e-05,
+      "loss": 0.7264,
+      "step": 17360
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 1.091138198837628e-05,
+      "loss": 0.6332,
+      "step": 17361
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 1.0908810368770251e-05,
+      "loss": 0.6084,
+      "step": 17362
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 1.0906238749164224e-05,
+      "loss": 0.5343,
+      "step": 17363
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 1.0903667129558197e-05,
+      "loss": 0.2948,
+      "step": 17364
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 1.0901095509952169e-05,
+      "loss": 0.6735,
+      "step": 17365
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 1.089852389034614e-05,
+      "loss": 0.4585,
+      "step": 17366
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 1.0895952270740113e-05,
+      "loss": 0.6874,
+      "step": 17367
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 1.0893380651134084e-05,
+      "loss": 0.428,
+      "step": 17368
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 1.0890809031528058e-05,
+      "loss": 0.7541,
+      "step": 17369
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 1.0888237411922029e-05,
+      "loss": 0.7094,
+      "step": 17370
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 1.0885665792316e-05,
+      "loss": 0.679,
+      "step": 17371
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 1.0883094172709973e-05,
+      "loss": 0.4644,
+      "step": 17372
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 1.0880522553103946e-05,
+      "loss": 0.6702,
+      "step": 17373
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 1.0877950933497918e-05,
+      "loss": 0.4804,
+      "step": 17374
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 1.087537931389189e-05,
+      "loss": 0.5105,
+      "step": 17375
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 1.0872807694285862e-05,
+      "loss": 0.699,
+      "step": 17376
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 1.0870236074679834e-05,
+      "loss": 0.4735,
+      "step": 17377
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 1.0867664455073807e-05,
+      "loss": 0.4892,
+      "step": 17378
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 1.0865092835467778e-05,
+      "loss": 0.716,
+      "step": 17379
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 1.086252121586175e-05,
+      "loss": 0.7106,
+      "step": 17380
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 1.0859949596255723e-05,
+      "loss": 0.72,
+      "step": 17381
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 1.0857377976649696e-05,
+      "loss": 0.7065,
+      "step": 17382
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 1.0854806357043667e-05,
+      "loss": 0.37,
+      "step": 17383
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 1.0852234737437638e-05,
+      "loss": 0.4131,
+      "step": 17384
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 1.0849663117831611e-05,
+      "loss": 0.4557,
+      "step": 17385
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 1.0847091498225583e-05,
+      "loss": 0.6433,
+      "step": 17386
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 1.0844519878619556e-05,
+      "loss": 0.6714,
+      "step": 17387
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 1.0841948259013527e-05,
+      "loss": 0.4457,
+      "step": 17388
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 1.0839376639407499e-05,
+      "loss": 0.4406,
+      "step": 17389
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 1.0836805019801472e-05,
+      "loss": 0.4525,
+      "step": 17390
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 1.0834233400195445e-05,
+      "loss": 0.5889,
+      "step": 17391
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 1.0831661780589414e-05,
+      "loss": 0.4676,
+      "step": 17392
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 1.0829090160983388e-05,
+      "loss": 0.871,
+      "step": 17393
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 1.082651854137736e-05,
+      "loss": 0.6495,
+      "step": 17394
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 1.0823946921771332e-05,
+      "loss": 0.2577,
+      "step": 17395
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 1.0821375302165303e-05,
+      "loss": 0.6454,
+      "step": 17396
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 1.0818803682559276e-05,
+      "loss": 0.8105,
+      "step": 17397
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 1.0816232062953248e-05,
+      "loss": 0.605,
+      "step": 17398
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 1.0813660443347221e-05,
+      "loss": 0.5152,
+      "step": 17399
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 1.0811088823741192e-05,
+      "loss": 0.4364,
+      "step": 17400
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 1.0808517204135164e-05,
+      "loss": 0.6509,
+      "step": 17401
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 1.0805945584529137e-05,
+      "loss": 0.4059,
+      "step": 17402
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 1.080337396492311e-05,
+      "loss": 0.27,
+      "step": 17403
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 1.0800802345317081e-05,
+      "loss": 0.789,
+      "step": 17404
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 1.0798230725711053e-05,
+      "loss": 0.7337,
+      "step": 17405
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 1.0795659106105026e-05,
+      "loss": 0.6732,
+      "step": 17406
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 1.0793087486498997e-05,
+      "loss": 0.5445,
+      "step": 17407
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 1.079051586689297e-05,
+      "loss": 0.562,
+      "step": 17408
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 1.0787944247286941e-05,
+      "loss": 0.5286,
+      "step": 17409
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 1.0785372627680913e-05,
+      "loss": 0.3018,
+      "step": 17410
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 1.0782801008074886e-05,
+      "loss": 0.5525,
+      "step": 17411
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 1.0780229388468859e-05,
+      "loss": 0.4778,
+      "step": 17412
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 1.077765776886283e-05,
+      "loss": 0.3865,
+      "step": 17413
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 1.0775086149256802e-05,
+      "loss": 0.5202,
+      "step": 17414
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 1.0772514529650775e-05,
+      "loss": 0.9101,
+      "step": 17415
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 1.0769942910044746e-05,
+      "loss": 0.7944,
+      "step": 17416
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 1.076737129043872e-05,
+      "loss": 0.8484,
+      "step": 17417
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 1.076479967083269e-05,
+      "loss": 0.7106,
+      "step": 17418
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 1.0762228051226662e-05,
+      "loss": 0.7713,
+      "step": 17419
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 1.0759656431620635e-05,
+      "loss": 0.4629,
+      "step": 17420
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 1.0757084812014608e-05,
+      "loss": 0.3651,
+      "step": 17421
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 1.075451319240858e-05,
+      "loss": 0.783,
+      "step": 17422
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 1.0751941572802551e-05,
+      "loss": 0.5926,
+      "step": 17423
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 1.0749369953196524e-05,
+      "loss": 0.6667,
+      "step": 17424
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 1.0746798333590497e-05,
+      "loss": 0.3844,
+      "step": 17425
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 1.0744226713984468e-05,
+      "loss": 0.7937,
+      "step": 17426
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 1.074165509437844e-05,
+      "loss": 1.1709,
+      "step": 17427
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 1.0739083474772413e-05,
+      "loss": 0.4482,
+      "step": 17428
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 1.0736511855166384e-05,
+      "loss": 0.4192,
+      "step": 17429
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 1.0733940235560357e-05,
+      "loss": 0.6349,
+      "step": 17430
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 1.0731368615954329e-05,
+      "loss": 0.5809,
+      "step": 17431
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 1.07287969963483e-05,
+      "loss": 0.3831,
+      "step": 17432
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 1.0726225376742273e-05,
+      "loss": 0.6026,
+      "step": 17433
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 1.0723653757136246e-05,
+      "loss": 0.4664,
+      "step": 17434
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 1.0721082137530216e-05,
+      "loss": 0.7043,
+      "step": 17435
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 1.0718510517924189e-05,
+      "loss": 0.3866,
+      "step": 17436
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 1.0715938898318162e-05,
+      "loss": 0.3385,
+      "step": 17437
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 1.0713367278712134e-05,
+      "loss": 0.3673,
+      "step": 17438
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 1.0710795659106105e-05,
+      "loss": 0.7573,
+      "step": 17439
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 1.0708224039500078e-05,
+      "loss": 0.6269,
+      "step": 17440
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 1.070565241989405e-05,
+      "loss": 0.7306,
+      "step": 17441
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 1.0703080800288022e-05,
+      "loss": 0.7102,
+      "step": 17442
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 1.0700509180681994e-05,
+      "loss": 0.4292,
+      "step": 17443
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 1.0697937561075965e-05,
+      "loss": 0.6594,
+      "step": 17444
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 1.0695365941469938e-05,
+      "loss": 0.7497,
+      "step": 17445
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 1.0692794321863911e-05,
+      "loss": 0.5646,
+      "step": 17446
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 1.0690222702257883e-05,
+      "loss": 0.7326,
+      "step": 17447
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 1.0687651082651854e-05,
+      "loss": 0.5356,
+      "step": 17448
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 1.0685079463045827e-05,
+      "loss": 0.3607,
+      "step": 17449
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 1.0682507843439799e-05,
+      "loss": 0.5174,
+      "step": 17450
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 1.0679936223833772e-05,
+      "loss": 0.58,
+      "step": 17451
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 1.0677364604227743e-05,
+      "loss": 0.6253,
+      "step": 17452
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 1.0674792984621714e-05,
+      "loss": 0.838,
+      "step": 17453
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 1.0672221365015687e-05,
+      "loss": 0.412,
+      "step": 17454
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 1.066964974540966e-05,
+      "loss": 0.3306,
+      "step": 17455
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 1.0667078125803632e-05,
+      "loss": 0.6477,
+      "step": 17456
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 1.0664506506197603e-05,
+      "loss": 0.7063,
+      "step": 17457
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 1.0661934886591576e-05,
+      "loss": 0.3932,
+      "step": 17458
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 1.0659363266985548e-05,
+      "loss": 0.3704,
+      "step": 17459
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 1.065679164737952e-05,
+      "loss": 0.4879,
+      "step": 17460
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 1.0654220027773492e-05,
+      "loss": 0.4876,
+      "step": 17461
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 1.0651648408167464e-05,
+      "loss": 0.276,
+      "step": 17462
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 1.0649076788561437e-05,
+      "loss": 0.9217,
+      "step": 17463
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 1.064650516895541e-05,
+      "loss": 0.6324,
+      "step": 17464
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 1.0643933549349381e-05,
+      "loss": 0.4422,
+      "step": 17465
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 1.0641361929743352e-05,
+      "loss": 0.566,
+      "step": 17466
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 1.0638790310137326e-05,
+      "loss": 0.5289,
+      "step": 17467
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 1.0636218690531297e-05,
+      "loss": 0.551,
+      "step": 17468
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 1.063364707092527e-05,
+      "loss": 0.7907,
+      "step": 17469
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 1.0631075451319241e-05,
+      "loss": 0.5474,
+      "step": 17470
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 1.0628503831713213e-05,
+      "loss": 0.6255,
+      "step": 17471
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 1.0625932212107186e-05,
+      "loss": 0.6253,
+      "step": 17472
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 1.0623360592501159e-05,
+      "loss": 0.5313,
+      "step": 17473
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 1.0620788972895129e-05,
+      "loss": 0.7896,
+      "step": 17474
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 1.0618217353289102e-05,
+      "loss": 0.4173,
+      "step": 17475
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 1.0615645733683075e-05,
+      "loss": 0.6008,
+      "step": 17476
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 1.0613074114077046e-05,
+      "loss": 0.4348,
+      "step": 17477
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 1.0610502494471017e-05,
+      "loss": 0.7003,
+      "step": 17478
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 1.060793087486499e-05,
+      "loss": 0.4563,
+      "step": 17479
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 1.0605359255258962e-05,
+      "loss": 0.3937,
+      "step": 17480
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 1.0602787635652935e-05,
+      "loss": 0.4886,
+      "step": 17481
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 1.0600216016046906e-05,
+      "loss": 0.2604,
+      "step": 17482
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 1.0597644396440878e-05,
+      "loss": 0.532,
+      "step": 17483
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 1.059507277683485e-05,
+      "loss": 0.8255,
+      "step": 17484
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 1.0592501157228824e-05,
+      "loss": 0.826,
+      "step": 17485
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 1.0589929537622795e-05,
+      "loss": 0.3741,
+      "step": 17486
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 1.0587357918016767e-05,
+      "loss": 1.0885,
+      "step": 17487
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 1.058478629841074e-05,
+      "loss": 0.7378,
+      "step": 17488
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 1.0582214678804711e-05,
+      "loss": 0.5848,
+      "step": 17489
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 1.0579643059198684e-05,
+      "loss": 0.5817,
+      "step": 17490
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 1.0577071439592656e-05,
+      "loss": 0.7547,
+      "step": 17491
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 1.0574499819986627e-05,
+      "loss": 0.9819,
+      "step": 17492
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 1.05719282003806e-05,
+      "loss": 0.6006,
+      "step": 17493
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 1.0569356580774573e-05,
+      "loss": 0.4167,
+      "step": 17494
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 1.0566784961168544e-05,
+      "loss": 0.611,
+      "step": 17495
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 1.0564213341562516e-05,
+      "loss": 0.6305,
+      "step": 17496
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 1.0561641721956489e-05,
+      "loss": 0.6731,
+      "step": 17497
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 1.0559070102350462e-05,
+      "loss": 0.9523,
+      "step": 17498
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 1.0556498482744433e-05,
+      "loss": 0.7139,
+      "step": 17499
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 1.0553926863138405e-05,
+      "loss": 0.6219,
+      "step": 17500
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 1.0551355243532378e-05,
+      "loss": 0.5882,
+      "step": 17501
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 1.054878362392635e-05,
+      "loss": 0.673,
+      "step": 17502
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 1.0546212004320322e-05,
+      "loss": 0.4885,
+      "step": 17503
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 1.0543640384714294e-05,
+      "loss": 0.478,
+      "step": 17504
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 1.0541068765108265e-05,
+      "loss": 0.4686,
+      "step": 17505
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 1.0538497145502238e-05,
+      "loss": 0.6744,
+      "step": 17506
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 1.0535925525896211e-05,
+      "loss": 0.6851,
+      "step": 17507
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 1.0533353906290183e-05,
+      "loss": 0.4348,
+      "step": 17508
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 1.0530782286684154e-05,
+      "loss": 0.7118,
+      "step": 17509
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 1.0528210667078127e-05,
+      "loss": 0.3846,
+      "step": 17510
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 1.0525639047472098e-05,
+      "loss": 0.535,
+      "step": 17511
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 1.0523067427866071e-05,
+      "loss": 0.6452,
+      "step": 17512
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 1.0520495808260043e-05,
+      "loss": 0.6345,
+      "step": 17513
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 1.0517924188654014e-05,
+      "loss": 0.5226,
+      "step": 17514
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 1.0515352569047987e-05,
+      "loss": 0.7393,
+      "step": 17515
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 1.051278094944196e-05,
+      "loss": 0.5699,
+      "step": 17516
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 1.051020932983593e-05,
+      "loss": 0.452,
+      "step": 17517
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 1.0507637710229903e-05,
+      "loss": 0.4926,
+      "step": 17518
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 1.0505066090623876e-05,
+      "loss": 0.4907,
+      "step": 17519
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 1.0502494471017848e-05,
+      "loss": 0.6114,
+      "step": 17520
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 1.0499922851411819e-05,
+      "loss": 0.675,
+      "step": 17521
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 1.0497351231805792e-05,
+      "loss": 0.7344,
+      "step": 17522
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 1.0494779612199763e-05,
+      "loss": 0.6357,
+      "step": 17523
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 1.0492207992593737e-05,
+      "loss": 0.5517,
+      "step": 17524
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 1.0489636372987708e-05,
+      "loss": 0.6923,
+      "step": 17525
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 1.048706475338168e-05,
+      "loss": 0.3707,
+      "step": 17526
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 1.0484493133775652e-05,
+      "loss": 0.3717,
+      "step": 17527
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 1.0481921514169625e-05,
+      "loss": 0.6447,
+      "step": 17528
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 1.0479349894563597e-05,
+      "loss": 0.5808,
+      "step": 17529
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 1.0476778274957568e-05,
+      "loss": 0.6532,
+      "step": 17530
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 1.0474206655351541e-05,
+      "loss": 0.5733,
+      "step": 17531
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 1.0471635035745513e-05,
+      "loss": 0.5266,
+      "step": 17532
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 1.0469063416139486e-05,
+      "loss": 0.2987,
+      "step": 17533
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 1.0466491796533457e-05,
+      "loss": 0.6605,
+      "step": 17534
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 1.0463920176927428e-05,
+      "loss": 0.566,
+      "step": 17535
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 1.0461348557321402e-05,
+      "loss": 0.4986,
+      "step": 17536
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 1.0458776937715375e-05,
+      "loss": 0.4597,
+      "step": 17537
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 1.0456205318109346e-05,
+      "loss": 0.6043,
+      "step": 17538
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 1.0453633698503317e-05,
+      "loss": 0.8793,
+      "step": 17539
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 1.045106207889729e-05,
+      "loss": 0.6202,
+      "step": 17540
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 1.0448490459291262e-05,
+      "loss": 0.5951,
+      "step": 17541
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 1.0445918839685235e-05,
+      "loss": 0.454,
+      "step": 17542
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 1.0443347220079206e-05,
+      "loss": 0.5628,
+      "step": 17543
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 1.0440775600473178e-05,
+      "loss": 0.8017,
+      "step": 17544
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 1.043820398086715e-05,
+      "loss": 0.6887,
+      "step": 17545
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 1.0435632361261124e-05,
+      "loss": 0.4453,
+      "step": 17546
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 1.0433060741655095e-05,
+      "loss": 0.3774,
+      "step": 17547
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 1.0430489122049067e-05,
+      "loss": 0.5312,
+      "step": 17548
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 1.042791750244304e-05,
+      "loss": 0.5234,
+      "step": 17549
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 1.0425345882837011e-05,
+      "loss": 0.3127,
+      "step": 17550
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 1.0422774263230984e-05,
+      "loss": 0.7116,
+      "step": 17551
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 1.0420202643624955e-05,
+      "loss": 0.3541,
+      "step": 17552
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 1.0417631024018927e-05,
+      "loss": 0.5724,
+      "step": 17553
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 1.04150594044129e-05,
+      "loss": 0.5228,
+      "step": 17554
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 1.0412487784806873e-05,
+      "loss": 0.7055,
+      "step": 17555
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 1.0409916165200843e-05,
+      "loss": 0.6907,
+      "step": 17556
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 1.0407344545594816e-05,
+      "loss": 0.639,
+      "step": 17557
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 1.0404772925988789e-05,
+      "loss": 0.5908,
+      "step": 17558
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 1.040220130638276e-05,
+      "loss": 0.5619,
+      "step": 17559
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 1.0399629686776732e-05,
+      "loss": 0.4392,
+      "step": 17560
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 1.0397058067170705e-05,
+      "loss": 0.5316,
+      "step": 17561
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 1.0394486447564676e-05,
+      "loss": 0.8625,
+      "step": 17562
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 1.0391914827958649e-05,
+      "loss": 0.5715,
+      "step": 17563
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 1.038934320835262e-05,
+      "loss": 0.4562,
+      "step": 17564
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 1.0386771588746592e-05,
+      "loss": 0.2606,
+      "step": 17565
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 1.0384199969140565e-05,
+      "loss": 0.9478,
+      "step": 17566
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 1.0381628349534538e-05,
+      "loss": 0.3516,
+      "step": 17567
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 1.037905672992851e-05,
+      "loss": 0.488,
+      "step": 17568
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 1.037648511032248e-05,
+      "loss": 0.7036,
+      "step": 17569
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 1.0373913490716454e-05,
+      "loss": 0.5695,
+      "step": 17570
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 1.0371341871110425e-05,
+      "loss": 0.6309,
+      "step": 17571
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 1.0368770251504398e-05,
+      "loss": 0.4442,
+      "step": 17572
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 1.036619863189837e-05,
+      "loss": 0.5617,
+      "step": 17573
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 1.0363627012292343e-05,
+      "loss": 0.4759,
+      "step": 17574
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 1.0361055392686314e-05,
+      "loss": 0.6856,
+      "step": 17575
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 1.0358483773080287e-05,
+      "loss": 0.4225,
+      "step": 17576
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 1.0355912153474259e-05,
+      "loss": 0.4326,
+      "step": 17577
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 1.035334053386823e-05,
+      "loss": 0.4211,
+      "step": 17578
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 1.0350768914262203e-05,
+      "loss": 0.4186,
+      "step": 17579
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 1.0348197294656176e-05,
+      "loss": 0.4558,
+      "step": 17580
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 1.0345625675050147e-05,
+      "loss": 0.692,
+      "step": 17581
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 1.0343054055444119e-05,
+      "loss": 0.3518,
+      "step": 17582
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 1.0340482435838092e-05,
+      "loss": 0.9658,
+      "step": 17583
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 1.0337910816232063e-05,
+      "loss": 0.5563,
+      "step": 17584
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 1.0335339196626036e-05,
+      "loss": 0.4183,
+      "step": 17585
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 1.0332767577020008e-05,
+      "loss": 0.343,
+      "step": 17586
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 1.0330195957413979e-05,
+      "loss": 0.4195,
+      "step": 17587
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 1.0327624337807952e-05,
+      "loss": 0.5709,
+      "step": 17588
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 1.0325052718201925e-05,
+      "loss": 0.5471,
+      "step": 17589
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 1.0322481098595897e-05,
+      "loss": 0.6618,
+      "step": 17590
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 1.0319909478989868e-05,
+      "loss": 0.6864,
+      "step": 17591
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 1.0317337859383841e-05,
+      "loss": 0.8099,
+      "step": 17592
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 1.0314766239777813e-05,
+      "loss": 0.5122,
+      "step": 17593
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 1.0312194620171786e-05,
+      "loss": 0.4973,
+      "step": 17594
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 1.0309623000565757e-05,
+      "loss": 0.5684,
+      "step": 17595
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 1.0307051380959728e-05,
+      "loss": 0.4851,
+      "step": 17596
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 1.0304479761353701e-05,
+      "loss": 0.6685,
+      "step": 17597
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 1.0301908141747674e-05,
+      "loss": 0.3967,
+      "step": 17598
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 1.0299336522141644e-05,
+      "loss": 0.3067,
+      "step": 17599
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 1.0296764902535617e-05,
+      "loss": 0.5836,
+      "step": 17600
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 1.029419328292959e-05,
+      "loss": 0.9396,
+      "step": 17601
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 1.0291621663323562e-05,
+      "loss": 0.5977,
+      "step": 17602
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 1.0289050043717533e-05,
+      "loss": 0.882,
+      "step": 17603
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 1.0286478424111506e-05,
+      "loss": 0.4255,
+      "step": 17604
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 1.0283906804505478e-05,
+      "loss": 0.7806,
+      "step": 17605
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 1.028133518489945e-05,
+      "loss": 0.6742,
+      "step": 17606
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 1.0278763565293422e-05,
+      "loss": 0.6576,
+      "step": 17607
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 1.0276191945687393e-05,
+      "loss": 0.4353,
+      "step": 17608
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 1.0273620326081366e-05,
+      "loss": 0.5149,
+      "step": 17609
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 1.027104870647534e-05,
+      "loss": 0.6219,
+      "step": 17610
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 1.0268477086869311e-05,
+      "loss": 0.3435,
+      "step": 17611
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 1.0265905467263282e-05,
+      "loss": 0.7233,
+      "step": 17612
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 1.0263333847657255e-05,
+      "loss": 0.5807,
+      "step": 17613
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 1.0260762228051227e-05,
+      "loss": 0.7098,
+      "step": 17614
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 1.02581906084452e-05,
+      "loss": 0.759,
+      "step": 17615
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 1.0255618988839171e-05,
+      "loss": 0.6083,
+      "step": 17616
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 1.0253047369233143e-05,
+      "loss": 0.5687,
+      "step": 17617
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 1.0250475749627116e-05,
+      "loss": 0.7473,
+      "step": 17618
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 1.0247904130021089e-05,
+      "loss": 0.7058,
+      "step": 17619
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 1.024533251041506e-05,
+      "loss": 0.5932,
+      "step": 17620
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 1.0242760890809031e-05,
+      "loss": 0.4492,
+      "step": 17621
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 1.0240189271203005e-05,
+      "loss": 0.7619,
+      "step": 17622
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 1.0237617651596976e-05,
+      "loss": 0.6239,
+      "step": 17623
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 1.0235046031990949e-05,
+      "loss": 0.4389,
+      "step": 17624
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 1.023247441238492e-05,
+      "loss": 0.3352,
+      "step": 17625
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 1.0229902792778892e-05,
+      "loss": 0.6422,
+      "step": 17626
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 1.0227331173172865e-05,
+      "loss": 0.3616,
+      "step": 17627
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 1.0224759553566838e-05,
+      "loss": 0.4592,
+      "step": 17628
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 1.022218793396081e-05,
+      "loss": 0.892,
+      "step": 17629
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 1.021961631435478e-05,
+      "loss": 0.4191,
+      "step": 17630
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 1.0217044694748754e-05,
+      "loss": 0.4906,
+      "step": 17631
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 1.0214473075142725e-05,
+      "loss": 0.5848,
+      "step": 17632
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 1.0211901455536698e-05,
+      "loss": 0.6608,
+      "step": 17633
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 1.020932983593067e-05,
+      "loss": 0.6523,
+      "step": 17634
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 1.0206758216324641e-05,
+      "loss": 0.5603,
+      "step": 17635
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 1.0204186596718614e-05,
+      "loss": 0.4083,
+      "step": 17636
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 1.0201614977112587e-05,
+      "loss": 0.4476,
+      "step": 17637
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 1.0199043357506557e-05,
+      "loss": 0.7928,
+      "step": 17638
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 1.019647173790053e-05,
+      "loss": 0.6902,
+      "step": 17639
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 1.0193900118294503e-05,
+      "loss": 0.6155,
+      "step": 17640
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 1.0191328498688474e-05,
+      "loss": 0.3417,
+      "step": 17641
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 1.0188756879082446e-05,
+      "loss": 0.5715,
+      "step": 17642
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 1.0186185259476419e-05,
+      "loss": 0.6314,
+      "step": 17643
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 1.018361363987039e-05,
+      "loss": 0.5277,
+      "step": 17644
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 1.0181042020264363e-05,
+      "loss": 0.6063,
+      "step": 17645
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 1.0178470400658335e-05,
+      "loss": 0.711,
+      "step": 17646
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 1.0175898781052306e-05,
+      "loss": 0.5068,
+      "step": 17647
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 1.0173327161446279e-05,
+      "loss": 0.5216,
+      "step": 17648
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 1.0170755541840252e-05,
+      "loss": 0.6487,
+      "step": 17649
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 1.0168183922234223e-05,
+      "loss": 0.8168,
+      "step": 17650
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 1.0165612302628195e-05,
+      "loss": 0.3405,
+      "step": 17651
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 1.0163040683022168e-05,
+      "loss": 0.5628,
+      "step": 17652
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 1.0160469063416141e-05,
+      "loss": 0.387,
+      "step": 17653
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 1.0157897443810112e-05,
+      "loss": 0.7332,
+      "step": 17654
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 1.0155325824204084e-05,
+      "loss": 0.7407,
+      "step": 17655
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 1.0152754204598057e-05,
+      "loss": 0.4964,
+      "step": 17656
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 1.0150182584992028e-05,
+      "loss": 0.3934,
+      "step": 17657
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 1.0147610965386001e-05,
+      "loss": 0.4583,
+      "step": 17658
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 1.0145039345779973e-05,
+      "loss": 0.5634,
+      "step": 17659
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 1.0142467726173944e-05,
+      "loss": 0.8521,
+      "step": 17660
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 1.0139896106567917e-05,
+      "loss": 0.4805,
+      "step": 17661
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 1.013732448696189e-05,
+      "loss": 0.5431,
+      "step": 17662
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 1.0134752867355862e-05,
+      "loss": 0.6913,
+      "step": 17663
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 1.0132181247749833e-05,
+      "loss": 0.3417,
+      "step": 17664
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 1.0129609628143806e-05,
+      "loss": 0.5716,
+      "step": 17665
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 1.0127038008537777e-05,
+      "loss": 0.6095,
+      "step": 17666
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 1.012446638893175e-05,
+      "loss": 0.7166,
+      "step": 17667
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 1.0121894769325722e-05,
+      "loss": 0.326,
+      "step": 17668
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 1.0119323149719693e-05,
+      "loss": 0.5298,
+      "step": 17669
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 1.0116751530113666e-05,
+      "loss": 0.3828,
+      "step": 17670
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 1.011417991050764e-05,
+      "loss": 0.5222,
+      "step": 17671
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 1.011160829090161e-05,
+      "loss": 0.5503,
+      "step": 17672
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 1.0109036671295582e-05,
+      "loss": 0.8766,
+      "step": 17673
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 1.0106465051689555e-05,
+      "loss": 0.6311,
+      "step": 17674
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 1.0103893432083527e-05,
+      "loss": 0.5248,
+      "step": 17675
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 1.01013218124775e-05,
+      "loss": 0.6402,
+      "step": 17676
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 1.0098750192871471e-05,
+      "loss": 0.6097,
+      "step": 17677
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 1.0096178573265442e-05,
+      "loss": 0.4687,
+      "step": 17678
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 1.0093606953659416e-05,
+      "loss": 0.5276,
+      "step": 17679
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 1.0091035334053389e-05,
+      "loss": 0.7232,
+      "step": 17680
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 1.0088463714447358e-05,
+      "loss": 0.4175,
+      "step": 17681
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 1.0085892094841331e-05,
+      "loss": 0.5345,
+      "step": 17682
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 1.0083320475235304e-05,
+      "loss": 0.6618,
+      "step": 17683
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 1.0080748855629276e-05,
+      "loss": 0.4386,
+      "step": 17684
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 1.0078177236023247e-05,
+      "loss": 0.5284,
+      "step": 17685
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 1.007560561641722e-05,
+      "loss": 0.5214,
+      "step": 17686
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 1.0073033996811192e-05,
+      "loss": 0.5166,
+      "step": 17687
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 1.0070462377205165e-05,
+      "loss": 0.4739,
+      "step": 17688
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 1.0067890757599136e-05,
+      "loss": 0.8369,
+      "step": 17689
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 1.0065319137993107e-05,
+      "loss": 0.573,
+      "step": 17690
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 1.006274751838708e-05,
+      "loss": 0.515,
+      "step": 17691
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 1.0060175898781054e-05,
+      "loss": 0.7448,
+      "step": 17692
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 1.0057604279175025e-05,
+      "loss": 0.8961,
+      "step": 17693
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 1.0055032659568996e-05,
+      "loss": 0.7881,
+      "step": 17694
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 1.005246103996297e-05,
+      "loss": 0.7239,
+      "step": 17695
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 1.004988942035694e-05,
+      "loss": 0.7585,
+      "step": 17696
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 1.0047317800750914e-05,
+      "loss": 0.4811,
+      "step": 17697
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 1.0044746181144885e-05,
+      "loss": 0.9449,
+      "step": 17698
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 1.0042174561538857e-05,
+      "loss": 0.536,
+      "step": 17699
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 1.003960294193283e-05,
+      "loss": 0.3416,
+      "step": 17700
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 1.0037031322326803e-05,
+      "loss": 0.5492,
+      "step": 17701
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 1.0034459702720774e-05,
+      "loss": 0.6284,
+      "step": 17702
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 1.0031888083114746e-05,
+      "loss": 0.51,
+      "step": 17703
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 1.0029316463508719e-05,
+      "loss": 0.6256,
+      "step": 17704
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 1.002674484390269e-05,
+      "loss": 0.4495,
+      "step": 17705
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 1.0024173224296663e-05,
+      "loss": 0.6722,
+      "step": 17706
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 1.0021601604690634e-05,
+      "loss": 0.8103,
+      "step": 17707
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 1.0019029985084606e-05,
+      "loss": 0.6491,
+      "step": 17708
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 1.0016458365478579e-05,
+      "loss": 0.6913,
+      "step": 17709
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 1.0013886745872552e-05,
+      "loss": 0.3343,
+      "step": 17710
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 1.0011315126266523e-05,
+      "loss": 0.7114,
+      "step": 17711
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 1.0008743506660495e-05,
+      "loss": 0.743,
+      "step": 17712
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 1.0006171887054468e-05,
+      "loss": 1.2726,
+      "step": 17713
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 1.000360026744844e-05,
+      "loss": 0.5568,
+      "step": 17714
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 1.0001028647842412e-05,
+      "loss": 0.6739,
+      "step": 17715
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 9.998457028236384e-06,
+      "loss": 0.571,
+      "step": 17716
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 9.995885408630355e-06,
+      "loss": 0.7246,
+      "step": 17717
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 9.993313789024328e-06,
+      "loss": 0.6814,
+      "step": 17718
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 9.990742169418301e-06,
+      "loss": 0.5983,
+      "step": 17719
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 9.988170549812271e-06,
+      "loss": 0.425,
+      "step": 17720
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 9.985598930206244e-06,
+      "loss": 0.7513,
+      "step": 17721
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 9.983027310600217e-06,
+      "loss": 0.6049,
+      "step": 17722
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 9.98045569099419e-06,
+      "loss": 0.7397,
+      "step": 17723
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 9.97788407138816e-06,
+      "loss": 0.5626,
+      "step": 17724
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 9.975312451782133e-06,
+      "loss": 0.6224,
+      "step": 17725
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 9.972740832176106e-06,
+      "loss": 0.5384,
+      "step": 17726
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 9.970169212570077e-06,
+      "loss": 1.0022,
+      "step": 17727
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 9.967597592964049e-06,
+      "loss": 0.5734,
+      "step": 17728
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 9.965025973358022e-06,
+      "loss": 0.4645,
+      "step": 17729
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 9.962454353751993e-06,
+      "loss": 0.7739,
+      "step": 17730
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 9.959882734145966e-06,
+      "loss": 0.955,
+      "step": 17731
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 9.957311114539938e-06,
+      "loss": 0.5748,
+      "step": 17732
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 9.954739494933909e-06,
+      "loss": 0.3933,
+      "step": 17733
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 9.952167875327882e-06,
+      "loss": 0.4837,
+      "step": 17734
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 9.949596255721855e-06,
+      "loss": 0.5866,
+      "step": 17735
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 9.947024636115826e-06,
+      "loss": 0.6752,
+      "step": 17736
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 9.944453016509798e-06,
+      "loss": 0.4068,
+      "step": 17737
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 9.941881396903771e-06,
+      "loss": 0.8175,
+      "step": 17738
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 9.939309777297742e-06,
+      "loss": 0.701,
+      "step": 17739
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 9.936738157691715e-06,
+      "loss": 0.3691,
+      "step": 17740
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 9.934166538085687e-06,
+      "loss": 0.4592,
+      "step": 17741
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 9.931594918479658e-06,
+      "loss": 0.5019,
+      "step": 17742
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 9.929023298873631e-06,
+      "loss": 0.6166,
+      "step": 17743
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 9.926451679267604e-06,
+      "loss": 0.8891,
+      "step": 17744
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 9.923880059661576e-06,
+      "loss": 0.7642,
+      "step": 17745
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 9.921308440055547e-06,
+      "loss": 0.5229,
+      "step": 17746
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 9.91873682044952e-06,
+      "loss": 0.3834,
+      "step": 17747
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 9.916165200843492e-06,
+      "loss": 0.6099,
+      "step": 17748
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 9.913593581237465e-06,
+      "loss": 0.7154,
+      "step": 17749
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 9.911021961631436e-06,
+      "loss": 0.6275,
+      "step": 17750
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 9.908450342025407e-06,
+      "loss": 0.574,
+      "step": 17751
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 9.90587872241938e-06,
+      "loss": 0.5416,
+      "step": 17752
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 9.903307102813353e-06,
+      "loss": 0.5013,
+      "step": 17753
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 9.900735483207325e-06,
+      "loss": 0.6687,
+      "step": 17754
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 9.898163863601296e-06,
+      "loss": 0.6322,
+      "step": 17755
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 9.89559224399527e-06,
+      "loss": 0.4534,
+      "step": 17756
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 9.89302062438924e-06,
+      "loss": 0.5848,
+      "step": 17757
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 9.890449004783214e-06,
+      "loss": 0.4623,
+      "step": 17758
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 9.887877385177185e-06,
+      "loss": 0.6157,
+      "step": 17759
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 9.885305765571157e-06,
+      "loss": 0.4294,
+      "step": 17760
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 9.88273414596513e-06,
+      "loss": 0.4332,
+      "step": 17761
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 9.880162526359103e-06,
+      "loss": 0.5758,
+      "step": 17762
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 9.877590906753072e-06,
+      "loss": 0.5437,
+      "step": 17763
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 9.875019287147045e-06,
+      "loss": 0.6078,
+      "step": 17764
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 9.872447667541019e-06,
+      "loss": 0.8934,
+      "step": 17765
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 9.86987604793499e-06,
+      "loss": 0.9684,
+      "step": 17766
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 9.867304428328961e-06,
+      "loss": 0.698,
+      "step": 17767
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 9.864732808722934e-06,
+      "loss": 0.5159,
+      "step": 17768
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 9.862161189116906e-06,
+      "loss": 0.4455,
+      "step": 17769
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 9.859589569510879e-06,
+      "loss": 0.8989,
+      "step": 17770
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 9.85701794990485e-06,
+      "loss": 0.4862,
+      "step": 17771
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 9.854446330298822e-06,
+      "loss": 0.8675,
+      "step": 17772
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 9.851874710692795e-06,
+      "loss": 0.8232,
+      "step": 17773
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 9.849303091086768e-06,
+      "loss": 0.5635,
+      "step": 17774
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 9.846731471480739e-06,
+      "loss": 0.59,
+      "step": 17775
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 9.84415985187471e-06,
+      "loss": 0.7603,
+      "step": 17776
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 9.841588232268684e-06,
+      "loss": 1.0192,
+      "step": 17777
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 9.839016612662655e-06,
+      "loss": 0.2931,
+      "step": 17778
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 9.836444993056628e-06,
+      "loss": 0.4243,
+      "step": 17779
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 9.8338733734506e-06,
+      "loss": 0.4161,
+      "step": 17780
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 9.83130175384457e-06,
+      "loss": 0.4276,
+      "step": 17781
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 9.828730134238544e-06,
+      "loss": 0.5623,
+      "step": 17782
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 9.826158514632517e-06,
+      "loss": 0.6954,
+      "step": 17783
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 9.823586895026488e-06,
+      "loss": 0.804,
+      "step": 17784
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 9.82101527542046e-06,
+      "loss": 0.3168,
+      "step": 17785
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 9.818443655814433e-06,
+      "loss": 0.403,
+      "step": 17786
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 9.815872036208404e-06,
+      "loss": 0.5096,
+      "step": 17787
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 9.813300416602377e-06,
+      "loss": 0.4205,
+      "step": 17788
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 9.810728796996349e-06,
+      "loss": 0.8772,
+      "step": 17789
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 9.80815717739032e-06,
+      "loss": 0.5387,
+      "step": 17790
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 9.805585557784293e-06,
+      "loss": 0.3969,
+      "step": 17791
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 9.803013938178266e-06,
+      "loss": 0.3423,
+      "step": 17792
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 9.800442318572236e-06,
+      "loss": 0.7103,
+      "step": 17793
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 9.797870698966209e-06,
+      "loss": 0.5964,
+      "step": 17794
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 9.795299079360182e-06,
+      "loss": 0.7785,
+      "step": 17795
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 9.792727459754153e-06,
+      "loss": 0.5336,
+      "step": 17796
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 9.790155840148126e-06,
+      "loss": 0.3798,
+      "step": 17797
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 9.787584220542098e-06,
+      "loss": 0.7336,
+      "step": 17798
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 9.78501260093607e-06,
+      "loss": 0.4621,
+      "step": 17799
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 9.782440981330042e-06,
+      "loss": 0.4824,
+      "step": 17800
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 9.779869361724015e-06,
+      "loss": 0.3939,
+      "step": 17801
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 9.777297742117987e-06,
+      "loss": 0.7992,
+      "step": 17802
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 9.774726122511958e-06,
+      "loss": 0.6301,
+      "step": 17803
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 9.772154502905931e-06,
+      "loss": 0.5114,
+      "step": 17804
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 9.769582883299904e-06,
+      "loss": 0.4158,
+      "step": 17805
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 9.767011263693874e-06,
+      "loss": 0.4125,
+      "step": 17806
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 9.764439644087847e-06,
+      "loss": 0.7014,
+      "step": 17807
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 9.76186802448182e-06,
+      "loss": 0.5473,
+      "step": 17808
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 9.759296404875791e-06,
+      "loss": 0.4187,
+      "step": 17809
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 9.756724785269763e-06,
+      "loss": 0.5627,
+      "step": 17810
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 9.754153165663736e-06,
+      "loss": 0.7599,
+      "step": 17811
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 9.751581546057707e-06,
+      "loss": 0.4435,
+      "step": 17812
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 9.74900992645168e-06,
+      "loss": 0.4587,
+      "step": 17813
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 9.746438306845652e-06,
+      "loss": 0.5864,
+      "step": 17814
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 9.743866687239623e-06,
+      "loss": 0.507,
+      "step": 17815
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 9.741295067633596e-06,
+      "loss": 0.5882,
+      "step": 17816
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 9.73872344802757e-06,
+      "loss": 0.5422,
+      "step": 17817
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 9.73615182842154e-06,
+      "loss": 0.9296,
+      "step": 17818
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 9.733580208815512e-06,
+      "loss": 0.826,
+      "step": 17819
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 9.731008589209485e-06,
+      "loss": 0.5267,
+      "step": 17820
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 9.728436969603456e-06,
+      "loss": 0.4761,
+      "step": 17821
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 9.72586534999743e-06,
+      "loss": 0.6029,
+      "step": 17822
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 9.723293730391401e-06,
+      "loss": 0.5593,
+      "step": 17823
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 9.720722110785372e-06,
+      "loss": 0.3863,
+      "step": 17824
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 9.718150491179345e-06,
+      "loss": 0.5914,
+      "step": 17825
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 9.715578871573318e-06,
+      "loss": 0.6177,
+      "step": 17826
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 9.71300725196729e-06,
+      "loss": 0.5406,
+      "step": 17827
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 9.710435632361261e-06,
+      "loss": 0.6872,
+      "step": 17828
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 9.707864012755234e-06,
+      "loss": 0.9672,
+      "step": 17829
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 9.705292393149206e-06,
+      "loss": 0.5389,
+      "step": 17830
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 9.702720773543179e-06,
+      "loss": 0.8356,
+      "step": 17831
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 9.70014915393715e-06,
+      "loss": 0.697,
+      "step": 17832
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 9.697577534331121e-06,
+      "loss": 0.6131,
+      "step": 17833
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 9.695005914725094e-06,
+      "loss": 0.3873,
+      "step": 17834
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 9.692434295119068e-06,
+      "loss": 0.706,
+      "step": 17835
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 9.689862675513039e-06,
+      "loss": 0.8396,
+      "step": 17836
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 9.68729105590701e-06,
+      "loss": 0.3923,
+      "step": 17837
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 9.684719436300983e-06,
+      "loss": 0.5093,
+      "step": 17838
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 9.682147816694955e-06,
+      "loss": 0.4656,
+      "step": 17839
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 9.679576197088928e-06,
+      "loss": 0.364,
+      "step": 17840
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 9.6770045774829e-06,
+      "loss": 0.5626,
+      "step": 17841
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 9.67443295787687e-06,
+      "loss": 0.9771,
+      "step": 17842
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 9.671861338270844e-06,
+      "loss": 0.953,
+      "step": 17843
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 9.669289718664817e-06,
+      "loss": 0.5379,
+      "step": 17844
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 9.666718099058786e-06,
+      "loss": 0.5395,
+      "step": 17845
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 9.66414647945276e-06,
+      "loss": 0.4448,
+      "step": 17846
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 9.661574859846733e-06,
+      "loss": 0.4733,
+      "step": 17847
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 9.659003240240704e-06,
+      "loss": 0.4377,
+      "step": 17848
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 9.656431620634675e-06,
+      "loss": 0.5876,
+      "step": 17849
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 9.653860001028648e-06,
+      "loss": 0.4247,
+      "step": 17850
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 9.65128838142262e-06,
+      "loss": 0.5177,
+      "step": 17851
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 9.648716761816593e-06,
+      "loss": 0.4613,
+      "step": 17852
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 9.646145142210564e-06,
+      "loss": 0.4433,
+      "step": 17853
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 9.643573522604536e-06,
+      "loss": 0.6234,
+      "step": 17854
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 9.641001902998509e-06,
+      "loss": 0.7559,
+      "step": 17855
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 9.638430283392482e-06,
+      "loss": 0.8776,
+      "step": 17856
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 9.635858663786453e-06,
+      "loss": 0.5069,
+      "step": 17857
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 9.633287044180425e-06,
+      "loss": 0.3586,
+      "step": 17858
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 9.630715424574398e-06,
+      "loss": 0.3006,
+      "step": 17859
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 9.628143804968369e-06,
+      "loss": 0.7363,
+      "step": 17860
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 9.625572185362342e-06,
+      "loss": 0.7634,
+      "step": 17861
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 9.623000565756313e-06,
+      "loss": 0.4917,
+      "step": 17862
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 9.620428946150285e-06,
+      "loss": 0.7406,
+      "step": 17863
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 9.617857326544258e-06,
+      "loss": 0.4609,
+      "step": 17864
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 9.615285706938231e-06,
+      "loss": 0.4176,
+      "step": 17865
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 9.612714087332202e-06,
+      "loss": 0.9289,
+      "step": 17866
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 9.610142467726174e-06,
+      "loss": 0.658,
+      "step": 17867
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 9.607570848120147e-06,
+      "loss": 0.8802,
+      "step": 17868
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 9.604999228514118e-06,
+      "loss": 0.7863,
+      "step": 17869
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 9.602427608908091e-06,
+      "loss": 0.7695,
+      "step": 17870
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 9.599855989302063e-06,
+      "loss": 0.3927,
+      "step": 17871
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 9.597284369696034e-06,
+      "loss": 0.3093,
+      "step": 17872
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 9.594712750090007e-06,
+      "loss": 0.6897,
+      "step": 17873
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 9.59214113048398e-06,
+      "loss": 0.7119,
+      "step": 17874
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 9.589569510877952e-06,
+      "loss": 0.895,
+      "step": 17875
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 9.586997891271923e-06,
+      "loss": 0.4298,
+      "step": 17876
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 9.584426271665896e-06,
+      "loss": 0.5451,
+      "step": 17877
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 9.581854652059869e-06,
+      "loss": 0.6097,
+      "step": 17878
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 9.57928303245384e-06,
+      "loss": 0.6832,
+      "step": 17879
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 9.576711412847812e-06,
+      "loss": 0.6833,
+      "step": 17880
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 9.574139793241785e-06,
+      "loss": 0.6397,
+      "step": 17881
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 9.571568173635756e-06,
+      "loss": 0.6422,
+      "step": 17882
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 9.56899655402973e-06,
+      "loss": 0.4271,
+      "step": 17883
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 9.5664249344237e-06,
+      "loss": 0.7894,
+      "step": 17884
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 9.563853314817672e-06,
+      "loss": 0.7601,
+      "step": 17885
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 9.561281695211645e-06,
+      "loss": 0.5745,
+      "step": 17886
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 9.558710075605618e-06,
+      "loss": 0.7308,
+      "step": 17887
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 9.556138455999588e-06,
+      "loss": 0.6709,
+      "step": 17888
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 9.553566836393561e-06,
+      "loss": 1.0347,
+      "step": 17889
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 9.550995216787534e-06,
+      "loss": 0.7129,
+      "step": 17890
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 9.548423597181505e-06,
+      "loss": 0.6242,
+      "step": 17891
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 9.545851977575477e-06,
+      "loss": 0.3167,
+      "step": 17892
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 9.54328035796945e-06,
+      "loss": 0.7669,
+      "step": 17893
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 9.540708738363421e-06,
+      "loss": 0.6618,
+      "step": 17894
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 9.538137118757394e-06,
+      "loss": 0.9394,
+      "step": 17895
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 9.535565499151366e-06,
+      "loss": 0.6942,
+      "step": 17896
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 9.532993879545337e-06,
+      "loss": 0.3087,
+      "step": 17897
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 9.53042225993931e-06,
+      "loss": 0.6524,
+      "step": 17898
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 9.527850640333283e-06,
+      "loss": 0.2506,
+      "step": 17899
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 9.525279020727255e-06,
+      "loss": 0.4303,
+      "step": 17900
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 9.522707401121226e-06,
+      "loss": 0.9203,
+      "step": 17901
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 9.520135781515199e-06,
+      "loss": 0.4168,
+      "step": 17902
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 9.51756416190917e-06,
+      "loss": 0.5477,
+      "step": 17903
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 9.514992542303144e-06,
+      "loss": 0.766,
+      "step": 17904
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 9.512420922697115e-06,
+      "loss": 0.3151,
+      "step": 17905
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 9.509849303091086e-06,
+      "loss": 0.3712,
+      "step": 17906
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 9.50727768348506e-06,
+      "loss": 0.6639,
+      "step": 17907
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 9.504706063879032e-06,
+      "loss": 0.2987,
+      "step": 17908
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 9.502134444273004e-06,
+      "loss": 0.5111,
+      "step": 17909
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 9.499562824666975e-06,
+      "loss": 0.5259,
+      "step": 17910
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 9.496991205060948e-06,
+      "loss": 0.8127,
+      "step": 17911
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 9.49441958545492e-06,
+      "loss": 0.543,
+      "step": 17912
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 9.491847965848893e-06,
+      "loss": 0.6365,
+      "step": 17913
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 9.489276346242864e-06,
+      "loss": 0.4629,
+      "step": 17914
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 9.486704726636836e-06,
+      "loss": 0.6369,
+      "step": 17915
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 9.484133107030809e-06,
+      "loss": 0.7618,
+      "step": 17916
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 9.481561487424782e-06,
+      "loss": 0.6232,
+      "step": 17917
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 9.478989867818753e-06,
+      "loss": 0.534,
+      "step": 17918
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 9.476418248212724e-06,
+      "loss": 0.7585,
+      "step": 17919
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 9.473846628606697e-06,
+      "loss": 0.8022,
+      "step": 17920
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 9.471275009000669e-06,
+      "loss": 0.4127,
+      "step": 17921
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 9.468703389394642e-06,
+      "loss": 0.5553,
+      "step": 17922
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 9.466131769788613e-06,
+      "loss": 0.3386,
+      "step": 17923
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 9.463560150182585e-06,
+      "loss": 0.6587,
+      "step": 17924
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 9.460988530576558e-06,
+      "loss": 0.416,
+      "step": 17925
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 9.458416910970531e-06,
+      "loss": 0.4048,
+      "step": 17926
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 9.4558452913645e-06,
+      "loss": 0.6155,
+      "step": 17927
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 9.453273671758474e-06,
+      "loss": 0.5005,
+      "step": 17928
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 9.450702052152447e-06,
+      "loss": 0.4912,
+      "step": 17929
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 9.448130432546418e-06,
+      "loss": 0.2946,
+      "step": 17930
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 9.44555881294039e-06,
+      "loss": 1.0272,
+      "step": 17931
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 9.442987193334363e-06,
+      "loss": 0.6536,
+      "step": 17932
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 9.440415573728334e-06,
+      "loss": 0.6013,
+      "step": 17933
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 9.437843954122307e-06,
+      "loss": 0.7205,
+      "step": 17934
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 9.435272334516278e-06,
+      "loss": 0.8185,
+      "step": 17935
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 9.43270071491025e-06,
+      "loss": 0.4458,
+      "step": 17936
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 9.430129095304223e-06,
+      "loss": 0.7015,
+      "step": 17937
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 9.427557475698196e-06,
+      "loss": 0.512,
+      "step": 17938
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 9.424985856092167e-06,
+      "loss": 0.4402,
+      "step": 17939
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 9.422414236486139e-06,
+      "loss": 0.4602,
+      "step": 17940
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 9.419842616880112e-06,
+      "loss": 0.7704,
+      "step": 17941
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 9.417270997274083e-06,
+      "loss": 0.5484,
+      "step": 17942
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 9.414699377668056e-06,
+      "loss": 0.6544,
+      "step": 17943
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 9.412127758062028e-06,
+      "loss": 0.5625,
+      "step": 17944
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 9.409556138455999e-06,
+      "loss": 0.5932,
+      "step": 17945
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 9.406984518849972e-06,
+      "loss": 0.5503,
+      "step": 17946
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 9.404412899243945e-06,
+      "loss": 0.8339,
+      "step": 17947
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 9.401841279637916e-06,
+      "loss": 0.5795,
+      "step": 17948
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 9.399269660031888e-06,
+      "loss": 0.4381,
+      "step": 17949
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 9.396698040425861e-06,
+      "loss": 0.4383,
+      "step": 17950
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 9.394126420819834e-06,
+      "loss": 0.4881,
+      "step": 17951
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 9.391554801213805e-06,
+      "loss": 0.4886,
+      "step": 17952
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 9.388983181607777e-06,
+      "loss": 0.7434,
+      "step": 17953
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 9.38641156200175e-06,
+      "loss": 0.5113,
+      "step": 17954
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 9.383839942395721e-06,
+      "loss": 0.593,
+      "step": 17955
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 9.381268322789694e-06,
+      "loss": 0.8954,
+      "step": 17956
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 9.378696703183666e-06,
+      "loss": 0.7744,
+      "step": 17957
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 9.376125083577637e-06,
+      "loss": 0.8539,
+      "step": 17958
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 9.37355346397161e-06,
+      "loss": 0.5732,
+      "step": 17959
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 9.370981844365583e-06,
+      "loss": 0.4853,
+      "step": 17960
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 9.368410224759555e-06,
+      "loss": 0.869,
+      "step": 17961
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 9.365838605153526e-06,
+      "loss": 0.6941,
+      "step": 17962
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 9.363266985547499e-06,
+      "loss": 0.5586,
+      "step": 17963
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 9.36069536594147e-06,
+      "loss": 0.4097,
+      "step": 17964
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 9.358123746335443e-06,
+      "loss": 0.6601,
+      "step": 17965
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 9.355552126729415e-06,
+      "loss": 0.5689,
+      "step": 17966
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 9.352980507123386e-06,
+      "loss": 0.5984,
+      "step": 17967
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 9.35040888751736e-06,
+      "loss": 0.512,
+      "step": 17968
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 9.347837267911332e-06,
+      "loss": 0.6363,
+      "step": 17969
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 9.345265648305302e-06,
+      "loss": 0.6296,
+      "step": 17970
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 9.342694028699275e-06,
+      "loss": 0.5945,
+      "step": 17971
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 9.340122409093248e-06,
+      "loss": 0.4788,
+      "step": 17972
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 9.33755078948722e-06,
+      "loss": 0.8496,
+      "step": 17973
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 9.334979169881191e-06,
+      "loss": 0.5812,
+      "step": 17974
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 9.332407550275164e-06,
+      "loss": 0.5191,
+      "step": 17975
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 9.329835930669135e-06,
+      "loss": 0.5717,
+      "step": 17976
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 9.327264311063108e-06,
+      "loss": 0.6341,
+      "step": 17977
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 9.32469269145708e-06,
+      "loss": 0.5154,
+      "step": 17978
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 9.322121071851051e-06,
+      "loss": 0.8306,
+      "step": 17979
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 9.319549452245024e-06,
+      "loss": 0.4171,
+      "step": 17980
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 9.316977832638997e-06,
+      "loss": 0.779,
+      "step": 17981
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 9.314406213032969e-06,
+      "loss": 0.563,
+      "step": 17982
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 9.31183459342694e-06,
+      "loss": 0.4731,
+      "step": 17983
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 9.309262973820913e-06,
+      "loss": 0.8518,
+      "step": 17984
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 9.306691354214885e-06,
+      "loss": 0.5358,
+      "step": 17985
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 9.304119734608858e-06,
+      "loss": 0.4245,
+      "step": 17986
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 9.301548115002829e-06,
+      "loss": 0.5334,
+      "step": 17987
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 9.2989764953968e-06,
+      "loss": 0.7499,
+      "step": 17988
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 9.296404875790773e-06,
+      "loss": 0.6054,
+      "step": 17989
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 9.293833256184747e-06,
+      "loss": 0.4715,
+      "step": 17990
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 9.291261636578718e-06,
+      "loss": 0.5329,
+      "step": 17991
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 9.28869001697269e-06,
+      "loss": 0.6867,
+      "step": 17992
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 9.286118397366662e-06,
+      "loss": 0.4539,
+      "step": 17993
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 9.283546777760634e-06,
+      "loss": 0.7412,
+      "step": 17994
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 9.280975158154607e-06,
+      "loss": 0.4576,
+      "step": 17995
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 9.278403538548578e-06,
+      "loss": 0.6777,
+      "step": 17996
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 9.27583191894255e-06,
+      "loss": 0.7838,
+      "step": 17997
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 9.273260299336523e-06,
+      "loss": 0.6108,
+      "step": 17998
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 9.270688679730496e-06,
+      "loss": 0.5331,
+      "step": 17999
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 9.268117060124465e-06,
+      "loss": 0.6222,
+      "step": 18000
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 9.265545440518439e-06,
+      "loss": 0.5177,
+      "step": 18001
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 9.262973820912412e-06,
+      "loss": 0.6884,
+      "step": 18002
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 9.260402201306383e-06,
+      "loss": 0.3867,
+      "step": 18003
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 9.257830581700356e-06,
+      "loss": 0.7341,
+      "step": 18004
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 9.255258962094327e-06,
+      "loss": 0.5286,
+      "step": 18005
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 9.252687342488299e-06,
+      "loss": 0.3421,
+      "step": 18006
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 9.250115722882272e-06,
+      "loss": 0.8227,
+      "step": 18007
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 9.247544103276245e-06,
+      "loss": 0.6168,
+      "step": 18008
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 9.244972483670215e-06,
+      "loss": 0.621,
+      "step": 18009
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 9.242400864064188e-06,
+      "loss": 0.5046,
+      "step": 18010
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 9.23982924445816e-06,
+      "loss": 0.6268,
+      "step": 18011
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 9.237257624852132e-06,
+      "loss": 0.5232,
+      "step": 18012
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 9.234686005246104e-06,
+      "loss": 0.394,
+      "step": 18013
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 9.232114385640077e-06,
+      "loss": 0.5715,
+      "step": 18014
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 9.229542766034048e-06,
+      "loss": 0.5608,
+      "step": 18015
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 9.226971146428021e-06,
+      "loss": 0.6355,
+      "step": 18016
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 9.224399526821992e-06,
+      "loss": 0.8218,
+      "step": 18017
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 9.221827907215964e-06,
+      "loss": 0.8067,
+      "step": 18018
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 9.219256287609937e-06,
+      "loss": 0.9278,
+      "step": 18019
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 9.21668466800391e-06,
+      "loss": 0.87,
+      "step": 18020
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 9.214113048397881e-06,
+      "loss": 0.1904,
+      "step": 18021
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 9.211541428791853e-06,
+      "loss": 0.5333,
+      "step": 18022
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 9.208969809185826e-06,
+      "loss": 0.6724,
+      "step": 18023
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 9.206398189579797e-06,
+      "loss": 0.4413,
+      "step": 18024
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 9.20382656997377e-06,
+      "loss": 0.6314,
+      "step": 18025
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 9.201254950367742e-06,
+      "loss": 0.368,
+      "step": 18026
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 9.198683330761715e-06,
+      "loss": 0.5393,
+      "step": 18027
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 9.196111711155686e-06,
+      "loss": 0.5958,
+      "step": 18028
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 9.19354009154966e-06,
+      "loss": 0.6046,
+      "step": 18029
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 9.19096847194363e-06,
+      "loss": 0.4828,
+      "step": 18030
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 9.188396852337602e-06,
+      "loss": 0.492,
+      "step": 18031
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 9.185825232731575e-06,
+      "loss": 0.6266,
+      "step": 18032
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 9.183253613125548e-06,
+      "loss": 0.8853,
+      "step": 18033
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 9.18068199351952e-06,
+      "loss": 0.8072,
+      "step": 18034
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 9.17811037391349e-06,
+      "loss": 0.3836,
+      "step": 18035
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 9.175538754307464e-06,
+      "loss": 0.3428,
+      "step": 18036
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 9.172967134701435e-06,
+      "loss": 0.6638,
+      "step": 18037
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 9.170395515095408e-06,
+      "loss": 0.7327,
+      "step": 18038
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 9.16782389548938e-06,
+      "loss": 0.5874,
+      "step": 18039
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 9.165252275883351e-06,
+      "loss": 0.5,
+      "step": 18040
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 9.162680656277324e-06,
+      "loss": 0.6311,
+      "step": 18041
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 9.160109036671297e-06,
+      "loss": 0.8867,
+      "step": 18042
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 9.157537417065269e-06,
+      "loss": 0.5654,
+      "step": 18043
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 9.15496579745924e-06,
+      "loss": 0.7623,
+      "step": 18044
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 9.152394177853213e-06,
+      "loss": 0.7545,
+      "step": 18045
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 9.149822558247184e-06,
+      "loss": 0.4009,
+      "step": 18046
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 9.147250938641158e-06,
+      "loss": 0.5405,
+      "step": 18047
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 9.144679319035129e-06,
+      "loss": 0.8419,
+      "step": 18048
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 9.1421076994291e-06,
+      "loss": 0.4364,
+      "step": 18049
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 9.139536079823073e-06,
+      "loss": 0.6115,
+      "step": 18050
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 9.136964460217046e-06,
+      "loss": 0.4733,
+      "step": 18051
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 9.134392840611016e-06,
+      "loss": 0.5047,
+      "step": 18052
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 9.13182122100499e-06,
+      "loss": 0.7234,
+      "step": 18053
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 9.129249601398962e-06,
+      "loss": 0.5108,
+      "step": 18054
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 9.126677981792934e-06,
+      "loss": 0.5019,
+      "step": 18055
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 9.124106362186905e-06,
+      "loss": 0.5971,
+      "step": 18056
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 9.121534742580878e-06,
+      "loss": 0.8941,
+      "step": 18057
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 9.11896312297485e-06,
+      "loss": 0.3366,
+      "step": 18058
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 9.116391503368823e-06,
+      "loss": 0.6624,
+      "step": 18059
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 9.113819883762794e-06,
+      "loss": 0.6026,
+      "step": 18060
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 9.111248264156765e-06,
+      "loss": 0.6955,
+      "step": 18061
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 9.108676644550738e-06,
+      "loss": 0.6328,
+      "step": 18062
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 9.106105024944711e-06,
+      "loss": 0.5711,
+      "step": 18063
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 9.103533405338683e-06,
+      "loss": 0.9369,
+      "step": 18064
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 9.100961785732654e-06,
+      "loss": 0.6373,
+      "step": 18065
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 9.098390166126627e-06,
+      "loss": 0.624,
+      "step": 18066
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 9.095818546520599e-06,
+      "loss": 0.9727,
+      "step": 18067
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 9.093246926914572e-06,
+      "loss": 0.5563,
+      "step": 18068
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 9.090675307308543e-06,
+      "loss": 0.3644,
+      "step": 18069
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 9.088103687702515e-06,
+      "loss": 0.6354,
+      "step": 18070
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 9.085532068096488e-06,
+      "loss": 0.6303,
+      "step": 18071
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 9.08296044849046e-06,
+      "loss": 0.5162,
+      "step": 18072
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 9.080388828884432e-06,
+      "loss": 0.6147,
+      "step": 18073
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 9.077817209278403e-06,
+      "loss": 0.6404,
+      "step": 18074
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 9.075245589672376e-06,
+      "loss": 0.5065,
+      "step": 18075
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 9.072673970066348e-06,
+      "loss": 0.7194,
+      "step": 18076
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 9.070102350460321e-06,
+      "loss": 0.7087,
+      "step": 18077
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 9.067530730854292e-06,
+      "loss": 0.5905,
+      "step": 18078
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 9.064959111248264e-06,
+      "loss": 0.7278,
+      "step": 18079
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 9.062387491642237e-06,
+      "loss": 0.3087,
+      "step": 18080
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 9.05981587203621e-06,
+      "loss": 0.6762,
+      "step": 18081
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 9.05724425243018e-06,
+      "loss": 0.5283,
+      "step": 18082
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 9.054672632824153e-06,
+      "loss": 0.5094,
+      "step": 18083
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 9.052101013218126e-06,
+      "loss": 0.4904,
+      "step": 18084
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 9.049529393612097e-06,
+      "loss": 0.4279,
+      "step": 18085
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 9.04695777400607e-06,
+      "loss": 0.552,
+      "step": 18086
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 9.044386154400042e-06,
+      "loss": 0.5857,
+      "step": 18087
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 9.041814534794013e-06,
+      "loss": 0.9755,
+      "step": 18088
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 9.039242915187986e-06,
+      "loss": 0.7271,
+      "step": 18089
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 9.036671295581959e-06,
+      "loss": 0.6526,
+      "step": 18090
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 9.034099675975929e-06,
+      "loss": 0.3999,
+      "step": 18091
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 9.031528056369902e-06,
+      "loss": 0.4149,
+      "step": 18092
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 9.028956436763875e-06,
+      "loss": 0.5599,
+      "step": 18093
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 9.026384817157846e-06,
+      "loss": 0.5613,
+      "step": 18094
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 9.023813197551818e-06,
+      "loss": 0.7211,
+      "step": 18095
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 9.02124157794579e-06,
+      "loss": 0.4221,
+      "step": 18096
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 9.018669958339762e-06,
+      "loss": 0.432,
+      "step": 18097
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 9.016098338733735e-06,
+      "loss": 0.4715,
+      "step": 18098
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 9.013526719127707e-06,
+      "loss": 0.3889,
+      "step": 18099
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 9.01095509952168e-06,
+      "loss": 0.6176,
+      "step": 18100
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 9.008383479915651e-06,
+      "loss": 0.5164,
+      "step": 18101
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 9.005811860309624e-06,
+      "loss": 0.4935,
+      "step": 18102
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 9.003240240703595e-06,
+      "loss": 0.7084,
+      "step": 18103
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 9.000668621097567e-06,
+      "loss": 0.8235,
+      "step": 18104
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 8.99809700149154e-06,
+      "loss": 0.503,
+      "step": 18105
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 8.995525381885513e-06,
+      "loss": 0.5714,
+      "step": 18106
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 8.992953762279484e-06,
+      "loss": 0.7405,
+      "step": 18107
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 8.990382142673456e-06,
+      "loss": 0.6163,
+      "step": 18108
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 8.987810523067429e-06,
+      "loss": 0.5571,
+      "step": 18109
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 8.9852389034614e-06,
+      "loss": 0.4858,
+      "step": 18110
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 8.982667283855373e-06,
+      "loss": 0.5181,
+      "step": 18111
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 8.980095664249345e-06,
+      "loss": 0.6198,
+      "step": 18112
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 8.977524044643316e-06,
+      "loss": 0.6707,
+      "step": 18113
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 8.974952425037289e-06,
+      "loss": 0.41,
+      "step": 18114
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 8.972380805431262e-06,
+      "loss": 0.9015,
+      "step": 18115
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 8.969809185825234e-06,
+      "loss": 0.8383,
+      "step": 18116
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 8.967237566219205e-06,
+      "loss": 0.548,
+      "step": 18117
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 8.964665946613178e-06,
+      "loss": 0.6559,
+      "step": 18118
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 8.96209432700715e-06,
+      "loss": 0.6132,
+      "step": 18119
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 8.959522707401122e-06,
+      "loss": 0.5393,
+      "step": 18120
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 8.956951087795094e-06,
+      "loss": 0.4454,
+      "step": 18121
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 8.954379468189065e-06,
+      "loss": 0.9003,
+      "step": 18122
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 8.951807848583038e-06,
+      "loss": 0.8815,
+      "step": 18123
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 8.949236228977011e-06,
+      "loss": 0.6385,
+      "step": 18124
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 8.946664609370983e-06,
+      "loss": 0.6455,
+      "step": 18125
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 8.944092989764954e-06,
+      "loss": 0.5503,
+      "step": 18126
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 8.941521370158927e-06,
+      "loss": 0.5036,
+      "step": 18127
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 8.938949750552899e-06,
+      "loss": 0.4764,
+      "step": 18128
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 8.936378130946872e-06,
+      "loss": 0.7735,
+      "step": 18129
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 8.933806511340843e-06,
+      "loss": 0.6448,
+      "step": 18130
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 8.931234891734814e-06,
+      "loss": 0.4996,
+      "step": 18131
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 8.928663272128787e-06,
+      "loss": 0.7017,
+      "step": 18132
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 8.92609165252276e-06,
+      "loss": 0.7591,
+      "step": 18133
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 8.92352003291673e-06,
+      "loss": 0.4726,
+      "step": 18134
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 8.920948413310703e-06,
+      "loss": 0.4819,
+      "step": 18135
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 8.918376793704676e-06,
+      "loss": 0.4496,
+      "step": 18136
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 8.915805174098648e-06,
+      "loss": 0.5148,
+      "step": 18137
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 8.913233554492619e-06,
+      "loss": 0.6589,
+      "step": 18138
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 8.910661934886592e-06,
+      "loss": 0.4742,
+      "step": 18139
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 8.908090315280564e-06,
+      "loss": 0.3606,
+      "step": 18140
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 8.905518695674537e-06,
+      "loss": 0.5177,
+      "step": 18141
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 8.902947076068508e-06,
+      "loss": 0.6947,
+      "step": 18142
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 8.90037545646248e-06,
+      "loss": 0.39,
+      "step": 18143
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 8.897803836856452e-06,
+      "loss": 0.47,
+      "step": 18144
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 8.895232217250426e-06,
+      "loss": 0.5816,
+      "step": 18145
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 8.892660597644397e-06,
+      "loss": 0.5416,
+      "step": 18146
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 8.890088978038368e-06,
+      "loss": 0.6272,
+      "step": 18147
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 8.887517358432341e-06,
+      "loss": 0.5222,
+      "step": 18148
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 8.884945738826313e-06,
+      "loss": 0.5854,
+      "step": 18149
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 8.882374119220286e-06,
+      "loss": 0.565,
+      "step": 18150
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 8.879802499614257e-06,
+      "loss": 0.4911,
+      "step": 18151
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 8.877230880008229e-06,
+      "loss": 0.7606,
+      "step": 18152
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 8.874659260402202e-06,
+      "loss": 0.4666,
+      "step": 18153
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 8.872087640796175e-06,
+      "loss": 0.6654,
+      "step": 18154
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 8.869516021190146e-06,
+      "loss": 0.6626,
+      "step": 18155
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 8.866944401584118e-06,
+      "loss": 0.8716,
+      "step": 18156
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 8.86437278197809e-06,
+      "loss": 0.6397,
+      "step": 18157
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 8.861801162372062e-06,
+      "loss": 0.5874,
+      "step": 18158
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 8.859229542766035e-06,
+      "loss": 0.6194,
+      "step": 18159
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 8.856657923160006e-06,
+      "loss": 0.5627,
+      "step": 18160
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 8.854086303553978e-06,
+      "loss": 0.7022,
+      "step": 18161
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 8.851514683947951e-06,
+      "loss": 0.3362,
+      "step": 18162
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 8.848943064341924e-06,
+      "loss": 0.4212,
+      "step": 18163
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 8.846371444735894e-06,
+      "loss": 0.4657,
+      "step": 18164
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 8.843799825129867e-06,
+      "loss": 0.6086,
+      "step": 18165
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 8.84122820552384e-06,
+      "loss": 0.4837,
+      "step": 18166
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 8.838656585917811e-06,
+      "loss": 0.566,
+      "step": 18167
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 8.836084966311784e-06,
+      "loss": 0.5525,
+      "step": 18168
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 8.833513346705756e-06,
+      "loss": 0.535,
+      "step": 18169
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 8.830941727099727e-06,
+      "loss": 0.4738,
+      "step": 18170
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 8.8283701074937e-06,
+      "loss": 0.8103,
+      "step": 18171
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 8.825798487887673e-06,
+      "loss": 0.535,
+      "step": 18172
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 8.823226868281643e-06,
+      "loss": 0.6484,
+      "step": 18173
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 8.820655248675616e-06,
+      "loss": 0.5427,
+      "step": 18174
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 8.818083629069589e-06,
+      "loss": 0.6453,
+      "step": 18175
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 8.815512009463562e-06,
+      "loss": 0.6855,
+      "step": 18176
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 8.812940389857532e-06,
+      "loss": 0.3927,
+      "step": 18177
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 8.810368770251505e-06,
+      "loss": 0.6796,
+      "step": 18178
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 8.807797150645478e-06,
+      "loss": 0.3555,
+      "step": 18179
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 8.80522553103945e-06,
+      "loss": 0.5903,
+      "step": 18180
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 8.80265391143342e-06,
+      "loss": 0.7214,
+      "step": 18181
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 8.800082291827394e-06,
+      "loss": 0.6665,
+      "step": 18182
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 8.797510672221365e-06,
+      "loss": 0.5887,
+      "step": 18183
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 8.794939052615338e-06,
+      "loss": 0.7762,
+      "step": 18184
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 8.79236743300931e-06,
+      "loss": 0.8141,
+      "step": 18185
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 8.789795813403281e-06,
+      "loss": 0.732,
+      "step": 18186
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 8.787224193797254e-06,
+      "loss": 0.8442,
+      "step": 18187
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 8.784652574191227e-06,
+      "loss": 0.8264,
+      "step": 18188
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 8.782080954585198e-06,
+      "loss": 0.7809,
+      "step": 18189
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 8.77950933497917e-06,
+      "loss": 0.8488,
+      "step": 18190
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 8.776937715373143e-06,
+      "loss": 0.5088,
+      "step": 18191
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 8.774366095767114e-06,
+      "loss": 0.3814,
+      "step": 18192
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 8.771794476161087e-06,
+      "loss": 0.4779,
+      "step": 18193
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 8.769222856555059e-06,
+      "loss": 0.6886,
+      "step": 18194
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 8.76665123694903e-06,
+      "loss": 0.8079,
+      "step": 18195
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 8.764079617343003e-06,
+      "loss": 0.489,
+      "step": 18196
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 8.761507997736976e-06,
+      "loss": 0.6671,
+      "step": 18197
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 8.758936378130948e-06,
+      "loss": 0.7751,
+      "step": 18198
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 8.756364758524919e-06,
+      "loss": 0.33,
+      "step": 18199
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 8.753793138918892e-06,
+      "loss": 0.5618,
+      "step": 18200
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 8.751221519312863e-06,
+      "loss": 0.493,
+      "step": 18201
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 8.748649899706837e-06,
+      "loss": 0.7218,
+      "step": 18202
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 8.746078280100808e-06,
+      "loss": 0.5541,
+      "step": 18203
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 8.74350666049478e-06,
+      "loss": 0.2864,
+      "step": 18204
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 8.740935040888752e-06,
+      "loss": 0.6461,
+      "step": 18205
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 8.738363421282725e-06,
+      "loss": 0.6957,
+      "step": 18206
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 8.735791801676695e-06,
+      "loss": 0.6677,
+      "step": 18207
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 8.733220182070668e-06,
+      "loss": 0.7343,
+      "step": 18208
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 8.730648562464641e-06,
+      "loss": 0.5235,
+      "step": 18209
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 8.728076942858613e-06,
+      "loss": 0.7059,
+      "step": 18210
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 8.725505323252586e-06,
+      "loss": 0.6247,
+      "step": 18211
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 8.722933703646557e-06,
+      "loss": 0.5505,
+      "step": 18212
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 8.720362084040528e-06,
+      "loss": 0.5094,
+      "step": 18213
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 8.717790464434502e-06,
+      "loss": 0.7054,
+      "step": 18214
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 8.715218844828475e-06,
+      "loss": 0.622,
+      "step": 18215
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 8.712647225222444e-06,
+      "loss": 0.7813,
+      "step": 18216
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 8.710075605616417e-06,
+      "loss": 0.48,
+      "step": 18217
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 8.70750398601039e-06,
+      "loss": 0.401,
+      "step": 18218
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 8.704932366404362e-06,
+      "loss": 0.8237,
+      "step": 18219
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 8.702360746798333e-06,
+      "loss": 0.5867,
+      "step": 18220
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 8.699789127192306e-06,
+      "loss": 0.4775,
+      "step": 18221
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 8.697217507586278e-06,
+      "loss": 0.4864,
+      "step": 18222
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 8.69464588798025e-06,
+      "loss": 0.5116,
+      "step": 18223
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 8.692074268374222e-06,
+      "loss": 0.6167,
+      "step": 18224
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 8.689502648768194e-06,
+      "loss": 0.5268,
+      "step": 18225
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 8.686931029162167e-06,
+      "loss": 0.5607,
+      "step": 18226
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 8.68435940955614e-06,
+      "loss": 0.6129,
+      "step": 18227
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 8.681787789950111e-06,
+      "loss": 0.449,
+      "step": 18228
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 8.679216170344082e-06,
+      "loss": 0.516,
+      "step": 18229
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 8.676644550738055e-06,
+      "loss": 0.4753,
+      "step": 18230
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 8.674072931132027e-06,
+      "loss": 0.6637,
+      "step": 18231
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 8.671501311526e-06,
+      "loss": 0.6359,
+      "step": 18232
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 8.668929691919971e-06,
+      "loss": 0.9211,
+      "step": 18233
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 8.666358072313943e-06,
+      "loss": 0.7189,
+      "step": 18234
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 8.663786452707916e-06,
+      "loss": 0.8107,
+      "step": 18235
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 8.661214833101889e-06,
+      "loss": 0.3465,
+      "step": 18236
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 8.65864321349586e-06,
+      "loss": 0.5157,
+      "step": 18237
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 8.656071593889832e-06,
+      "loss": 0.5845,
+      "step": 18238
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 8.653499974283805e-06,
+      "loss": 0.4515,
+      "step": 18239
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 8.650928354677776e-06,
+      "loss": 0.4581,
+      "step": 18240
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 8.648356735071749e-06,
+      "loss": 0.6002,
+      "step": 18241
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 8.64578511546572e-06,
+      "loss": 0.7453,
+      "step": 18242
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 8.643213495859692e-06,
+      "loss": 0.3332,
+      "step": 18243
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 8.640641876253665e-06,
+      "loss": 0.7068,
+      "step": 18244
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 8.638070256647638e-06,
+      "loss": 0.5504,
+      "step": 18245
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 8.635498637041608e-06,
+      "loss": 0.551,
+      "step": 18246
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 8.63292701743558e-06,
+      "loss": 1.0151,
+      "step": 18247
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 8.630355397829554e-06,
+      "loss": 0.8472,
+      "step": 18248
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 8.627783778223525e-06,
+      "loss": 0.6375,
+      "step": 18249
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 8.625212158617498e-06,
+      "loss": 0.509,
+      "step": 18250
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 8.62264053901147e-06,
+      "loss": 0.7342,
+      "step": 18251
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 8.620068919405443e-06,
+      "loss": 0.4251,
+      "step": 18252
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 8.617497299799414e-06,
+      "loss": 0.608,
+      "step": 18253
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 8.614925680193387e-06,
+      "loss": 0.5256,
+      "step": 18254
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 8.612354060587359e-06,
+      "loss": 0.5205,
+      "step": 18255
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 8.60978244098133e-06,
+      "loss": 0.6234,
+      "step": 18256
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 8.607210821375303e-06,
+      "loss": 0.5326,
+      "step": 18257
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 8.604639201769276e-06,
+      "loss": 0.4659,
+      "step": 18258
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 8.602067582163246e-06,
+      "loss": 0.2958,
+      "step": 18259
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 8.599495962557219e-06,
+      "loss": 0.6718,
+      "step": 18260
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 8.596924342951192e-06,
+      "loss": 0.4009,
+      "step": 18261
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 8.594352723345163e-06,
+      "loss": 0.7233,
+      "step": 18262
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 8.591781103739135e-06,
+      "loss": 0.7143,
+      "step": 18263
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 8.589209484133108e-06,
+      "loss": 0.412,
+      "step": 18264
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 8.58663786452708e-06,
+      "loss": 0.5273,
+      "step": 18265
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 8.584066244921052e-06,
+      "loss": 0.6483,
+      "step": 18266
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 8.581494625315024e-06,
+      "loss": 0.6,
+      "step": 18267
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 8.578923005708995e-06,
+      "loss": 0.2653,
+      "step": 18268
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 8.576351386102968e-06,
+      "loss": 0.4122,
+      "step": 18269
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 8.573779766496941e-06,
+      "loss": 0.5121,
+      "step": 18270
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 8.571208146890913e-06,
+      "loss": 0.5161,
+      "step": 18271
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 8.568636527284884e-06,
+      "loss": 0.6144,
+      "step": 18272
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 8.566064907678857e-06,
+      "loss": 0.6585,
+      "step": 18273
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 8.563493288072828e-06,
+      "loss": 0.6863,
+      "step": 18274
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 8.560921668466801e-06,
+      "loss": 0.5573,
+      "step": 18275
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 8.558350048860773e-06,
+      "loss": 0.371,
+      "step": 18276
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 8.555778429254744e-06,
+      "loss": 0.6811,
+      "step": 18277
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 8.553206809648717e-06,
+      "loss": 0.6194,
+      "step": 18278
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 8.55063519004269e-06,
+      "loss": 0.367,
+      "step": 18279
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 8.548063570436662e-06,
+      "loss": 0.887,
+      "step": 18280
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 8.545491950830633e-06,
+      "loss": 0.8611,
+      "step": 18281
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 8.542920331224606e-06,
+      "loss": 0.697,
+      "step": 18282
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 8.540348711618578e-06,
+      "loss": 0.5942,
+      "step": 18283
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 8.53777709201255e-06,
+      "loss": 0.5222,
+      "step": 18284
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 8.535205472406522e-06,
+      "loss": 0.4407,
+      "step": 18285
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 8.532633852800493e-06,
+      "loss": 0.3639,
+      "step": 18286
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 8.530062233194466e-06,
+      "loss": 0.5931,
+      "step": 18287
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 8.52749061358844e-06,
+      "loss": 0.6679,
+      "step": 18288
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 8.52491899398241e-06,
+      "loss": 0.7269,
+      "step": 18289
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 8.522347374376382e-06,
+      "loss": 0.5906,
+      "step": 18290
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 8.519775754770355e-06,
+      "loss": 0.8486,
+      "step": 18291
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 8.517204135164327e-06,
+      "loss": 0.4842,
+      "step": 18292
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 8.5146325155583e-06,
+      "loss": 0.6606,
+      "step": 18293
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 8.512060895952271e-06,
+      "loss": 0.6946,
+      "step": 18294
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 8.509489276346243e-06,
+      "loss": 0.5521,
+      "step": 18295
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 8.506917656740216e-06,
+      "loss": 0.5601,
+      "step": 18296
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 8.504346037134189e-06,
+      "loss": 0.5071,
+      "step": 18297
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 8.501774417528158e-06,
+      "loss": 0.6291,
+      "step": 18298
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 8.499202797922131e-06,
+      "loss": 0.3885,
+      "step": 18299
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 8.496631178316105e-06,
+      "loss": 0.5962,
+      "step": 18300
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 8.494059558710076e-06,
+      "loss": 0.8767,
+      "step": 18301
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 8.491487939104047e-06,
+      "loss": 0.6738,
+      "step": 18302
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 8.48891631949802e-06,
+      "loss": 0.4192,
+      "step": 18303
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 8.486344699891992e-06,
+      "loss": 0.3216,
+      "step": 18304
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 8.483773080285965e-06,
+      "loss": 0.584,
+      "step": 18305
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 8.481201460679936e-06,
+      "loss": 0.7351,
+      "step": 18306
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 8.478629841073908e-06,
+      "loss": 0.503,
+      "step": 18307
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 8.47605822146788e-06,
+      "loss": 0.6779,
+      "step": 18308
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 8.473486601861854e-06,
+      "loss": 0.6583,
+      "step": 18309
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 8.470914982255825e-06,
+      "loss": 0.5624,
+      "step": 18310
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 8.468343362649797e-06,
+      "loss": 0.881,
+      "step": 18311
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 8.46577174304377e-06,
+      "loss": 0.5826,
+      "step": 18312
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 8.463200123437741e-06,
+      "loss": 0.8184,
+      "step": 18313
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 8.460628503831714e-06,
+      "loss": 1.1899,
+      "step": 18314
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 8.458056884225685e-06,
+      "loss": 0.6107,
+      "step": 18315
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 8.455485264619657e-06,
+      "loss": 0.6909,
+      "step": 18316
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 8.45291364501363e-06,
+      "loss": 0.7828,
+      "step": 18317
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 8.450342025407603e-06,
+      "loss": 0.7694,
+      "step": 18318
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 8.447770405801574e-06,
+      "loss": 0.6472,
+      "step": 18319
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 8.445198786195546e-06,
+      "loss": 0.7121,
+      "step": 18320
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 8.442627166589519e-06,
+      "loss": 0.6935,
+      "step": 18321
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 8.44005554698349e-06,
+      "loss": 0.8275,
+      "step": 18322
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 8.437483927377463e-06,
+      "loss": 0.6759,
+      "step": 18323
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 8.434912307771435e-06,
+      "loss": 0.4277,
+      "step": 18324
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 8.432340688165406e-06,
+      "loss": 0.4076,
+      "step": 18325
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 8.429769068559379e-06,
+      "loss": 0.8483,
+      "step": 18326
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 8.427197448953352e-06,
+      "loss": 0.7885,
+      "step": 18327
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 8.424625829347324e-06,
+      "loss": 0.714,
+      "step": 18328
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 8.422054209741295e-06,
+      "loss": 0.4961,
+      "step": 18329
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 8.419482590135268e-06,
+      "loss": 0.6204,
+      "step": 18330
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 8.416910970529241e-06,
+      "loss": 0.768,
+      "step": 18331
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 8.414339350923212e-06,
+      "loss": 0.5882,
+      "step": 18332
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 8.411767731317184e-06,
+      "loss": 0.4837,
+      "step": 18333
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 8.409196111711157e-06,
+      "loss": 0.6352,
+      "step": 18334
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 8.406624492105128e-06,
+      "loss": 0.5035,
+      "step": 18335
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 8.404052872499101e-06,
+      "loss": 0.5058,
+      "step": 18336
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 8.401481252893073e-06,
+      "loss": 0.4873,
+      "step": 18337
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 8.398909633287044e-06,
+      "loss": 0.6384,
+      "step": 18338
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 8.396338013681017e-06,
+      "loss": 0.4183,
+      "step": 18339
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 8.39376639407499e-06,
+      "loss": 0.5345,
+      "step": 18340
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 8.39119477446896e-06,
+      "loss": 0.7237,
+      "step": 18341
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 8.388623154862933e-06,
+      "loss": 0.6304,
+      "step": 18342
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 8.386051535256906e-06,
+      "loss": 0.7431,
+      "step": 18343
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 8.383479915650877e-06,
+      "loss": 0.5826,
+      "step": 18344
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 8.380908296044849e-06,
+      "loss": 0.531,
+      "step": 18345
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 8.378336676438822e-06,
+      "loss": 0.5985,
+      "step": 18346
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 8.375765056832793e-06,
+      "loss": 0.5519,
+      "step": 18347
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 8.373193437226766e-06,
+      "loss": 0.4227,
+      "step": 18348
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 8.370621817620738e-06,
+      "loss": 0.7186,
+      "step": 18349
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 8.368050198014709e-06,
+      "loss": 0.711,
+      "step": 18350
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 8.365478578408682e-06,
+      "loss": 0.6092,
+      "step": 18351
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 8.362906958802655e-06,
+      "loss": 0.5459,
+      "step": 18352
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 8.360335339196627e-06,
+      "loss": 0.6316,
+      "step": 18353
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 8.357763719590598e-06,
+      "loss": 0.3827,
+      "step": 18354
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 8.355192099984571e-06,
+      "loss": 0.464,
+      "step": 18355
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 8.352620480378542e-06,
+      "loss": 0.8122,
+      "step": 18356
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 8.350048860772516e-06,
+      "loss": 0.4388,
+      "step": 18357
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 8.347477241166487e-06,
+      "loss": 0.6065,
+      "step": 18358
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 8.344905621560458e-06,
+      "loss": 0.5945,
+      "step": 18359
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 8.342334001954431e-06,
+      "loss": 0.6726,
+      "step": 18360
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 8.339762382348404e-06,
+      "loss": 0.7116,
+      "step": 18361
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 8.337190762742376e-06,
+      "loss": 0.5717,
+      "step": 18362
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 8.334619143136347e-06,
+      "loss": 0.4104,
+      "step": 18363
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 8.33204752353032e-06,
+      "loss": 0.5878,
+      "step": 18364
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 8.329475903924292e-06,
+      "loss": 0.6015,
+      "step": 18365
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 8.326904284318265e-06,
+      "loss": 0.3354,
+      "step": 18366
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 8.324332664712236e-06,
+      "loss": 0.8282,
+      "step": 18367
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 8.321761045106207e-06,
+      "loss": 0.5238,
+      "step": 18368
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 8.31918942550018e-06,
+      "loss": 0.6772,
+      "step": 18369
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 8.316617805894154e-06,
+      "loss": 0.6179,
+      "step": 18370
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 8.314046186288123e-06,
+      "loss": 0.7847,
+      "step": 18371
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 8.311474566682096e-06,
+      "loss": 0.5468,
+      "step": 18372
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 8.30890294707607e-06,
+      "loss": 0.468,
+      "step": 18373
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 8.30633132747004e-06,
+      "loss": 0.5755,
+      "step": 18374
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 8.303759707864014e-06,
+      "loss": 0.7967,
+      "step": 18375
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 8.301188088257985e-06,
+      "loss": 0.5747,
+      "step": 18376
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 8.298616468651957e-06,
+      "loss": 0.7388,
+      "step": 18377
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 8.29604484904593e-06,
+      "loss": 0.3442,
+      "step": 18378
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 8.293473229439903e-06,
+      "loss": 0.6186,
+      "step": 18379
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 8.290901609833873e-06,
+      "loss": 0.2048,
+      "step": 18380
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 8.288329990227846e-06,
+      "loss": 0.5701,
+      "step": 18381
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 8.285758370621819e-06,
+      "loss": 0.4812,
+      "step": 18382
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 8.28318675101579e-06,
+      "loss": 0.5216,
+      "step": 18383
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 8.280615131409761e-06,
+      "loss": 0.6016,
+      "step": 18384
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 8.278043511803734e-06,
+      "loss": 0.6117,
+      "step": 18385
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 8.275471892197706e-06,
+      "loss": 0.5197,
+      "step": 18386
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 8.272900272591679e-06,
+      "loss": 0.671,
+      "step": 18387
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 8.27032865298565e-06,
+      "loss": 0.8287,
+      "step": 18388
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 8.267757033379622e-06,
+      "loss": 0.5815,
+      "step": 18389
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 8.265185413773595e-06,
+      "loss": 0.7006,
+      "step": 18390
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 8.262613794167568e-06,
+      "loss": 0.4561,
+      "step": 18391
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 8.26004217456154e-06,
+      "loss": 0.4656,
+      "step": 18392
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 8.25747055495551e-06,
+      "loss": 0.3983,
+      "step": 18393
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 8.254898935349484e-06,
+      "loss": 0.5442,
+      "step": 18394
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 8.252327315743455e-06,
+      "loss": 0.4815,
+      "step": 18395
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 8.249755696137428e-06,
+      "loss": 0.5449,
+      "step": 18396
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 8.2471840765314e-06,
+      "loss": 0.5157,
+      "step": 18397
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 8.244612456925371e-06,
+      "loss": 0.489,
+      "step": 18398
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 8.242040837319344e-06,
+      "loss": 0.4026,
+      "step": 18399
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 8.239469217713317e-06,
+      "loss": 0.672,
+      "step": 18400
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 8.236897598107288e-06,
+      "loss": 0.4096,
+      "step": 18401
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 8.23432597850126e-06,
+      "loss": 0.6282,
+      "step": 18402
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 8.231754358895233e-06,
+      "loss": 0.5253,
+      "step": 18403
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 8.229182739289206e-06,
+      "loss": 0.7306,
+      "step": 18404
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 8.226611119683177e-06,
+      "loss": 0.6458,
+      "step": 18405
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 8.224039500077149e-06,
+      "loss": 0.5634,
+      "step": 18406
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 8.221467880471122e-06,
+      "loss": 0.6655,
+      "step": 18407
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 8.218896260865093e-06,
+      "loss": 0.6763,
+      "step": 18408
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 8.216324641259066e-06,
+      "loss": 0.5357,
+      "step": 18409
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 8.213753021653038e-06,
+      "loss": 0.5636,
+      "step": 18410
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 8.211181402047009e-06,
+      "loss": 0.5816,
+      "step": 18411
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 8.208609782440982e-06,
+      "loss": 0.5535,
+      "step": 18412
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 8.206038162834955e-06,
+      "loss": 0.9436,
+      "step": 18413
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 8.203466543228927e-06,
+      "loss": 0.5937,
+      "step": 18414
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 8.200894923622898e-06,
+      "loss": 0.5466,
+      "step": 18415
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 8.198323304016871e-06,
+      "loss": 0.6667,
+      "step": 18416
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 8.195751684410842e-06,
+      "loss": 0.6241,
+      "step": 18417
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 8.193180064804815e-06,
+      "loss": 0.583,
+      "step": 18418
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 8.190608445198787e-06,
+      "loss": 0.5816,
+      "step": 18419
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 8.188036825592758e-06,
+      "loss": 0.5729,
+      "step": 18420
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 8.185465205986731e-06,
+      "loss": 0.4562,
+      "step": 18421
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 8.182893586380704e-06,
+      "loss": 0.3703,
+      "step": 18422
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 8.180321966774674e-06,
+      "loss": 0.6743,
+      "step": 18423
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 8.177750347168647e-06,
+      "loss": 0.5761,
+      "step": 18424
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 8.17517872756262e-06,
+      "loss": 0.7922,
+      "step": 18425
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 8.172607107956592e-06,
+      "loss": 0.4064,
+      "step": 18426
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 8.170035488350563e-06,
+      "loss": 0.4949,
+      "step": 18427
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 8.167463868744536e-06,
+      "loss": 0.6043,
+      "step": 18428
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 8.164892249138507e-06,
+      "loss": 0.5785,
+      "step": 18429
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 8.16232062953248e-06,
+      "loss": 0.6809,
+      "step": 18430
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 8.159749009926452e-06,
+      "loss": 0.5295,
+      "step": 18431
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 8.157177390320423e-06,
+      "loss": 0.5359,
+      "step": 18432
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 8.154605770714396e-06,
+      "loss": 0.8438,
+      "step": 18433
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 8.15203415110837e-06,
+      "loss": 0.7804,
+      "step": 18434
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 8.14946253150234e-06,
+      "loss": 0.4772,
+      "step": 18435
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 8.146890911896312e-06,
+      "loss": 0.5268,
+      "step": 18436
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 8.144319292290285e-06,
+      "loss": 0.6333,
+      "step": 18437
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 8.141747672684257e-06,
+      "loss": 0.3789,
+      "step": 18438
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 8.13917605307823e-06,
+      "loss": 0.476,
+      "step": 18439
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 8.136604433472201e-06,
+      "loss": 0.527,
+      "step": 18440
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 8.134032813866172e-06,
+      "loss": 0.4953,
+      "step": 18441
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 8.131461194260145e-06,
+      "loss": 0.5793,
+      "step": 18442
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 8.128889574654119e-06,
+      "loss": 1.0185,
+      "step": 18443
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 8.12631795504809e-06,
+      "loss": 0.3666,
+      "step": 18444
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 8.123746335442061e-06,
+      "loss": 0.8708,
+      "step": 18445
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 8.121174715836034e-06,
+      "loss": 0.4986,
+      "step": 18446
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 8.118603096230006e-06,
+      "loss": 0.5638,
+      "step": 18447
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 8.116031476623979e-06,
+      "loss": 0.4914,
+      "step": 18448
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 8.11345985701795e-06,
+      "loss": 0.6259,
+      "step": 18449
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 8.110888237411922e-06,
+      "loss": 0.4582,
+      "step": 18450
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 8.108316617805895e-06,
+      "loss": 0.4603,
+      "step": 18451
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 8.105744998199868e-06,
+      "loss": 0.6437,
+      "step": 18452
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 8.103173378593837e-06,
+      "loss": 0.6465,
+      "step": 18453
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 8.10060175898781e-06,
+      "loss": 0.8002,
+      "step": 18454
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 8.098030139381784e-06,
+      "loss": 0.6578,
+      "step": 18455
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 8.095458519775755e-06,
+      "loss": 0.6069,
+      "step": 18456
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 8.092886900169728e-06,
+      "loss": 0.7071,
+      "step": 18457
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 8.0903152805637e-06,
+      "loss": 0.35,
+      "step": 18458
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 8.08774366095767e-06,
+      "loss": 0.808,
+      "step": 18459
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 8.085172041351644e-06,
+      "loss": 0.8209,
+      "step": 18460
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 8.082600421745617e-06,
+      "loss": 0.3449,
+      "step": 18461
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 8.080028802139587e-06,
+      "loss": 0.6759,
+      "step": 18462
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 8.07745718253356e-06,
+      "loss": 0.6196,
+      "step": 18463
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 8.074885562927533e-06,
+      "loss": 0.535,
+      "step": 18464
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 8.072313943321504e-06,
+      "loss": 0.503,
+      "step": 18465
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 8.069742323715476e-06,
+      "loss": 0.6878,
+      "step": 18466
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 8.067170704109449e-06,
+      "loss": 0.6208,
+      "step": 18467
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 8.06459908450342e-06,
+      "loss": 0.3039,
+      "step": 18468
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 8.062027464897393e-06,
+      "loss": 0.5051,
+      "step": 18469
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 8.059455845291364e-06,
+      "loss": 0.7953,
+      "step": 18470
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 8.056884225685336e-06,
+      "loss": 0.2933,
+      "step": 18471
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 8.054312606079309e-06,
+      "loss": 0.6996,
+      "step": 18472
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 8.051740986473282e-06,
+      "loss": 0.5675,
+      "step": 18473
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 8.049169366867253e-06,
+      "loss": 0.495,
+      "step": 18474
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 8.046597747261225e-06,
+      "loss": 0.7405,
+      "step": 18475
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 8.044026127655198e-06,
+      "loss": 0.2907,
+      "step": 18476
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 8.04145450804917e-06,
+      "loss": 0.7552,
+      "step": 18477
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 8.038882888443142e-06,
+      "loss": 0.3674,
+      "step": 18478
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 8.036311268837114e-06,
+      "loss": 0.5499,
+      "step": 18479
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 8.033739649231087e-06,
+      "loss": 0.7022,
+      "step": 18480
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 8.031168029625058e-06,
+      "loss": 0.5633,
+      "step": 18481
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 8.028596410019031e-06,
+      "loss": 0.876,
+      "step": 18482
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 8.026024790413003e-06,
+      "loss": 0.3539,
+      "step": 18483
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 8.023453170806974e-06,
+      "loss": 0.7029,
+      "step": 18484
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 8.020881551200947e-06,
+      "loss": 0.5279,
+      "step": 18485
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 8.01830993159492e-06,
+      "loss": 0.6798,
+      "step": 18486
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 8.015738311988891e-06,
+      "loss": 0.3054,
+      "step": 18487
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 8.013166692382863e-06,
+      "loss": 0.438,
+      "step": 18488
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 8.010595072776836e-06,
+      "loss": 0.5467,
+      "step": 18489
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 8.008023453170807e-06,
+      "loss": 0.4265,
+      "step": 18490
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 8.00545183356478e-06,
+      "loss": 0.6601,
+      "step": 18491
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 8.002880213958752e-06,
+      "loss": 0.5436,
+      "step": 18492
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 8.000308594352723e-06,
+      "loss": 0.7484,
+      "step": 18493
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 7.997736974746696e-06,
+      "loss": 0.5997,
+      "step": 18494
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 7.99516535514067e-06,
+      "loss": 0.3936,
+      "step": 18495
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 7.992593735534639e-06,
+      "loss": 0.677,
+      "step": 18496
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 7.990022115928612e-06,
+      "loss": 0.4159,
+      "step": 18497
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 7.987450496322585e-06,
+      "loss": 0.4268,
+      "step": 18498
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 7.984878876716556e-06,
+      "loss": 0.6168,
+      "step": 18499
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 7.98230725711053e-06,
+      "loss": 0.6509,
+      "step": 18500
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 7.979735637504501e-06,
+      "loss": 0.492,
+      "step": 18501
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 7.977164017898472e-06,
+      "loss": 0.5554,
+      "step": 18502
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 7.974592398292445e-06,
+      "loss": 0.3462,
+      "step": 18503
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 7.972020778686418e-06,
+      "loss": 0.5703,
+      "step": 18504
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 7.969449159080388e-06,
+      "loss": 0.4674,
+      "step": 18505
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 7.966877539474361e-06,
+      "loss": 0.5185,
+      "step": 18506
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 7.964305919868334e-06,
+      "loss": 0.5779,
+      "step": 18507
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 7.961734300262306e-06,
+      "loss": 0.2218,
+      "step": 18508
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 7.959162680656277e-06,
+      "loss": 0.6411,
+      "step": 18509
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 7.95659106105025e-06,
+      "loss": 0.5632,
+      "step": 18510
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 7.954019441444221e-06,
+      "loss": 0.4398,
+      "step": 18511
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 7.951447821838195e-06,
+      "loss": 0.6731,
+      "step": 18512
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 7.948876202232166e-06,
+      "loss": 0.6496,
+      "step": 18513
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 7.946304582626137e-06,
+      "loss": 0.5829,
+      "step": 18514
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 7.94373296302011e-06,
+      "loss": 0.5671,
+      "step": 18515
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 7.941161343414083e-06,
+      "loss": 0.4626,
+      "step": 18516
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 7.938589723808055e-06,
+      "loss": 0.4394,
+      "step": 18517
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 7.936018104202026e-06,
+      "loss": 0.5212,
+      "step": 18518
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 7.933446484596e-06,
+      "loss": 0.4483,
+      "step": 18519
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 7.93087486498997e-06,
+      "loss": 0.8985,
+      "step": 18520
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 7.928303245383944e-06,
+      "loss": 0.4151,
+      "step": 18521
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 7.925731625777915e-06,
+      "loss": 0.5663,
+      "step": 18522
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 7.923160006171886e-06,
+      "loss": 0.6353,
+      "step": 18523
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 7.92058838656586e-06,
+      "loss": 0.5684,
+      "step": 18524
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 7.918016766959833e-06,
+      "loss": 0.5321,
+      "step": 18525
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 7.915445147353804e-06,
+      "loss": 0.4501,
+      "step": 18526
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 7.912873527747775e-06,
+      "loss": 0.7268,
+      "step": 18527
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 7.910301908141748e-06,
+      "loss": 0.4544,
+      "step": 18528
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 7.90773028853572e-06,
+      "loss": 0.5652,
+      "step": 18529
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 7.905158668929693e-06,
+      "loss": 0.7467,
+      "step": 18530
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 7.902587049323664e-06,
+      "loss": 0.5973,
+      "step": 18531
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 7.900015429717636e-06,
+      "loss": 0.4142,
+      "step": 18532
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 7.897443810111609e-06,
+      "loss": 0.5576,
+      "step": 18533
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 7.894872190505582e-06,
+      "loss": 0.5785,
+      "step": 18534
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 7.892300570899551e-06,
+      "loss": 0.3964,
+      "step": 18535
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 7.889728951293525e-06,
+      "loss": 0.784,
+      "step": 18536
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 7.887157331687498e-06,
+      "loss": 0.6084,
+      "step": 18537
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 7.884585712081469e-06,
+      "loss": 0.3194,
+      "step": 18538
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 7.882014092475442e-06,
+      "loss": 0.6352,
+      "step": 18539
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 7.879442472869413e-06,
+      "loss": 0.4202,
+      "step": 18540
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 7.876870853263385e-06,
+      "loss": 0.3912,
+      "step": 18541
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 7.874299233657358e-06,
+      "loss": 0.6824,
+      "step": 18542
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 7.871727614051331e-06,
+      "loss": 0.6736,
+      "step": 18543
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 7.8691559944453e-06,
+      "loss": 0.5387,
+      "step": 18544
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 7.866584374839274e-06,
+      "loss": 0.6782,
+      "step": 18545
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 7.864012755233247e-06,
+      "loss": 0.621,
+      "step": 18546
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 7.861441135627218e-06,
+      "loss": 0.4145,
+      "step": 18547
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 7.85886951602119e-06,
+      "loss": 0.5138,
+      "step": 18548
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 7.856297896415163e-06,
+      "loss": 0.8414,
+      "step": 18549
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 7.853726276809134e-06,
+      "loss": 0.3997,
+      "step": 18550
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 7.851154657203107e-06,
+      "loss": 0.6431,
+      "step": 18551
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 7.848583037597078e-06,
+      "loss": 0.5263,
+      "step": 18552
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 7.846011417991052e-06,
+      "loss": 0.6357,
+      "step": 18553
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 7.843439798385023e-06,
+      "loss": 0.8121,
+      "step": 18554
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 7.840868178778996e-06,
+      "loss": 0.7199,
+      "step": 18555
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 7.838296559172967e-06,
+      "loss": 0.6055,
+      "step": 18556
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 7.835724939566939e-06,
+      "loss": 0.4945,
+      "step": 18557
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 7.833153319960912e-06,
+      "loss": 0.51,
+      "step": 18558
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 7.830581700354885e-06,
+      "loss": 0.4523,
+      "step": 18559
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 7.828010080748856e-06,
+      "loss": 0.4718,
+      "step": 18560
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 7.825438461142828e-06,
+      "loss": 0.4838,
+      "step": 18561
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 7.8228668415368e-06,
+      "loss": 0.7372,
+      "step": 18562
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 7.820295221930772e-06,
+      "loss": 0.5277,
+      "step": 18563
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 7.817723602324745e-06,
+      "loss": 0.2636,
+      "step": 18564
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 7.815151982718717e-06,
+      "loss": 0.8894,
+      "step": 18565
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 7.812580363112688e-06,
+      "loss": 0.3432,
+      "step": 18566
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 7.810008743506661e-06,
+      "loss": 0.5315,
+      "step": 18567
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 7.807437123900634e-06,
+      "loss": 0.6522,
+      "step": 18568
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 7.804865504294606e-06,
+      "loss": 0.5375,
+      "step": 18569
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 7.802293884688577e-06,
+      "loss": 0.6729,
+      "step": 18570
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 7.79972226508255e-06,
+      "loss": 0.4866,
+      "step": 18571
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 7.797150645476521e-06,
+      "loss": 0.2771,
+      "step": 18572
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 7.794579025870494e-06,
+      "loss": 0.4866,
+      "step": 18573
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 7.792007406264466e-06,
+      "loss": 0.5424,
+      "step": 18574
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 7.789435786658437e-06,
+      "loss": 0.3277,
+      "step": 18575
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 7.78686416705241e-06,
+      "loss": 0.7967,
+      "step": 18576
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 7.784292547446383e-06,
+      "loss": 0.5155,
+      "step": 18577
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 7.781720927840353e-06,
+      "loss": 0.3173,
+      "step": 18578
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 7.779149308234326e-06,
+      "loss": 0.5764,
+      "step": 18579
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 7.776577688628299e-06,
+      "loss": 0.6287,
+      "step": 18580
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 7.77400606902227e-06,
+      "loss": 0.5043,
+      "step": 18581
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 7.771434449416244e-06,
+      "loss": 0.6024,
+      "step": 18582
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 7.768862829810215e-06,
+      "loss": 0.7362,
+      "step": 18583
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 7.766291210204186e-06,
+      "loss": 0.7786,
+      "step": 18584
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 7.76371959059816e-06,
+      "loss": 0.5102,
+      "step": 18585
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 7.761147970992133e-06,
+      "loss": 0.8088,
+      "step": 18586
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 7.758576351386102e-06,
+      "loss": 0.4485,
+      "step": 18587
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 7.756004731780075e-06,
+      "loss": 0.3761,
+      "step": 18588
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 7.753433112174048e-06,
+      "loss": 0.3452,
+      "step": 18589
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 7.75086149256802e-06,
+      "loss": 0.5155,
+      "step": 18590
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 7.748289872961991e-06,
+      "loss": 0.7293,
+      "step": 18591
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 7.745718253355964e-06,
+      "loss": 0.4233,
+      "step": 18592
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 7.743146633749936e-06,
+      "loss": 0.6113,
+      "step": 18593
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 7.740575014143909e-06,
+      "loss": 0.4257,
+      "step": 18594
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 7.73800339453788e-06,
+      "loss": 1.0675,
+      "step": 18595
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 7.735431774931851e-06,
+      "loss": 0.5713,
+      "step": 18596
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 7.732860155325824e-06,
+      "loss": 0.5725,
+      "step": 18597
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 7.730288535719798e-06,
+      "loss": 0.732,
+      "step": 18598
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 7.727716916113769e-06,
+      "loss": 0.4892,
+      "step": 18599
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 7.72514529650774e-06,
+      "loss": 0.2725,
+      "step": 18600
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 7.722573676901713e-06,
+      "loss": 0.637,
+      "step": 18601
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 7.720002057295685e-06,
+      "loss": 0.4111,
+      "step": 18602
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 7.717430437689658e-06,
+      "loss": 0.3783,
+      "step": 18603
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 7.71485881808363e-06,
+      "loss": 0.4122,
+      "step": 18604
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 7.7122871984776e-06,
+      "loss": 0.5199,
+      "step": 18605
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 7.709715578871574e-06,
+      "loss": 0.8019,
+      "step": 18606
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 7.707143959265547e-06,
+      "loss": 0.6088,
+      "step": 18607
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 7.704572339659518e-06,
+      "loss": 0.8463,
+      "step": 18608
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 7.70200072005349e-06,
+      "loss": 0.8302,
+      "step": 18609
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 7.699429100447463e-06,
+      "loss": 0.5903,
+      "step": 18610
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 7.696857480841434e-06,
+      "loss": 0.437,
+      "step": 18611
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 7.694285861235407e-06,
+      "loss": 0.5946,
+      "step": 18612
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 7.691714241629378e-06,
+      "loss": 0.5991,
+      "step": 18613
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 7.68914262202335e-06,
+      "loss": 0.3141,
+      "step": 18614
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 7.686571002417323e-06,
+      "loss": 0.7904,
+      "step": 18615
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 7.683999382811296e-06,
+      "loss": 0.4798,
+      "step": 18616
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 7.681427763205266e-06,
+      "loss": 0.8748,
+      "step": 18617
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 7.678856143599239e-06,
+      "loss": 0.6802,
+      "step": 18618
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 7.676284523993212e-06,
+      "loss": 0.3949,
+      "step": 18619
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 7.673712904387183e-06,
+      "loss": 0.2196,
+      "step": 18620
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 7.671141284781156e-06,
+      "loss": 0.4283,
+      "step": 18621
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 7.668569665175128e-06,
+      "loss": 1.0895,
+      "step": 18622
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 7.665998045569099e-06,
+      "loss": 0.6402,
+      "step": 18623
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 7.663426425963072e-06,
+      "loss": 0.4856,
+      "step": 18624
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 7.660854806357045e-06,
+      "loss": 0.5714,
+      "step": 18625
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 7.658283186751015e-06,
+      "loss": 0.3655,
+      "step": 18626
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 7.655711567144988e-06,
+      "loss": 0.5705,
+      "step": 18627
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 7.653139947538961e-06,
+      "loss": 0.39,
+      "step": 18628
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 7.650568327932934e-06,
+      "loss": 0.5945,
+      "step": 18629
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 7.647996708326904e-06,
+      "loss": 0.8148,
+      "step": 18630
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 7.645425088720877e-06,
+      "loss": 0.8991,
+      "step": 18631
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 7.64285346911485e-06,
+      "loss": 0.6324,
+      "step": 18632
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 7.640281849508821e-06,
+      "loss": 0.9081,
+      "step": 18633
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 7.637710229902793e-06,
+      "loss": 0.5088,
+      "step": 18634
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 7.635138610296766e-06,
+      "loss": 0.3553,
+      "step": 18635
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 7.632566990690737e-06,
+      "loss": 0.5509,
+      "step": 18636
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 7.62999537108471e-06,
+      "loss": 0.4767,
+      "step": 18637
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 7.627423751478682e-06,
+      "loss": 0.6321,
+      "step": 18638
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 7.624852131872654e-06,
+      "loss": 0.6762,
+      "step": 18639
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 7.622280512266626e-06,
+      "loss": 0.6953,
+      "step": 18640
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 7.619708892660599e-06,
+      "loss": 0.5098,
+      "step": 18641
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 7.6171372730545696e-06,
+      "loss": 0.655,
+      "step": 18642
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 7.614565653448543e-06,
+      "loss": 0.6852,
+      "step": 18643
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 7.611994033842515e-06,
+      "loss": 0.624,
+      "step": 18644
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 7.609422414236486e-06,
+      "loss": 0.6544,
+      "step": 18645
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 7.6068507946304585e-06,
+      "loss": 0.7758,
+      "step": 18646
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 7.6042791750244315e-06,
+      "loss": 0.6409,
+      "step": 18647
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 7.601707555418403e-06,
+      "loss": 0.3989,
+      "step": 18648
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 7.599135935812375e-06,
+      "loss": 0.4178,
+      "step": 18649
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 7.596564316206347e-06,
+      "loss": 0.783,
+      "step": 18650
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 7.593992696600319e-06,
+      "loss": 0.5355,
+      "step": 18651
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 7.591421076994292e-06,
+      "loss": 1.1054,
+      "step": 18652
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 7.588849457388264e-06,
+      "loss": 0.6174,
+      "step": 18653
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 7.5862778377822354e-06,
+      "loss": 0.4129,
+      "step": 18654
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 7.583706218176208e-06,
+      "loss": 0.5429,
+      "step": 18655
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 7.581134598570181e-06,
+      "loss": 1.0183,
+      "step": 18656
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 7.578562978964151e-06,
+      "loss": 0.2211,
+      "step": 18657
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 7.575991359358124e-06,
+      "loss": 0.4792,
+      "step": 18658
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 7.5734197397520966e-06,
+      "loss": 0.4713,
+      "step": 18659
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 7.570848120146068e-06,
+      "loss": 0.7092,
+      "step": 18660
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 7.56827650054004e-06,
+      "loss": 0.7172,
+      "step": 18661
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 7.565704880934013e-06,
+      "loss": 0.4983,
+      "step": 18662
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 7.563133261327985e-06,
+      "loss": 0.4752,
+      "step": 18663
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 7.560561641721957e-06,
+      "loss": 0.4252,
+      "step": 18664
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 7.557990022115929e-06,
+      "loss": 0.6134,
+      "step": 18665
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 7.5554184025099004e-06,
+      "loss": 0.5717,
+      "step": 18666
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 7.5528467829038735e-06,
+      "loss": 0.5691,
+      "step": 18667
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 7.550275163297846e-06,
+      "loss": 0.5632,
+      "step": 18668
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 7.547703543691817e-06,
+      "loss": 0.8638,
+      "step": 18669
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 7.545131924085789e-06,
+      "loss": 0.6074,
+      "step": 18670
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 7.5425603044797624e-06,
+      "loss": 0.8557,
+      "step": 18671
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 7.539988684873733e-06,
+      "loss": 0.627,
+      "step": 18672
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 7.537417065267706e-06,
+      "loss": 0.5289,
+      "step": 18673
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 7.534845445661678e-06,
+      "loss": 0.7855,
+      "step": 18674
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 7.53227382605565e-06,
+      "loss": 0.6828,
+      "step": 18675
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 7.529702206449622e-06,
+      "loss": 1.1459,
+      "step": 18676
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 7.527130586843595e-06,
+      "loss": 0.5076,
+      "step": 18677
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 7.524558967237566e-06,
+      "loss": 0.5745,
+      "step": 18678
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 7.5219873476315385e-06,
+      "loss": 0.6469,
+      "step": 18679
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 7.519415728025512e-06,
+      "loss": 0.4364,
+      "step": 18680
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 7.516844108419482e-06,
+      "loss": 0.5687,
+      "step": 18681
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 7.514272488813455e-06,
+      "loss": 0.6214,
+      "step": 18682
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 7.5117008692074274e-06,
+      "loss": 0.581,
+      "step": 18683
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 7.509129249601399e-06,
+      "loss": 0.5396,
+      "step": 18684
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 7.506557629995371e-06,
+      "loss": 0.6695,
+      "step": 18685
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 7.503986010389344e-06,
+      "loss": 0.6099,
+      "step": 18686
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 7.5014143907833155e-06,
+      "loss": 0.5715,
+      "step": 18687
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 7.498842771177288e-06,
+      "loss": 0.8016,
+      "step": 18688
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 7.49627115157126e-06,
+      "loss": 0.6854,
+      "step": 18689
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 7.493699531965231e-06,
+      "loss": 0.8171,
+      "step": 18690
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 7.491127912359204e-06,
+      "loss": 0.5385,
+      "step": 18691
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 7.488556292753177e-06,
+      "loss": 0.4735,
+      "step": 18692
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 7.485984673147148e-06,
+      "loss": 0.404,
+      "step": 18693
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 7.48341305354112e-06,
+      "loss": 0.5636,
+      "step": 18694
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 7.480841433935093e-06,
+      "loss": 0.4875,
+      "step": 18695
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 7.478269814329064e-06,
+      "loss": 0.4994,
+      "step": 18696
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 7.475698194723037e-06,
+      "loss": 0.4958,
+      "step": 18697
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 7.473126575117009e-06,
+      "loss": 0.5909,
+      "step": 18698
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 7.4705549555109805e-06,
+      "loss": 0.4557,
+      "step": 18699
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 7.467983335904953e-06,
+      "loss": 0.5565,
+      "step": 18700
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 7.465411716298926e-06,
+      "loss": 0.6691,
+      "step": 18701
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 7.462840096692897e-06,
+      "loss": 0.533,
+      "step": 18702
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 7.4602684770868694e-06,
+      "loss": 0.5201,
+      "step": 18703
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 7.457696857480842e-06,
+      "loss": 0.7452,
+      "step": 18704
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 7.455125237874815e-06,
+      "loss": 0.6282,
+      "step": 18705
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 7.452553618268786e-06,
+      "loss": 0.7283,
+      "step": 18706
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 7.449981998662758e-06,
+      "loss": 0.654,
+      "step": 18707
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 7.4474103790567306e-06,
+      "loss": 0.6867,
+      "step": 18708
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 7.444838759450702e-06,
+      "loss": 0.6705,
+      "step": 18709
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 7.442267139844675e-06,
+      "loss": 0.6622,
+      "step": 18710
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 7.439695520238647e-06,
+      "loss": 0.5014,
+      "step": 18711
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 7.437123900632619e-06,
+      "loss": 0.3721,
+      "step": 18712
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 7.434552281026591e-06,
+      "loss": 0.5769,
+      "step": 18713
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 7.431980661420564e-06,
+      "loss": 0.3814,
+      "step": 18714
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 7.4294090418145345e-06,
+      "loss": 0.7385,
+      "step": 18715
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 7.4268374222085075e-06,
+      "loss": 0.6943,
+      "step": 18716
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 7.42426580260248e-06,
+      "loss": 0.5093,
+      "step": 18717
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 7.421694182996451e-06,
+      "loss": 0.6119,
+      "step": 18718
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 7.419122563390423e-06,
+      "loss": 0.4685,
+      "step": 18719
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 7.4165509437843964e-06,
+      "loss": 0.4812,
+      "step": 18720
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 7.413979324178368e-06,
+      "loss": 0.6837,
+      "step": 18721
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 7.41140770457234e-06,
+      "loss": 0.4038,
+      "step": 18722
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 7.408836084966313e-06,
+      "loss": 0.5857,
+      "step": 18723
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 7.406264465360284e-06,
+      "loss": 0.2879,
+      "step": 18724
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 7.403692845754257e-06,
+      "loss": 0.4973,
+      "step": 18725
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 7.401121226148229e-06,
+      "loss": 0.6271,
+      "step": 18726
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 7.3985496065422e-06,
+      "loss": 0.5238,
+      "step": 18727
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 7.3959779869361726e-06,
+      "loss": 0.6175,
+      "step": 18728
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 7.393406367330146e-06,
+      "loss": 0.34,
+      "step": 18729
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 7.390834747724117e-06,
+      "loss": 0.903,
+      "step": 18730
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 7.388263128118089e-06,
+      "loss": 0.7279,
+      "step": 18731
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 7.3856915085120615e-06,
+      "loss": 0.4626,
+      "step": 18732
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 7.383119888906033e-06,
+      "loss": 0.7597,
+      "step": 18733
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 7.380548269300006e-06,
+      "loss": 0.8338,
+      "step": 18734
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 7.377976649693978e-06,
+      "loss": 0.7301,
+      "step": 18735
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 7.3754050300879495e-06,
+      "loss": 0.7619,
+      "step": 18736
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 7.372833410481922e-06,
+      "loss": 0.8972,
+      "step": 18737
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 7.370261790875895e-06,
+      "loss": 0.4182,
+      "step": 18738
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 7.367690171269865e-06,
+      "loss": 0.6421,
+      "step": 18739
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 7.3651185516638384e-06,
+      "loss": 0.8525,
+      "step": 18740
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 7.362546932057811e-06,
+      "loss": 0.7307,
+      "step": 18741
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 7.359975312451782e-06,
+      "loss": 0.7097,
+      "step": 18742
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 7.357403692845754e-06,
+      "loss": 0.5299,
+      "step": 18743
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 7.354832073239727e-06,
+      "loss": 0.6244,
+      "step": 18744
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 7.352260453633699e-06,
+      "loss": 0.6008,
+      "step": 18745
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 7.349688834027671e-06,
+      "loss": 0.6025,
+      "step": 18746
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 7.347117214421643e-06,
+      "loss": 0.7715,
+      "step": 18747
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 7.3445455948156145e-06,
+      "loss": 0.6437,
+      "step": 18748
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 7.341973975209588e-06,
+      "loss": 0.4245,
+      "step": 18749
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 7.33940235560356e-06,
+      "loss": 0.4258,
+      "step": 18750
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 7.336830735997531e-06,
+      "loss": 0.855,
+      "step": 18751
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 7.3342591163915034e-06,
+      "loss": 0.4403,
+      "step": 18752
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 7.3316874967854765e-06,
+      "loss": 0.3448,
+      "step": 18753
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 7.329115877179447e-06,
+      "loss": 0.6175,
+      "step": 18754
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 7.32654425757342e-06,
+      "loss": 0.4893,
+      "step": 18755
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 7.323972637967392e-06,
+      "loss": 0.4621,
+      "step": 18756
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 7.321401018361364e-06,
+      "loss": 0.4461,
+      "step": 18757
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 7.318829398755336e-06,
+      "loss": 0.5075,
+      "step": 18758
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 7.316257779149309e-06,
+      "loss": 0.4956,
+      "step": 18759
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 7.31368615954328e-06,
+      "loss": 0.4402,
+      "step": 18760
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 7.311114539937253e-06,
+      "loss": 0.7179,
+      "step": 18761
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 7.308542920331225e-06,
+      "loss": 0.7548,
+      "step": 18762
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 7.305971300725196e-06,
+      "loss": 0.3431,
+      "step": 18763
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 7.303399681119169e-06,
+      "loss": 0.7933,
+      "step": 18764
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 7.3008280615131415e-06,
+      "loss": 0.6764,
+      "step": 18765
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 7.298256441907113e-06,
+      "loss": 0.4394,
+      "step": 18766
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 7.295684822301085e-06,
+      "loss": 0.4225,
+      "step": 18767
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 7.293113202695058e-06,
+      "loss": 0.5447,
+      "step": 18768
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 7.29054158308903e-06,
+      "loss": 0.8044,
+      "step": 18769
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 7.287969963483002e-06,
+      "loss": 0.456,
+      "step": 18770
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 7.285398343876974e-06,
+      "loss": 0.4291,
+      "step": 18771
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 7.2828267242709454e-06,
+      "loss": 0.487,
+      "step": 18772
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 7.2802551046649185e-06,
+      "loss": 0.5861,
+      "step": 18773
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 7.277683485058891e-06,
+      "loss": 0.6161,
+      "step": 18774
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 7.275111865452862e-06,
+      "loss": 0.6616,
+      "step": 18775
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 7.272540245846834e-06,
+      "loss": 0.6991,
+      "step": 18776
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 7.269968626240807e-06,
+      "loss": 0.4817,
+      "step": 18777
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 7.26739700663478e-06,
+      "loss": 0.3572,
+      "step": 18778
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 7.264825387028751e-06,
+      "loss": 0.4207,
+      "step": 18779
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 7.262253767422723e-06,
+      "loss": 0.6597,
+      "step": 18780
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 7.259682147816696e-06,
+      "loss": 0.5646,
+      "step": 18781
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 7.257110528210667e-06,
+      "loss": 0.5317,
+      "step": 18782
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 7.25453890860464e-06,
+      "loss": 0.2633,
+      "step": 18783
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 7.251967288998612e-06,
+      "loss": 0.4086,
+      "step": 18784
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 7.2493956693925835e-06,
+      "loss": 0.5235,
+      "step": 18785
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 7.246824049786556e-06,
+      "loss": 0.3495,
+      "step": 18786
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 7.244252430180529e-06,
+      "loss": 0.4847,
+      "step": 18787
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 7.2416808105745e-06,
+      "loss": 0.3845,
+      "step": 18788
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 7.2391091909684724e-06,
+      "loss": 0.4363,
+      "step": 18789
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 7.236537571362445e-06,
+      "loss": 0.4024,
+      "step": 18790
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 7.233965951756416e-06,
+      "loss": 0.4343,
+      "step": 18791
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 7.231394332150389e-06,
+      "loss": 0.4902,
+      "step": 18792
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 7.228822712544361e-06,
+      "loss": 0.439,
+      "step": 18793
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 7.226251092938333e-06,
+      "loss": 0.6674,
+      "step": 18794
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 7.223679473332305e-06,
+      "loss": 0.6331,
+      "step": 18795
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 7.221107853726278e-06,
+      "loss": 0.3722,
+      "step": 18796
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 7.2185362341202486e-06,
+      "loss": 0.3414,
+      "step": 18797
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 7.215964614514222e-06,
+      "loss": 0.6348,
+      "step": 18798
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 7.213392994908194e-06,
+      "loss": 0.7294,
+      "step": 18799
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 7.210821375302165e-06,
+      "loss": 0.7532,
+      "step": 18800
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 7.2082497556961375e-06,
+      "loss": 0.8131,
+      "step": 18801
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 7.2056781360901105e-06,
+      "loss": 0.6633,
+      "step": 18802
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 7.203106516484082e-06,
+      "loss": 0.7932,
+      "step": 18803
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 7.200534896878054e-06,
+      "loss": 0.8053,
+      "step": 18804
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 7.197963277272027e-06,
+      "loss": 0.5615,
+      "step": 18805
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 7.195391657665998e-06,
+      "loss": 0.4596,
+      "step": 18806
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 7.192820038059971e-06,
+      "loss": 0.7436,
+      "step": 18807
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 7.190248418453943e-06,
+      "loss": 0.3554,
+      "step": 18808
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 7.187676798847914e-06,
+      "loss": 0.3786,
+      "step": 18809
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 7.185105179241887e-06,
+      "loss": 0.6188,
+      "step": 18810
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 7.18253355963586e-06,
+      "loss": 0.495,
+      "step": 18811
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 7.179961940029831e-06,
+      "loss": 0.545,
+      "step": 18812
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 7.177390320423803e-06,
+      "loss": 0.5615,
+      "step": 18813
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 7.1748187008177756e-06,
+      "loss": 0.377,
+      "step": 18814
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 7.172247081211747e-06,
+      "loss": 0.5146,
+      "step": 18815
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 7.16967546160572e-06,
+      "loss": 0.4494,
+      "step": 18816
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 7.167103841999692e-06,
+      "loss": 0.5865,
+      "step": 18817
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 7.164532222393664e-06,
+      "loss": 0.604,
+      "step": 18818
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 7.161960602787636e-06,
+      "loss": 0.5497,
+      "step": 18819
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 7.159388983181609e-06,
+      "loss": 0.5101,
+      "step": 18820
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 7.1568173635755794e-06,
+      "loss": 0.8312,
+      "step": 18821
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 7.1542457439695525e-06,
+      "loss": 0.6309,
+      "step": 18822
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 7.151674124363525e-06,
+      "loss": 0.8696,
+      "step": 18823
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 7.149102504757496e-06,
+      "loss": 0.4594,
+      "step": 18824
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 7.146530885151468e-06,
+      "loss": 0.3927,
+      "step": 18825
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 7.143959265545441e-06,
+      "loss": 0.7007,
+      "step": 18826
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 7.141387645939413e-06,
+      "loss": 0.6694,
+      "step": 18827
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 7.138816026333385e-06,
+      "loss": 0.5785,
+      "step": 18828
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 7.136244406727357e-06,
+      "loss": 0.6658,
+      "step": 18829
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 7.133672787121329e-06,
+      "loss": 0.6818,
+      "step": 18830
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 7.131101167515302e-06,
+      "loss": 0.5629,
+      "step": 18831
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 7.128529547909274e-06,
+      "loss": 0.8696,
+      "step": 18832
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 7.125957928303245e-06,
+      "loss": 0.4697,
+      "step": 18833
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 7.1233863086972175e-06,
+      "loss": 0.7924,
+      "step": 18834
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 7.120814689091191e-06,
+      "loss": 0.377,
+      "step": 18835
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 7.118243069485161e-06,
+      "loss": 0.6537,
+      "step": 18836
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 7.115671449879134e-06,
+      "loss": 0.3415,
+      "step": 18837
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 7.1130998302731064e-06,
+      "loss": 0.3798,
+      "step": 18838
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 7.110528210667078e-06,
+      "loss": 0.5624,
+      "step": 18839
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 7.10795659106105e-06,
+      "loss": 0.6338,
+      "step": 18840
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 7.105384971455023e-06,
+      "loss": 0.7481,
+      "step": 18841
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 7.1028133518489945e-06,
+      "loss": 0.286,
+      "step": 18842
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 7.100241732242967e-06,
+      "loss": 0.6068,
+      "step": 18843
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 7.097670112636939e-06,
+      "loss": 0.6283,
+      "step": 18844
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 7.09509849303091e-06,
+      "loss": 0.7961,
+      "step": 18845
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 7.092526873424883e-06,
+      "loss": 0.4068,
+      "step": 18846
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 7.089955253818856e-06,
+      "loss": 0.7798,
+      "step": 18847
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 7.087383634212827e-06,
+      "loss": 0.8727,
+      "step": 18848
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 7.084812014606799e-06,
+      "loss": 0.6779,
+      "step": 18849
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 7.082240395000772e-06,
+      "loss": 0.457,
+      "step": 18850
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 7.079668775394743e-06,
+      "loss": 0.3493,
+      "step": 18851
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 7.077097155788716e-06,
+      "loss": 0.5298,
+      "step": 18852
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 7.074525536182688e-06,
+      "loss": 0.3935,
+      "step": 18853
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 7.071953916576661e-06,
+      "loss": 0.5862,
+      "step": 18854
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 7.069382296970633e-06,
+      "loss": 0.7067,
+      "step": 18855
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 7.066810677364605e-06,
+      "loss": 0.5712,
+      "step": 18856
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 7.064239057758577e-06,
+      "loss": 0.7961,
+      "step": 18857
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 7.0616674381525484e-06,
+      "loss": 0.5742,
+      "step": 18858
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 7.0590958185465215e-06,
+      "loss": 0.7766,
+      "step": 18859
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 7.056524198940494e-06,
+      "loss": 0.4782,
+      "step": 18860
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 7.053952579334465e-06,
+      "loss": 0.5304,
+      "step": 18861
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 7.051380959728437e-06,
+      "loss": 0.5154,
+      "step": 18862
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 7.04880934012241e-06,
+      "loss": 0.5744,
+      "step": 18863
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 7.046237720516381e-06,
+      "loss": 0.691,
+      "step": 18864
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 7.043666100910354e-06,
+      "loss": 0.3692,
+      "step": 18865
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 7.041094481304326e-06,
+      "loss": 0.6514,
+      "step": 18866
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 7.038522861698298e-06,
+      "loss": 0.3508,
+      "step": 18867
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 7.03595124209227e-06,
+      "loss": 0.7129,
+      "step": 18868
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 7.033379622486243e-06,
+      "loss": 0.3617,
+      "step": 18869
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 7.030808002880214e-06,
+      "loss": 0.7144,
+      "step": 18870
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 7.0282363832741865e-06,
+      "loss": 0.5841,
+      "step": 18871
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 7.025664763668159e-06,
+      "loss": 0.7217,
+      "step": 18872
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 7.02309314406213e-06,
+      "loss": 0.6406,
+      "step": 18873
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 7.020521524456103e-06,
+      "loss": 0.5905,
+      "step": 18874
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 7.0179499048500754e-06,
+      "loss": 0.5339,
+      "step": 18875
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 7.015378285244047e-06,
+      "loss": 0.651,
+      "step": 18876
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 7.012806665638019e-06,
+      "loss": 0.6724,
+      "step": 18877
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 7.010235046031992e-06,
+      "loss": 0.3898,
+      "step": 18878
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 7.007663426425963e-06,
+      "loss": 1.2194,
+      "step": 18879
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 7.005091806819936e-06,
+      "loss": 0.6175,
+      "step": 18880
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 7.002520187213908e-06,
+      "loss": 0.8999,
+      "step": 18881
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 6.999948567607879e-06,
+      "loss": 0.856,
+      "step": 18882
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 6.9973769480018515e-06,
+      "loss": 0.4677,
+      "step": 18883
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 6.994805328395825e-06,
+      "loss": 0.4691,
+      "step": 18884
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 6.992233708789796e-06,
+      "loss": 0.5091,
+      "step": 18885
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 6.989662089183768e-06,
+      "loss": 0.5775,
+      "step": 18886
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 6.987090469577741e-06,
+      "loss": 0.3674,
+      "step": 18887
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 6.984518849971712e-06,
+      "loss": 0.7856,
+      "step": 18888
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 6.981947230365685e-06,
+      "loss": 0.6716,
+      "step": 18889
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 6.979375610759657e-06,
+      "loss": 0.5564,
+      "step": 18890
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 6.9768039911536285e-06,
+      "loss": 0.6093,
+      "step": 18891
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 6.974232371547601e-06,
+      "loss": 0.4125,
+      "step": 18892
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 6.971660751941574e-06,
+      "loss": 0.338,
+      "step": 18893
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 6.969089132335545e-06,
+      "loss": 0.4632,
+      "step": 18894
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 6.966517512729517e-06,
+      "loss": 0.5462,
+      "step": 18895
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 6.96394589312349e-06,
+      "loss": 0.3153,
+      "step": 18896
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 6.961374273517461e-06,
+      "loss": 0.5697,
+      "step": 18897
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 6.958802653911434e-06,
+      "loss": 0.9994,
+      "step": 18898
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 6.956231034305406e-06,
+      "loss": 0.5464,
+      "step": 18899
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 6.953659414699378e-06,
+      "loss": 0.7931,
+      "step": 18900
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 6.95108779509335e-06,
+      "loss": 0.676,
+      "step": 18901
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 6.948516175487323e-06,
+      "loss": 0.5157,
+      "step": 18902
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 6.9459445558812935e-06,
+      "loss": 0.9773,
+      "step": 18903
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 6.943372936275267e-06,
+      "loss": 0.6154,
+      "step": 18904
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 6.940801316669239e-06,
+      "loss": 0.3858,
+      "step": 18905
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 6.93822969706321e-06,
+      "loss": 0.5055,
+      "step": 18906
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 6.9356580774571824e-06,
+      "loss": 0.5793,
+      "step": 18907
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 6.9330864578511555e-06,
+      "loss": 0.4285,
+      "step": 18908
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 6.930514838245127e-06,
+      "loss": 0.8041,
+      "step": 18909
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 6.927943218639099e-06,
+      "loss": 0.3872,
+      "step": 18910
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 6.925371599033071e-06,
+      "loss": 0.8449,
+      "step": 18911
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 6.922799979427043e-06,
+      "loss": 0.5409,
+      "step": 18912
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 6.920228359821016e-06,
+      "loss": 0.5231,
+      "step": 18913
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 6.917656740214988e-06,
+      "loss": 0.7503,
+      "step": 18914
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 6.915085120608959e-06,
+      "loss": 0.6172,
+      "step": 18915
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 6.912513501002932e-06,
+      "loss": 0.722,
+      "step": 18916
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 6.909941881396905e-06,
+      "loss": 0.4377,
+      "step": 18917
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 6.907370261790875e-06,
+      "loss": 0.4022,
+      "step": 18918
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 6.904798642184848e-06,
+      "loss": 0.5024,
+      "step": 18919
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 6.9022270225788205e-06,
+      "loss": 0.8454,
+      "step": 18920
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 6.899655402972792e-06,
+      "loss": 0.7776,
+      "step": 18921
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 6.897083783366764e-06,
+      "loss": 0.5317,
+      "step": 18922
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 6.894512163760737e-06,
+      "loss": 0.4144,
+      "step": 18923
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 6.891940544154709e-06,
+      "loss": 0.7262,
+      "step": 18924
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 6.889368924548681e-06,
+      "loss": 0.5786,
+      "step": 18925
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 6.886797304942653e-06,
+      "loss": 0.5789,
+      "step": 18926
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 6.8842256853366244e-06,
+      "loss": 0.4596,
+      "step": 18927
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 6.8816540657305975e-06,
+      "loss": 0.7626,
+      "step": 18928
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 6.87908244612457e-06,
+      "loss": 0.5802,
+      "step": 18929
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 6.876510826518543e-06,
+      "loss": 0.5414,
+      "step": 18930
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 6.873939206912513e-06,
+      "loss": 0.3338,
+      "step": 18931
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 6.871367587306486e-06,
+      "loss": 0.5034,
+      "step": 18932
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 6.868795967700459e-06,
+      "loss": 0.7755,
+      "step": 18933
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 6.86622434809443e-06,
+      "loss": 0.4749,
+      "step": 18934
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 6.863652728488402e-06,
+      "loss": 0.7052,
+      "step": 18935
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 6.861081108882375e-06,
+      "loss": 0.4574,
+      "step": 18936
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 6.858509489276347e-06,
+      "loss": 0.4791,
+      "step": 18937
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 6.855937869670319e-06,
+      "loss": 0.571,
+      "step": 18938
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 6.853366250064291e-06,
+      "loss": 0.7859,
+      "step": 18939
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 6.8507946304582625e-06,
+      "loss": 0.5628,
+      "step": 18940
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 6.848223010852236e-06,
+      "loss": 0.5051,
+      "step": 18941
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 6.845651391246208e-06,
+      "loss": 0.6486,
+      "step": 18942
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 6.843079771640179e-06,
+      "loss": 0.4567,
+      "step": 18943
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 6.8405081520341514e-06,
+      "loss": 0.5078,
+      "step": 18944
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 6.8379365324281245e-06,
+      "loss": 0.4265,
+      "step": 18945
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 6.835364912822095e-06,
+      "loss": 0.7278,
+      "step": 18946
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 6.832793293216068e-06,
+      "loss": 0.6683,
+      "step": 18947
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 6.83022167361004e-06,
+      "loss": 0.5873,
+      "step": 18948
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 6.827650054004012e-06,
+      "loss": 0.8598,
+      "step": 18949
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 6.825078434397984e-06,
+      "loss": 0.373,
+      "step": 18950
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 6.822506814791957e-06,
+      "loss": 0.6158,
+      "step": 18951
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 6.819935195185928e-06,
+      "loss": 0.6783,
+      "step": 18952
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 6.817363575579901e-06,
+      "loss": 0.7286,
+      "step": 18953
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 6.814791955973873e-06,
+      "loss": 0.5295,
+      "step": 18954
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 6.812220336367844e-06,
+      "loss": 0.2956,
+      "step": 18955
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 6.809648716761817e-06,
+      "loss": 0.3682,
+      "step": 18956
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 6.8070770971557895e-06,
+      "loss": 0.5662,
+      "step": 18957
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 6.804505477549761e-06,
+      "loss": 0.5292,
+      "step": 18958
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 6.801933857943733e-06,
+      "loss": 0.4386,
+      "step": 18959
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 6.799362238337706e-06,
+      "loss": 0.5169,
+      "step": 18960
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 6.796790618731677e-06,
+      "loss": 0.7341,
+      "step": 18961
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 6.79421899912565e-06,
+      "loss": 0.5446,
+      "step": 18962
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 6.791647379519622e-06,
+      "loss": 0.5701,
+      "step": 18963
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 6.789075759913593e-06,
+      "loss": 0.5966,
+      "step": 18964
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 6.786504140307566e-06,
+      "loss": 0.3637,
+      "step": 18965
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 6.783932520701539e-06,
+      "loss": 0.5905,
+      "step": 18966
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 6.78136090109551e-06,
+      "loss": 0.5904,
+      "step": 18967
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 6.778789281489482e-06,
+      "loss": 0.6619,
+      "step": 18968
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 6.776217661883455e-06,
+      "loss": 0.2525,
+      "step": 18969
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 6.773646042277426e-06,
+      "loss": 0.4186,
+      "step": 18970
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 6.771074422671399e-06,
+      "loss": 0.429,
+      "step": 18971
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 6.768502803065371e-06,
+      "loss": 0.7146,
+      "step": 18972
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 6.765931183459343e-06,
+      "loss": 0.4798,
+      "step": 18973
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 6.763359563853315e-06,
+      "loss": 0.3552,
+      "step": 18974
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 6.760787944247288e-06,
+      "loss": 0.2687,
+      "step": 18975
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 6.758216324641259e-06,
+      "loss": 0.3946,
+      "step": 18976
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 6.7556447050352315e-06,
+      "loss": 0.6872,
+      "step": 18977
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 6.753073085429204e-06,
+      "loss": 0.595,
+      "step": 18978
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 6.750501465823175e-06,
+      "loss": 0.6326,
+      "step": 18979
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 6.747929846217148e-06,
+      "loss": 0.5146,
+      "step": 18980
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 6.74535822661112e-06,
+      "loss": 0.7723,
+      "step": 18981
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 6.742786607005092e-06,
+      "loss": 0.3433,
+      "step": 18982
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 6.740214987399064e-06,
+      "loss": 0.511,
+      "step": 18983
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 6.737643367793037e-06,
+      "loss": 0.4526,
+      "step": 18984
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 6.735071748187008e-06,
+      "loss": 0.4612,
+      "step": 18985
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 6.732500128580981e-06,
+      "loss": 0.7477,
+      "step": 18986
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 6.729928508974953e-06,
+      "loss": 0.4352,
+      "step": 18987
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 6.727356889368924e-06,
+      "loss": 0.5329,
+      "step": 18988
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 6.7247852697628965e-06,
+      "loss": 0.4615,
+      "step": 18989
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 6.72221365015687e-06,
+      "loss": 0.5834,
+      "step": 18990
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 6.719642030550841e-06,
+      "loss": 0.8348,
+      "step": 18991
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 6.717070410944813e-06,
+      "loss": 0.826,
+      "step": 18992
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 6.7144987913387854e-06,
+      "loss": 0.6319,
+      "step": 18993
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 6.711927171732757e-06,
+      "loss": 0.9033,
+      "step": 18994
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 6.70935555212673e-06,
+      "loss": 0.4338,
+      "step": 18995
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 6.706783932520702e-06,
+      "loss": 0.7719,
+      "step": 18996
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 6.7042123129146735e-06,
+      "loss": 0.4504,
+      "step": 18997
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 6.701640693308646e-06,
+      "loss": 0.4557,
+      "step": 18998
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 6.699069073702619e-06,
+      "loss": 0.4517,
+      "step": 18999
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 6.696497454096589e-06,
+      "loss": 0.3183,
+      "step": 19000
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 6.693925834490562e-06,
+      "loss": 0.6038,
+      "step": 19001
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 6.691354214884535e-06,
+      "loss": 0.5687,
+      "step": 19002
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 6.688782595278506e-06,
+      "loss": 0.5178,
+      "step": 19003
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 6.686210975672478e-06,
+      "loss": 0.3749,
+      "step": 19004
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 6.683639356066451e-06,
+      "loss": 0.8525,
+      "step": 19005
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 6.6810677364604235e-06,
+      "loss": 0.5933,
+      "step": 19006
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 6.678496116854395e-06,
+      "loss": 0.6804,
+      "step": 19007
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 6.675924497248367e-06,
+      "loss": 0.3625,
+      "step": 19008
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 6.67335287764234e-06,
+      "loss": 0.5339,
+      "step": 19009
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 6.670781258036312e-06,
+      "loss": 0.6221,
+      "step": 19010
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 6.668209638430284e-06,
+      "loss": 0.3085,
+      "step": 19011
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 6.665638018824257e-06,
+      "loss": 0.6212,
+      "step": 19012
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 6.6630663992182274e-06,
+      "loss": 0.8244,
+      "step": 19013
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 6.6604947796122005e-06,
+      "loss": 0.6777,
+      "step": 19014
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 6.657923160006173e-06,
+      "loss": 0.5585,
+      "step": 19015
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 6.655351540400144e-06,
+      "loss": 0.6111,
+      "step": 19016
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 6.652779920794116e-06,
+      "loss": 0.6808,
+      "step": 19017
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 6.650208301188089e-06,
+      "loss": 0.3609,
+      "step": 19018
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 6.647636681582061e-06,
+      "loss": 0.6937,
+      "step": 19019
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 6.645065061976033e-06,
+      "loss": 0.4833,
+      "step": 19020
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 6.642493442370005e-06,
+      "loss": 0.4944,
+      "step": 19021
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 6.639921822763977e-06,
+      "loss": 0.435,
+      "step": 19022
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 6.63735020315795e-06,
+      "loss": 0.3318,
+      "step": 19023
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 6.634778583551922e-06,
+      "loss": 0.5674,
+      "step": 19024
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 6.632206963945893e-06,
+      "loss": 0.8478,
+      "step": 19025
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 6.6296353443398655e-06,
+      "loss": 0.712,
+      "step": 19026
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 6.627063724733839e-06,
+      "loss": 0.7594,
+      "step": 19027
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 6.624492105127809e-06,
+      "loss": 0.439,
+      "step": 19028
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 6.621920485521782e-06,
+      "loss": 0.6904,
+      "step": 19029
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 6.6193488659157544e-06,
+      "loss": 0.5352,
+      "step": 19030
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 6.616777246309726e-06,
+      "loss": 0.5809,
+      "step": 19031
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 6.614205626703698e-06,
+      "loss": 0.6563,
+      "step": 19032
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 6.611634007097671e-06,
+      "loss": 0.869,
+      "step": 19033
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 6.6090623874916425e-06,
+      "loss": 0.6223,
+      "step": 19034
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 6.606490767885615e-06,
+      "loss": 0.4057,
+      "step": 19035
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 6.603919148279587e-06,
+      "loss": 0.465,
+      "step": 19036
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 6.601347528673558e-06,
+      "loss": 0.326,
+      "step": 19037
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 6.598775909067531e-06,
+      "loss": 0.4902,
+      "step": 19038
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 6.596204289461504e-06,
+      "loss": 0.7595,
+      "step": 19039
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 6.593632669855475e-06,
+      "loss": 0.4429,
+      "step": 19040
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 6.591061050249447e-06,
+      "loss": 0.4311,
+      "step": 19041
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 6.58848943064342e-06,
+      "loss": 0.6433,
+      "step": 19042
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 6.585917811037391e-06,
+      "loss": 0.4171,
+      "step": 19043
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 6.583346191431364e-06,
+      "loss": 0.4314,
+      "step": 19044
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 6.580774571825336e-06,
+      "loss": 0.3716,
+      "step": 19045
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 6.5782029522193075e-06,
+      "loss": 0.6357,
+      "step": 19046
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 6.57563133261328e-06,
+      "loss": 0.3315,
+      "step": 19047
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 6.573059713007253e-06,
+      "loss": 0.5594,
+      "step": 19048
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 6.570488093401224e-06,
+      "loss": 0.6991,
+      "step": 19049
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 6.567916473795196e-06,
+      "loss": 0.3556,
+      "step": 19050
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 6.565344854189169e-06,
+      "loss": 0.4801,
+      "step": 19051
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 6.56277323458314e-06,
+      "loss": 0.6049,
+      "step": 19052
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 6.560201614977113e-06,
+      "loss": 0.779,
+      "step": 19053
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 6.557629995371085e-06,
+      "loss": 0.6942,
+      "step": 19054
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 6.555058375765057e-06,
+      "loss": 0.5458,
+      "step": 19055
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 6.552486756159029e-06,
+      "loss": 0.4412,
+      "step": 19056
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 6.549915136553002e-06,
+      "loss": 0.553,
+      "step": 19057
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 6.547343516946973e-06,
+      "loss": 0.4915,
+      "step": 19058
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 6.544771897340946e-06,
+      "loss": 0.3696,
+      "step": 19059
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 6.542200277734918e-06,
+      "loss": 0.5755,
+      "step": 19060
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 6.539628658128889e-06,
+      "loss": 0.5458,
+      "step": 19061
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 6.537057038522862e-06,
+      "loss": 0.418,
+      "step": 19062
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 6.5344854189168345e-06,
+      "loss": 0.3915,
+      "step": 19063
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 6.531913799310806e-06,
+      "loss": 0.4821,
+      "step": 19064
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 6.529342179704778e-06,
+      "loss": 0.6091,
+      "step": 19065
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 6.526770560098751e-06,
+      "loss": 0.4137,
+      "step": 19066
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 6.524198940492722e-06,
+      "loss": 0.4955,
+      "step": 19067
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 6.521627320886695e-06,
+      "loss": 0.6011,
+      "step": 19068
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 6.519055701280667e-06,
+      "loss": 0.3892,
+      "step": 19069
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 6.516484081674638e-06,
+      "loss": 0.5173,
+      "step": 19070
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 6.513912462068611e-06,
+      "loss": 0.4931,
+      "step": 19071
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 6.511340842462584e-06,
+      "loss": 0.3684,
+      "step": 19072
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 6.508769222856555e-06,
+      "loss": 0.3123,
+      "step": 19073
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 6.506197603250527e-06,
+      "loss": 0.9168,
+      "step": 19074
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 6.5036259836444995e-06,
+      "loss": 0.4068,
+      "step": 19075
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 6.501054364038471e-06,
+      "loss": 0.3192,
+      "step": 19076
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 6.498482744432444e-06,
+      "loss": 0.5681,
+      "step": 19077
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 6.495911124826416e-06,
+      "loss": 0.5815,
+      "step": 19078
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 6.493339505220388e-06,
+      "loss": 0.659,
+      "step": 19079
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 6.49076788561436e-06,
+      "loss": 0.6104,
+      "step": 19080
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 6.488196266008333e-06,
+      "loss": 0.7238,
+      "step": 19081
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 6.485624646402305e-06,
+      "loss": 0.7062,
+      "step": 19082
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 6.4830530267962765e-06,
+      "loss": 0.3066,
+      "step": 19083
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 6.480481407190249e-06,
+      "loss": 0.6745,
+      "step": 19084
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 6.477909787584222e-06,
+      "loss": 0.7737,
+      "step": 19085
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 6.475338167978192e-06,
+      "loss": 0.5189,
+      "step": 19086
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 6.472766548372165e-06,
+      "loss": 0.9731,
+      "step": 19087
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 6.470194928766138e-06,
+      "loss": 0.7237,
+      "step": 19088
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 6.467623309160109e-06,
+      "loss": 0.437,
+      "step": 19089
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 6.465051689554081e-06,
+      "loss": 0.6487,
+      "step": 19090
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 6.462480069948054e-06,
+      "loss": 0.5601,
+      "step": 19091
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 6.459908450342026e-06,
+      "loss": 0.6734,
+      "step": 19092
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 6.457336830735998e-06,
+      "loss": 0.4945,
+      "step": 19093
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 6.454765211129971e-06,
+      "loss": 0.5875,
+      "step": 19094
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 6.4521935915239415e-06,
+      "loss": 0.8918,
+      "step": 19095
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 6.449621971917915e-06,
+      "loss": 0.5497,
+      "step": 19096
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 6.447050352311887e-06,
+      "loss": 0.6368,
+      "step": 19097
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 6.444478732705858e-06,
+      "loss": 0.7058,
+      "step": 19098
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 6.4419071130998304e-06,
+      "loss": 0.7315,
+      "step": 19099
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 6.4393354934938035e-06,
+      "loss": 0.6323,
+      "step": 19100
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 6.436763873887775e-06,
+      "loss": 0.4337,
+      "step": 19101
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 6.434192254281747e-06,
+      "loss": 0.4891,
+      "step": 19102
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 6.431620634675719e-06,
+      "loss": 0.7985,
+      "step": 19103
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 6.429049015069691e-06,
+      "loss": 0.7048,
+      "step": 19104
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 6.426477395463664e-06,
+      "loss": 0.6207,
+      "step": 19105
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 6.423905775857636e-06,
+      "loss": 0.5153,
+      "step": 19106
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 6.421334156251607e-06,
+      "loss": 0.4554,
+      "step": 19107
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 6.41876253664558e-06,
+      "loss": 0.6137,
+      "step": 19108
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 6.416190917039553e-06,
+      "loss": 0.5712,
+      "step": 19109
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 6.413619297433523e-06,
+      "loss": 0.5675,
+      "step": 19110
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 6.411047677827496e-06,
+      "loss": 0.6457,
+      "step": 19111
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 6.4084760582214685e-06,
+      "loss": 0.5044,
+      "step": 19112
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 6.40590443861544e-06,
+      "loss": 0.8316,
+      "step": 19113
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 6.403332819009412e-06,
+      "loss": 0.5833,
+      "step": 19114
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 6.400761199403385e-06,
+      "loss": 0.573,
+      "step": 19115
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 6.398189579797357e-06,
+      "loss": 0.4626,
+      "step": 19116
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 6.395617960191329e-06,
+      "loss": 0.5032,
+      "step": 19117
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 6.393046340585301e-06,
+      "loss": 0.543,
+      "step": 19118
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 6.390474720979272e-06,
+      "loss": 0.5323,
+      "step": 19119
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 6.3879031013732455e-06,
+      "loss": 0.5763,
+      "step": 19120
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 6.385331481767218e-06,
+      "loss": 0.6131,
+      "step": 19121
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 6.382759862161189e-06,
+      "loss": 0.5187,
+      "step": 19122
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 6.380188242555161e-06,
+      "loss": 0.5603,
+      "step": 19123
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 6.377616622949134e-06,
+      "loss": 0.3548,
+      "step": 19124
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 6.375045003343105e-06,
+      "loss": 0.8226,
+      "step": 19125
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 6.372473383737078e-06,
+      "loss": 0.6188,
+      "step": 19126
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 6.36990176413105e-06,
+      "loss": 0.453,
+      "step": 19127
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 6.367330144525022e-06,
+      "loss": 0.792,
+      "step": 19128
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 6.364758524918994e-06,
+      "loss": 0.8175,
+      "step": 19129
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 6.362186905312967e-06,
+      "loss": 0.8245,
+      "step": 19130
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 6.359615285706938e-06,
+      "loss": 0.5669,
+      "step": 19131
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 6.3570436661009105e-06,
+      "loss": 0.6566,
+      "step": 19132
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 6.354472046494883e-06,
+      "loss": 0.6057,
+      "step": 19133
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 6.351900426888854e-06,
+      "loss": 0.5114,
+      "step": 19134
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 6.349328807282827e-06,
+      "loss": 0.6029,
+      "step": 19135
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 6.346757187676799e-06,
+      "loss": 0.6098,
+      "step": 19136
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 6.344185568070771e-06,
+      "loss": 0.753,
+      "step": 19137
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 6.341613948464743e-06,
+      "loss": 0.5668,
+      "step": 19138
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 6.339042328858716e-06,
+      "loss": 0.4108,
+      "step": 19139
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 6.336470709252687e-06,
+      "loss": 0.4389,
+      "step": 19140
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 6.33389908964666e-06,
+      "loss": 0.6345,
+      "step": 19141
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 6.331327470040632e-06,
+      "loss": 0.5683,
+      "step": 19142
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 6.328755850434603e-06,
+      "loss": 0.5311,
+      "step": 19143
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 6.326184230828576e-06,
+      "loss": 0.6276,
+      "step": 19144
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 6.323612611222549e-06,
+      "loss": 0.2134,
+      "step": 19145
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 6.32104099161652e-06,
+      "loss": 0.5407,
+      "step": 19146
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 6.318469372010492e-06,
+      "loss": 0.6793,
+      "step": 19147
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 6.315897752404465e-06,
+      "loss": 0.2781,
+      "step": 19148
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 6.313326132798436e-06,
+      "loss": 0.8544,
+      "step": 19149
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 6.310754513192409e-06,
+      "loss": 0.7116,
+      "step": 19150
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 6.308182893586381e-06,
+      "loss": 0.6799,
+      "step": 19151
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 6.3056112739803525e-06,
+      "loss": 0.582,
+      "step": 19152
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 6.303039654374325e-06,
+      "loss": 0.874,
+      "step": 19153
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 6.300468034768298e-06,
+      "loss": 0.848,
+      "step": 19154
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 6.29789641516227e-06,
+      "loss": 0.4153,
+      "step": 19155
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 6.295324795556241e-06,
+      "loss": 0.5788,
+      "step": 19156
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 6.292753175950214e-06,
+      "loss": 1.0304,
+      "step": 19157
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 6.290181556344187e-06,
+      "loss": 0.8245,
+      "step": 19158
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 6.287609936738158e-06,
+      "loss": 0.5399,
+      "step": 19159
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 6.28503831713213e-06,
+      "loss": 0.4993,
+      "step": 19160
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 6.2824666975261025e-06,
+      "loss": 0.3817,
+      "step": 19161
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 6.279895077920074e-06,
+      "loss": 0.4844,
+      "step": 19162
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 6.277323458314047e-06,
+      "loss": 0.5858,
+      "step": 19163
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 6.274751838708019e-06,
+      "loss": 0.4637,
+      "step": 19164
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 6.272180219101991e-06,
+      "loss": 0.5506,
+      "step": 19165
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 6.269608599495963e-06,
+      "loss": 0.9562,
+      "step": 19166
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 6.267036979889936e-06,
+      "loss": 0.4771,
+      "step": 19167
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 6.264465360283906e-06,
+      "loss": 0.5253,
+      "step": 19168
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 6.2618937406778795e-06,
+      "loss": 0.8228,
+      "step": 19169
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 6.259322121071852e-06,
+      "loss": 0.6133,
+      "step": 19170
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 6.256750501465823e-06,
+      "loss": 0.7714,
+      "step": 19171
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 6.254178881859795e-06,
+      "loss": 0.6449,
+      "step": 19172
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 6.251607262253768e-06,
+      "loss": 0.5401,
+      "step": 19173
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 6.24903564264774e-06,
+      "loss": 0.6978,
+      "step": 19174
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 6.246464023041712e-06,
+      "loss": 0.638,
+      "step": 19175
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 6.243892403435684e-06,
+      "loss": 0.6512,
+      "step": 19176
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 6.2413207838296565e-06,
+      "loss": 0.5541,
+      "step": 19177
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 6.238749164223629e-06,
+      "loss": 0.6325,
+      "step": 19178
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 6.2361775446176e-06,
+      "loss": 0.7017,
+      "step": 19179
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 6.233605925011573e-06,
+      "loss": 0.5297,
+      "step": 19180
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 6.2310343054055445e-06,
+      "loss": 0.5348,
+      "step": 19181
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 6.228462685799517e-06,
+      "loss": 0.858,
+      "step": 19182
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 6.225891066193489e-06,
+      "loss": 0.5417,
+      "step": 19183
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 6.223319446587461e-06,
+      "loss": 0.5411,
+      "step": 19184
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 6.220747826981433e-06,
+      "loss": 0.3968,
+      "step": 19185
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 6.218176207375406e-06,
+      "loss": 0.4684,
+      "step": 19186
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 6.215604587769378e-06,
+      "loss": 0.5007,
+      "step": 19187
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 6.213032968163349e-06,
+      "loss": 0.5513,
+      "step": 19188
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 6.210461348557322e-06,
+      "loss": 0.5137,
+      "step": 19189
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 6.207889728951294e-06,
+      "loss": 0.4141,
+      "step": 19190
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 6.205318109345266e-06,
+      "loss": 0.4189,
+      "step": 19191
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 6.202746489739238e-06,
+      "loss": 0.5474,
+      "step": 19192
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 6.20017487013321e-06,
+      "loss": 0.5915,
+      "step": 19193
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 6.197603250527182e-06,
+      "loss": 0.7702,
+      "step": 19194
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 6.195031630921155e-06,
+      "loss": 0.5785,
+      "step": 19195
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 6.192460011315126e-06,
+      "loss": 0.41,
+      "step": 19196
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 6.1898883917090984e-06,
+      "loss": 0.2939,
+      "step": 19197
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 6.187316772103071e-06,
+      "loss": 0.4,
+      "step": 19198
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 6.184745152497043e-06,
+      "loss": 0.5538,
+      "step": 19199
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 6.182173532891015e-06,
+      "loss": 0.5497,
+      "step": 19200
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 6.179601913284987e-06,
+      "loss": 0.5421,
+      "step": 19201
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 6.1770302936789596e-06,
+      "loss": 0.6147,
+      "step": 19202
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 6.174458674072931e-06,
+      "loss": 0.7035,
+      "step": 19203
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 6.171887054466904e-06,
+      "loss": 0.4816,
+      "step": 19204
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 6.169315434860875e-06,
+      "loss": 0.4222,
+      "step": 19205
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 6.166743815254848e-06,
+      "loss": 0.6668,
+      "step": 19206
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 6.16417219564882e-06,
+      "loss": 1.0158,
+      "step": 19207
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 6.161600576042792e-06,
+      "loss": 0.7975,
+      "step": 19208
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 6.1590289564367635e-06,
+      "loss": 0.5711,
+      "step": 19209
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 6.1564573368307365e-06,
+      "loss": 0.5165,
+      "step": 19210
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 6.153885717224708e-06,
+      "loss": 0.433,
+      "step": 19211
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 6.151314097618681e-06,
+      "loss": 0.3924,
+      "step": 19212
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 6.148742478012652e-06,
+      "loss": 0.7832,
+      "step": 19213
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 6.146170858406625e-06,
+      "loss": 0.9006,
+      "step": 19214
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 6.143599238800597e-06,
+      "loss": 0.298,
+      "step": 19215
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 6.141027619194569e-06,
+      "loss": 0.6385,
+      "step": 19216
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 6.138455999588541e-06,
+      "loss": 0.5121,
+      "step": 19217
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 6.1358843799825135e-06,
+      "loss": 0.3507,
+      "step": 19218
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 6.133312760376486e-06,
+      "loss": 0.7122,
+      "step": 19219
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 6.130741140770457e-06,
+      "loss": 0.9406,
+      "step": 19220
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 6.12816952116443e-06,
+      "loss": 0.61,
+      "step": 19221
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 6.1255979015584016e-06,
+      "loss": 0.9018,
+      "step": 19222
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 6.123026281952374e-06,
+      "loss": 0.7599,
+      "step": 19223
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 6.120454662346346e-06,
+      "loss": 1.0676,
+      "step": 19224
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 6.117883042740318e-06,
+      "loss": 0.3815,
+      "step": 19225
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 6.1153114231342905e-06,
+      "loss": 0.3528,
+      "step": 19226
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 6.112739803528263e-06,
+      "loss": 0.6143,
+      "step": 19227
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 6.110168183922235e-06,
+      "loss": 0.8092,
+      "step": 19228
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 6.107596564316206e-06,
+      "loss": 0.6878,
+      "step": 19229
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 6.105024944710179e-06,
+      "loss": 0.5747,
+      "step": 19230
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 6.102453325104151e-06,
+      "loss": 0.5832,
+      "step": 19231
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 6.099881705498123e-06,
+      "loss": 0.4829,
+      "step": 19232
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 6.097310085892095e-06,
+      "loss": 0.684,
+      "step": 19233
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 6.0947384662860674e-06,
+      "loss": 0.4949,
+      "step": 19234
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 6.092166846680039e-06,
+      "loss": 0.6493,
+      "step": 19235
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 6.089595227074012e-06,
+      "loss": 0.4759,
+      "step": 19236
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 6.087023607467983e-06,
+      "loss": 0.506,
+      "step": 19237
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 6.0844519878619555e-06,
+      "loss": 0.5507,
+      "step": 19238
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 6.081880368255928e-06,
+      "loss": 0.4172,
+      "step": 19239
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 6.0793087486499e-06,
+      "loss": 0.5312,
+      "step": 19240
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 6.076737129043872e-06,
+      "loss": 0.7173,
+      "step": 19241
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 6.074165509437844e-06,
+      "loss": 0.4794,
+      "step": 19242
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 6.071593889831817e-06,
+      "loss": 0.5089,
+      "step": 19243
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 6.069022270225788e-06,
+      "loss": 0.8612,
+      "step": 19244
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 6.066450650619761e-06,
+      "loss": 0.7511,
+      "step": 19245
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 6.0638790310137325e-06,
+      "loss": 0.6953,
+      "step": 19246
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 6.061307411407705e-06,
+      "loss": 0.6682,
+      "step": 19247
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 6.058735791801677e-06,
+      "loss": 0.7287,
+      "step": 19248
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 6.056164172195649e-06,
+      "loss": 0.5137,
+      "step": 19249
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 6.053592552589621e-06,
+      "loss": 1.0239,
+      "step": 19250
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 6.051020932983594e-06,
+      "loss": 0.444,
+      "step": 19251
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 6.048449313377565e-06,
+      "loss": 0.4331,
+      "step": 19252
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 6.045877693771538e-06,
+      "loss": 0.5194,
+      "step": 19253
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 6.043306074165509e-06,
+      "loss": 0.5327,
+      "step": 19254
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 6.040734454559482e-06,
+      "loss": 0.7696,
+      "step": 19255
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 6.038162834953454e-06,
+      "loss": 0.6159,
+      "step": 19256
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 6.035591215347426e-06,
+      "loss": 0.9196,
+      "step": 19257
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 6.033019595741398e-06,
+      "loss": 0.5353,
+      "step": 19258
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 6.0304479761353706e-06,
+      "loss": 0.8372,
+      "step": 19259
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 6.027876356529343e-06,
+      "loss": 0.7262,
+      "step": 19260
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 6.025304736923314e-06,
+      "loss": 0.6299,
+      "step": 19261
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 6.022733117317287e-06,
+      "loss": 0.3235,
+      "step": 19262
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 6.020161497711259e-06,
+      "loss": 0.3501,
+      "step": 19263
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 6.017589878105231e-06,
+      "loss": 0.4979,
+      "step": 19264
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 6.015018258499203e-06,
+      "loss": 0.6971,
+      "step": 19265
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 6.012446638893175e-06,
+      "loss": 0.6242,
+      "step": 19266
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 6.0098750192871475e-06,
+      "loss": 0.6035,
+      "step": 19267
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 6.00730339968112e-06,
+      "loss": 0.6957,
+      "step": 19268
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 6.004731780075092e-06,
+      "loss": 0.6553,
+      "step": 19269
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 6.002160160469063e-06,
+      "loss": 0.6588,
+      "step": 19270
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 5.999588540863036e-06,
+      "loss": 0.7072,
+      "step": 19271
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 5.997016921257008e-06,
+      "loss": 0.5701,
+      "step": 19272
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 5.99444530165098e-06,
+      "loss": 0.4678,
+      "step": 19273
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 5.991873682044952e-06,
+      "loss": 0.7664,
+      "step": 19274
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 5.9893020624389245e-06,
+      "loss": 0.6151,
+      "step": 19275
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 5.986730442832896e-06,
+      "loss": 0.9022,
+      "step": 19276
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 5.984158823226869e-06,
+      "loss": 0.3909,
+      "step": 19277
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 5.98158720362084e-06,
+      "loss": 0.4317,
+      "step": 19278
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 5.9790155840148125e-06,
+      "loss": 0.6565,
+      "step": 19279
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 5.976443964408785e-06,
+      "loss": 0.3293,
+      "step": 19280
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 5.973872344802757e-06,
+      "loss": 0.3587,
+      "step": 19281
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 5.971300725196729e-06,
+      "loss": 0.9658,
+      "step": 19282
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 5.9687291055907014e-06,
+      "loss": 0.8179,
+      "step": 19283
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 5.966157485984674e-06,
+      "loss": 0.5059,
+      "step": 19284
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 5.963585866378645e-06,
+      "loss": 0.445,
+      "step": 19285
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 5.961014246772618e-06,
+      "loss": 0.8546,
+      "step": 19286
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 5.9584426271665895e-06,
+      "loss": 0.5175,
+      "step": 19287
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 5.9558710075605626e-06,
+      "loss": 0.7413,
+      "step": 19288
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 5.953299387954534e-06,
+      "loss": 0.5267,
+      "step": 19289
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 5.950727768348506e-06,
+      "loss": 0.5967,
+      "step": 19290
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 5.948156148742478e-06,
+      "loss": 0.4271,
+      "step": 19291
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 5.945584529136451e-06,
+      "loss": 0.6661,
+      "step": 19292
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 5.943012909530422e-06,
+      "loss": 0.6801,
+      "step": 19293
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 5.940441289924395e-06,
+      "loss": 0.6008,
+      "step": 19294
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 5.9378696703183665e-06,
+      "loss": 0.5815,
+      "step": 19295
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 5.935298050712339e-06,
+      "loss": 0.7014,
+      "step": 19296
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 5.932726431106311e-06,
+      "loss": 0.6079,
+      "step": 19297
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 5.930154811500283e-06,
+      "loss": 0.8589,
+      "step": 19298
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 5.927583191894255e-06,
+      "loss": 0.2237,
+      "step": 19299
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 5.925011572288228e-06,
+      "loss": 0.4836,
+      "step": 19300
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 5.9224399526822e-06,
+      "loss": 0.3563,
+      "step": 19301
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 5.919868333076171e-06,
+      "loss": 0.611,
+      "step": 19302
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 5.917296713470144e-06,
+      "loss": 0.6086,
+      "step": 19303
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 5.914725093864116e-06,
+      "loss": 0.592,
+      "step": 19304
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 5.912153474258088e-06,
+      "loss": 0.8324,
+      "step": 19305
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 5.90958185465206e-06,
+      "loss": 0.5492,
+      "step": 19306
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 5.907010235046032e-06,
+      "loss": 0.9504,
+      "step": 19307
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 5.9044386154400046e-06,
+      "loss": 0.4487,
+      "step": 19308
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 5.901866995833977e-06,
+      "loss": 0.7346,
+      "step": 19309
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 5.899295376227949e-06,
+      "loss": 0.2332,
+      "step": 19310
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 5.89672375662192e-06,
+      "loss": 0.5565,
+      "step": 19311
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 5.8941521370158935e-06,
+      "loss": 0.476,
+      "step": 19312
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 5.891580517409865e-06,
+      "loss": 0.4808,
+      "step": 19313
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 5.889008897803837e-06,
+      "loss": 0.7424,
+      "step": 19314
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 5.886437278197809e-06,
+      "loss": 0.647,
+      "step": 19315
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 5.8838656585917815e-06,
+      "loss": 0.2878,
+      "step": 19316
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 5.881294038985753e-06,
+      "loss": 0.554,
+      "step": 19317
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 5.878722419379726e-06,
+      "loss": 0.4135,
+      "step": 19318
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 5.876150799773697e-06,
+      "loss": 0.4891,
+      "step": 19319
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 5.87357918016767e-06,
+      "loss": 0.4193,
+      "step": 19320
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 5.871007560561642e-06,
+      "loss": 0.5204,
+      "step": 19321
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 5.868435940955614e-06,
+      "loss": 0.6597,
+      "step": 19322
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 5.865864321349586e-06,
+      "loss": 0.4861,
+      "step": 19323
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 5.8632927017435585e-06,
+      "loss": 0.8408,
+      "step": 19324
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 5.860721082137531e-06,
+      "loss": 0.6223,
+      "step": 19325
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 5.858149462531503e-06,
+      "loss": 1.1034,
+      "step": 19326
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 5.855577842925475e-06,
+      "loss": 0.5338,
+      "step": 19327
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 5.8530062233194465e-06,
+      "loss": 0.6625,
+      "step": 19328
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 5.85043460371342e-06,
+      "loss": 0.4368,
+      "step": 19329
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 5.847862984107391e-06,
+      "loss": 0.3312,
+      "step": 19330
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 5.845291364501363e-06,
+      "loss": 0.6046,
+      "step": 19331
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 5.8427197448953355e-06,
+      "loss": 0.3876,
+      "step": 19332
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 5.840148125289308e-06,
+      "loss": 0.6638,
+      "step": 19333
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 5.837576505683279e-06,
+      "loss": 1.01,
+      "step": 19334
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 5.835004886077252e-06,
+      "loss": 0.6654,
+      "step": 19335
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 5.8324332664712235e-06,
+      "loss": 0.611,
+      "step": 19336
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 5.829861646865196e-06,
+      "loss": 0.7334,
+      "step": 19337
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 5.827290027259168e-06,
+      "loss": 0.6429,
+      "step": 19338
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 5.82471840765314e-06,
+      "loss": 0.512,
+      "step": 19339
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 5.822146788047112e-06,
+      "loss": 0.7754,
+      "step": 19340
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 5.819575168441085e-06,
+      "loss": 0.6679,
+      "step": 19341
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 5.817003548835057e-06,
+      "loss": 0.9323,
+      "step": 19342
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 5.814431929229028e-06,
+      "loss": 0.8046,
+      "step": 19343
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 5.811860309623001e-06,
+      "loss": 0.6795,
+      "step": 19344
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 5.809288690016973e-06,
+      "loss": 0.8781,
+      "step": 19345
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 5.806717070410945e-06,
+      "loss": 0.7264,
+      "step": 19346
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 5.804145450804917e-06,
+      "loss": 0.7613,
+      "step": 19347
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 5.801573831198889e-06,
+      "loss": 0.5794,
+      "step": 19348
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 5.799002211592862e-06,
+      "loss": 0.6281,
+      "step": 19349
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 5.796430591986834e-06,
+      "loss": 0.5319,
+      "step": 19350
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 5.793858972380806e-06,
+      "loss": 0.3876,
+      "step": 19351
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 5.7912873527747774e-06,
+      "loss": 0.5024,
+      "step": 19352
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 5.7887157331687505e-06,
+      "loss": 0.3862,
+      "step": 19353
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 5.786144113562722e-06,
+      "loss": 0.6075,
+      "step": 19354
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 5.783572493956694e-06,
+      "loss": 0.3727,
+      "step": 19355
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 5.781000874350666e-06,
+      "loss": 0.5115,
+      "step": 19356
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 5.7784292547446386e-06,
+      "loss": 0.6543,
+      "step": 19357
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 5.77585763513861e-06,
+      "loss": 0.2754,
+      "step": 19358
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 5.773286015532583e-06,
+      "loss": 0.5406,
+      "step": 19359
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 5.770714395926554e-06,
+      "loss": 0.6978,
+      "step": 19360
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 5.768142776320527e-06,
+      "loss": 0.5405,
+      "step": 19361
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 5.765571156714499e-06,
+      "loss": 0.6075,
+      "step": 19362
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 5.762999537108471e-06,
+      "loss": 0.5571,
+      "step": 19363
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 5.760427917502443e-06,
+      "loss": 1.0423,
+      "step": 19364
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 5.7578562978964155e-06,
+      "loss": 0.3379,
+      "step": 19365
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 5.755284678290388e-06,
+      "loss": 0.6313,
+      "step": 19366
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 5.75271305868436e-06,
+      "loss": 1.0692,
+      "step": 19367
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 5.750141439078332e-06,
+      "loss": 0.415,
+      "step": 19368
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 5.747569819472304e-06,
+      "loss": 0.5643,
+      "step": 19369
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 5.744998199866277e-06,
+      "loss": 0.763,
+      "step": 19370
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 5.742426580260248e-06,
+      "loss": 0.6657,
+      "step": 19371
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 5.73985496065422e-06,
+      "loss": 0.5526,
+      "step": 19372
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 5.7372833410481925e-06,
+      "loss": 0.6087,
+      "step": 19373
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 5.734711721442165e-06,
+      "loss": 0.3079,
+      "step": 19374
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 5.732140101836136e-06,
+      "loss": 0.6393,
+      "step": 19375
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 5.729568482230109e-06,
+      "loss": 0.509,
+      "step": 19376
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 5.7269968626240806e-06,
+      "loss": 0.5233,
+      "step": 19377
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 5.724425243018053e-06,
+      "loss": 0.7715,
+      "step": 19378
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 5.721853623412025e-06,
+      "loss": 0.7613,
+      "step": 19379
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 5.719282003805997e-06,
+      "loss": 0.6343,
+      "step": 19380
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 5.7167103841999695e-06,
+      "loss": 0.6314,
+      "step": 19381
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 5.714138764593942e-06,
+      "loss": 0.4834,
+      "step": 19382
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 5.711567144987914e-06,
+      "loss": 0.6733,
+      "step": 19383
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 5.708995525381885e-06,
+      "loss": 0.6102,
+      "step": 19384
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 5.706423905775858e-06,
+      "loss": 0.6372,
+      "step": 19385
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 5.70385228616983e-06,
+      "loss": 0.557,
+      "step": 19386
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 5.701280666563802e-06,
+      "loss": 0.6178,
+      "step": 19387
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 5.698709046957774e-06,
+      "loss": 0.3985,
+      "step": 19388
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 5.6961374273517464e-06,
+      "loss": 0.2893,
+      "step": 19389
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 5.693565807745719e-06,
+      "loss": 0.8949,
+      "step": 19390
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 5.690994188139691e-06,
+      "loss": 0.6273,
+      "step": 19391
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 5.688422568533663e-06,
+      "loss": 0.5429,
+      "step": 19392
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 5.6858509489276345e-06,
+      "loss": 0.615,
+      "step": 19393
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 5.6832793293216076e-06,
+      "loss": 0.5455,
+      "step": 19394
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 5.680707709715579e-06,
+      "loss": 0.8334,
+      "step": 19395
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 5.678136090109551e-06,
+      "loss": 0.6165,
+      "step": 19396
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 5.675564470503523e-06,
+      "loss": 0.6481,
+      "step": 19397
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 5.672992850897496e-06,
+      "loss": 0.6427,
+      "step": 19398
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 5.670421231291467e-06,
+      "loss": 0.4418,
+      "step": 19399
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 5.66784961168544e-06,
+      "loss": 0.4298,
+      "step": 19400
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 5.6652779920794114e-06,
+      "loss": 0.7098,
+      "step": 19401
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 5.6627063724733845e-06,
+      "loss": 0.4621,
+      "step": 19402
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 5.660134752867356e-06,
+      "loss": 0.6579,
+      "step": 19403
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 5.657563133261328e-06,
+      "loss": 0.7971,
+      "step": 19404
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 5.6549915136553e-06,
+      "loss": 0.6233,
+      "step": 19405
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 5.652419894049273e-06,
+      "loss": 0.442,
+      "step": 19406
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 5.649848274443245e-06,
+      "loss": 0.4859,
+      "step": 19407
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 5.647276654837217e-06,
+      "loss": 0.4206,
+      "step": 19408
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 5.644705035231189e-06,
+      "loss": 0.655,
+      "step": 19409
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 5.642133415625161e-06,
+      "loss": 0.919,
+      "step": 19410
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 5.639561796019134e-06,
+      "loss": 0.4803,
+      "step": 19411
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 5.636990176413105e-06,
+      "loss": 0.6828,
+      "step": 19412
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 5.634418556807077e-06,
+      "loss": 0.422,
+      "step": 19413
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 5.6318469372010495e-06,
+      "loss": 0.3971,
+      "step": 19414
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 5.629275317595022e-06,
+      "loss": 0.5254,
+      "step": 19415
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 5.626703697988993e-06,
+      "loss": 0.4107,
+      "step": 19416
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 5.624132078382966e-06,
+      "loss": 0.3082,
+      "step": 19417
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 5.621560458776938e-06,
+      "loss": 0.5837,
+      "step": 19418
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 5.61898883917091e-06,
+      "loss": 0.5841,
+      "step": 19419
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 5.616417219564882e-06,
+      "loss": 0.4683,
+      "step": 19420
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 5.613845599958854e-06,
+      "loss": 0.457,
+      "step": 19421
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 5.6112739803528265e-06,
+      "loss": 0.5658,
+      "step": 19422
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 5.608702360746799e-06,
+      "loss": 0.4253,
+      "step": 19423
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 5.606130741140771e-06,
+      "loss": 0.6156,
+      "step": 19424
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 5.603559121534742e-06,
+      "loss": 0.4701,
+      "step": 19425
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 5.600987501928715e-06,
+      "loss": 0.8417,
+      "step": 19426
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 5.598415882322687e-06,
+      "loss": 0.4942,
+      "step": 19427
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 5.595844262716659e-06,
+      "loss": 0.3348,
+      "step": 19428
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 5.593272643110631e-06,
+      "loss": 0.6686,
+      "step": 19429
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 5.5907010235046035e-06,
+      "loss": 0.9191,
+      "step": 19430
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 5.588129403898576e-06,
+      "loss": 0.6532,
+      "step": 19431
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 5.585557784292548e-06,
+      "loss": 0.3565,
+      "step": 19432
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 5.58298616468652e-06,
+      "loss": 0.6938,
+      "step": 19433
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 5.5804145450804915e-06,
+      "loss": 0.3781,
+      "step": 19434
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 5.577842925474465e-06,
+      "loss": 0.1136,
+      "step": 19435
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 5.575271305868436e-06,
+      "loss": 0.8047,
+      "step": 19436
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 5.572699686262408e-06,
+      "loss": 0.7466,
+      "step": 19437
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 5.5701280666563804e-06,
+      "loss": 0.9152,
+      "step": 19438
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 5.567556447050353e-06,
+      "loss": 0.7672,
+      "step": 19439
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 5.564984827444325e-06,
+      "loss": 0.2518,
+      "step": 19440
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 5.562413207838297e-06,
+      "loss": 0.6988,
+      "step": 19441
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 5.5598415882322685e-06,
+      "loss": 0.4877,
+      "step": 19442
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 5.5572699686262416e-06,
+      "loss": 0.9098,
+      "step": 19443
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 5.554698349020213e-06,
+      "loss": 0.946,
+      "step": 19444
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 5.552126729414185e-06,
+      "loss": 0.5397,
+      "step": 19445
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 5.549555109808157e-06,
+      "loss": 0.5977,
+      "step": 19446
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 5.54698349020213e-06,
+      "loss": 0.5427,
+      "step": 19447
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 5.544411870596102e-06,
+      "loss": 0.6039,
+      "step": 19448
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 5.541840250990074e-06,
+      "loss": 0.6275,
+      "step": 19449
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 5.539268631384046e-06,
+      "loss": 0.4733,
+      "step": 19450
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 5.536697011778018e-06,
+      "loss": 0.431,
+      "step": 19451
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 5.534125392171991e-06,
+      "loss": 0.2259,
+      "step": 19452
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 5.531553772565962e-06,
+      "loss": 0.3844,
+      "step": 19453
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 5.528982152959934e-06,
+      "loss": 0.4379,
+      "step": 19454
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 5.526410533353907e-06,
+      "loss": 0.2634,
+      "step": 19455
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 5.523838913747879e-06,
+      "loss": 0.585,
+      "step": 19456
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 5.52126729414185e-06,
+      "loss": 0.5759,
+      "step": 19457
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 5.518695674535823e-06,
+      "loss": 0.4535,
+      "step": 19458
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 5.516124054929795e-06,
+      "loss": 0.2364,
+      "step": 19459
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 5.513552435323767e-06,
+      "loss": 0.4392,
+      "step": 19460
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 5.510980815717739e-06,
+      "loss": 0.3595,
+      "step": 19461
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 5.508409196111711e-06,
+      "loss": 0.4578,
+      "step": 19462
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 5.5058375765056836e-06,
+      "loss": 0.7783,
+      "step": 19463
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 5.503265956899656e-06,
+      "loss": 0.7858,
+      "step": 19464
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 5.500694337293628e-06,
+      "loss": 0.3831,
+      "step": 19465
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 5.498122717687599e-06,
+      "loss": 0.8549,
+      "step": 19466
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 5.4955510980815725e-06,
+      "loss": 0.6196,
+      "step": 19467
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 5.492979478475544e-06,
+      "loss": 0.5873,
+      "step": 19468
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 5.490407858869516e-06,
+      "loss": 0.5564,
+      "step": 19469
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 5.487836239263488e-06,
+      "loss": 0.8937,
+      "step": 19470
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 5.4852646196574605e-06,
+      "loss": 0.5114,
+      "step": 19471
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 5.482693000051433e-06,
+      "loss": 0.3275,
+      "step": 19472
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 5.480121380445405e-06,
+      "loss": 0.6609,
+      "step": 19473
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 5.477549760839377e-06,
+      "loss": 0.343,
+      "step": 19474
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 5.474978141233349e-06,
+      "loss": 0.422,
+      "step": 19475
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 5.472406521627322e-06,
+      "loss": 0.6046,
+      "step": 19476
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 5.469834902021293e-06,
+      "loss": 0.6291,
+      "step": 19477
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 5.467263282415266e-06,
+      "loss": 0.4548,
+      "step": 19478
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 5.4646916628092375e-06,
+      "loss": 0.2465,
+      "step": 19479
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 5.46212004320321e-06,
+      "loss": 0.3035,
+      "step": 19480
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 5.459548423597182e-06,
+      "loss": 0.3325,
+      "step": 19481
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 5.456976803991154e-06,
+      "loss": 0.4049,
+      "step": 19482
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 5.4544051843851255e-06,
+      "loss": 0.5039,
+      "step": 19483
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 5.451833564779099e-06,
+      "loss": 0.4189,
+      "step": 19484
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 5.44926194517307e-06,
+      "loss": 0.8129,
+      "step": 19485
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 5.446690325567042e-06,
+      "loss": 0.2564,
+      "step": 19486
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 5.4441187059610144e-06,
+      "loss": 0.8246,
+      "step": 19487
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 5.441547086354987e-06,
+      "loss": 0.6082,
+      "step": 19488
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 5.438975466748959e-06,
+      "loss": 0.4005,
+      "step": 19489
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 5.436403847142931e-06,
+      "loss": 0.557,
+      "step": 19490
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 5.433832227536903e-06,
+      "loss": 0.61,
+      "step": 19491
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 5.431260607930875e-06,
+      "loss": 0.5692,
+      "step": 19492
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 5.428688988324848e-06,
+      "loss": 0.7311,
+      "step": 19493
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 5.426117368718819e-06,
+      "loss": 0.8879,
+      "step": 19494
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 5.423545749112791e-06,
+      "loss": 0.6795,
+      "step": 19495
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 5.420974129506764e-06,
+      "loss": 1.0238,
+      "step": 19496
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 5.418402509900736e-06,
+      "loss": 0.3884,
+      "step": 19497
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 5.415830890294707e-06,
+      "loss": 0.6112,
+      "step": 19498
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 5.41325927068868e-06,
+      "loss": 0.5066,
+      "step": 19499
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 5.410687651082652e-06,
+      "loss": 0.9476,
+      "step": 19500
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 5.408116031476624e-06,
+      "loss": 0.5581,
+      "step": 19501
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 5.405544411870596e-06,
+      "loss": 0.5407,
+      "step": 19502
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 5.402972792264568e-06,
+      "loss": 0.5668,
+      "step": 19503
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 5.400401172658541e-06,
+      "loss": 0.5399,
+      "step": 19504
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 5.397829553052513e-06,
+      "loss": 0.5753,
+      "step": 19505
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 5.395257933446485e-06,
+      "loss": 0.6361,
+      "step": 19506
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 5.3926863138404564e-06,
+      "loss": 0.5135,
+      "step": 19507
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 5.3901146942344295e-06,
+      "loss": 0.6557,
+      "step": 19508
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 5.387543074628401e-06,
+      "loss": 0.4468,
+      "step": 19509
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 5.384971455022373e-06,
+      "loss": 1.0399,
+      "step": 19510
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 5.382399835416345e-06,
+      "loss": 0.6572,
+      "step": 19511
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 5.3798282158103176e-06,
+      "loss": 0.6617,
+      "step": 19512
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 5.37725659620429e-06,
+      "loss": 0.6288,
+      "step": 19513
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 5.374684976598262e-06,
+      "loss": 0.6091,
+      "step": 19514
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 5.372113356992234e-06,
+      "loss": 0.5884,
+      "step": 19515
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 5.3695417373862065e-06,
+      "loss": 0.3184,
+      "step": 19516
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 5.366970117780179e-06,
+      "loss": 0.7683,
+      "step": 19517
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 5.36439849817415e-06,
+      "loss": 0.3443,
+      "step": 19518
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 5.361826878568123e-06,
+      "loss": 0.5424,
+      "step": 19519
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 5.3592552589620945e-06,
+      "loss": 0.5125,
+      "step": 19520
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 5.356683639356067e-06,
+      "loss": 0.5195,
+      "step": 19521
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 5.354112019750039e-06,
+      "loss": 0.7623,
+      "step": 19522
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 5.351540400144011e-06,
+      "loss": 0.4462,
+      "step": 19523
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 5.348968780537983e-06,
+      "loss": 0.6537,
+      "step": 19524
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 5.346397160931956e-06,
+      "loss": 0.3991,
+      "step": 19525
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 5.343825541325927e-06,
+      "loss": 0.644,
+      "step": 19526
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 5.341253921719899e-06,
+      "loss": 0.5852,
+      "step": 19527
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 5.3386823021138715e-06,
+      "loss": 0.3978,
+      "step": 19528
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 5.336110682507844e-06,
+      "loss": 0.4058,
+      "step": 19529
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 5.333539062901816e-06,
+      "loss": 0.9444,
+      "step": 19530
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 5.330967443295788e-06,
+      "loss": 0.8197,
+      "step": 19531
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 5.32839582368976e-06,
+      "loss": 0.5153,
+      "step": 19532
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 5.325824204083732e-06,
+      "loss": 0.5243,
+      "step": 19533
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 5.323252584477705e-06,
+      "loss": 0.6582,
+      "step": 19534
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 5.320680964871676e-06,
+      "loss": 0.5913,
+      "step": 19535
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 5.3181093452656485e-06,
+      "loss": 0.5126,
+      "step": 19536
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 5.315537725659621e-06,
+      "loss": 0.726,
+      "step": 19537
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 5.312966106053593e-06,
+      "loss": 0.4736,
+      "step": 19538
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 5.310394486447564e-06,
+      "loss": 0.3777,
+      "step": 19539
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 5.307822866841537e-06,
+      "loss": 0.5195,
+      "step": 19540
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 5.305251247235509e-06,
+      "loss": 0.4328,
+      "step": 19541
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 5.302679627629481e-06,
+      "loss": 0.6142,
+      "step": 19542
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 5.300108008023453e-06,
+      "loss": 0.4246,
+      "step": 19543
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 5.297536388417425e-06,
+      "loss": 0.5457,
+      "step": 19544
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 5.294964768811398e-06,
+      "loss": 0.5818,
+      "step": 19545
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 5.29239314920537e-06,
+      "loss": 0.5011,
+      "step": 19546
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 5.289821529599342e-06,
+      "loss": 0.5767,
+      "step": 19547
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 5.2872499099933135e-06,
+      "loss": 0.6075,
+      "step": 19548
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 5.2846782903872866e-06,
+      "loss": 0.7091,
+      "step": 19549
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 5.282106670781258e-06,
+      "loss": 0.6819,
+      "step": 19550
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 5.279535051175231e-06,
+      "loss": 1.0821,
+      "step": 19551
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 5.276963431569202e-06,
+      "loss": 0.3272,
+      "step": 19552
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 5.274391811963175e-06,
+      "loss": 0.5627,
+      "step": 19553
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 5.271820192357147e-06,
+      "loss": 0.4749,
+      "step": 19554
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 5.269248572751119e-06,
+      "loss": 0.444,
+      "step": 19555
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 5.266676953145091e-06,
+      "loss": 0.7068,
+      "step": 19556
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 5.2641053335390635e-06,
+      "loss": 0.6566,
+      "step": 19557
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 5.261533713933036e-06,
+      "loss": 0.8012,
+      "step": 19558
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 5.258962094327007e-06,
+      "loss": 0.4065,
+      "step": 19559
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 5.25639047472098e-06,
+      "loss": 0.3862,
+      "step": 19560
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 5.2538188551149516e-06,
+      "loss": 0.3822,
+      "step": 19561
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 5.251247235508924e-06,
+      "loss": 0.5209,
+      "step": 19562
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 5.248675615902896e-06,
+      "loss": 0.5467,
+      "step": 19563
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 5.246103996296868e-06,
+      "loss": 0.7046,
+      "step": 19564
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 5.24353237669084e-06,
+      "loss": 0.7496,
+      "step": 19565
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 5.240960757084813e-06,
+      "loss": 0.4876,
+      "step": 19566
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 5.238389137478784e-06,
+      "loss": 0.2802,
+      "step": 19567
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 5.235817517872756e-06,
+      "loss": 0.5505,
+      "step": 19568
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 5.2332458982667285e-06,
+      "loss": 0.5487,
+      "step": 19569
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 5.230674278660701e-06,
+      "loss": 0.7509,
+      "step": 19570
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 5.228102659054673e-06,
+      "loss": 0.228,
+      "step": 19571
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 5.225531039448645e-06,
+      "loss": 0.5329,
+      "step": 19572
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 5.2229594198426174e-06,
+      "loss": 0.7698,
+      "step": 19573
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 5.220387800236589e-06,
+      "loss": 0.6771,
+      "step": 19574
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 5.217816180630562e-06,
+      "loss": 0.5758,
+      "step": 19575
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 5.215244561024533e-06,
+      "loss": 0.7325,
+      "step": 19576
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 5.2126729414185055e-06,
+      "loss": 0.3456,
+      "step": 19577
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 5.210101321812478e-06,
+      "loss": 0.2456,
+      "step": 19578
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 5.20752970220645e-06,
+      "loss": 0.5609,
+      "step": 19579
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 5.204958082600421e-06,
+      "loss": 0.3828,
+      "step": 19580
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 5.202386462994394e-06,
+      "loss": 0.6912,
+      "step": 19581
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 5.199814843388366e-06,
+      "loss": 0.409,
+      "step": 19582
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 5.197243223782338e-06,
+      "loss": 0.558,
+      "step": 19583
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 5.19467160417631e-06,
+      "loss": 0.819,
+      "step": 19584
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 5.1920999845702825e-06,
+      "loss": 0.292,
+      "step": 19585
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 5.189528364964255e-06,
+      "loss": 0.5956,
+      "step": 19586
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 5.186956745358227e-06,
+      "loss": 0.7251,
+      "step": 19587
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 5.184385125752199e-06,
+      "loss": 0.6105,
+      "step": 19588
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 5.181813506146171e-06,
+      "loss": 0.4286,
+      "step": 19589
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 5.179241886540144e-06,
+      "loss": 0.9564,
+      "step": 19590
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 5.176670266934115e-06,
+      "loss": 0.8229,
+      "step": 19591
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 5.174098647328088e-06,
+      "loss": 0.7167,
+      "step": 19592
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 5.1715270277220594e-06,
+      "loss": 0.5073,
+      "step": 19593
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 5.168955408116032e-06,
+      "loss": 0.4219,
+      "step": 19594
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 5.166383788510004e-06,
+      "loss": 0.8356,
+      "step": 19595
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 5.163812168903976e-06,
+      "loss": 0.8579,
+      "step": 19596
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 5.161240549297948e-06,
+      "loss": 0.5234,
+      "step": 19597
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 5.1586689296919206e-06,
+      "loss": 0.8679,
+      "step": 19598
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 5.156097310085893e-06,
+      "loss": 0.5009,
+      "step": 19599
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 5.153525690479864e-06,
+      "loss": 0.8572,
+      "step": 19600
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 5.150954070873837e-06,
+      "loss": 0.6186,
+      "step": 19601
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 5.148382451267809e-06,
+      "loss": 0.7093,
+      "step": 19602
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 5.145810831661781e-06,
+      "loss": 0.4864,
+      "step": 19603
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 5.143239212055753e-06,
+      "loss": 0.4911,
+      "step": 19604
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 5.140667592449725e-06,
+      "loss": 0.4457,
+      "step": 19605
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 5.138095972843697e-06,
+      "loss": 0.7311,
+      "step": 19606
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 5.13552435323767e-06,
+      "loss": 0.5864,
+      "step": 19607
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 5.132952733631641e-06,
+      "loss": 0.4972,
+      "step": 19608
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 5.130381114025613e-06,
+      "loss": 0.4647,
+      "step": 19609
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 5.127809494419586e-06,
+      "loss": 0.6742,
+      "step": 19610
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 5.125237874813558e-06,
+      "loss": 0.3171,
+      "step": 19611
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 5.12266625520753e-06,
+      "loss": 0.5934,
+      "step": 19612
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 5.120094635601502e-06,
+      "loss": 0.7037,
+      "step": 19613
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 5.1175230159954745e-06,
+      "loss": 0.3256,
+      "step": 19614
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 5.114951396389446e-06,
+      "loss": 0.2881,
+      "step": 19615
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 5.112379776783419e-06,
+      "loss": 0.6852,
+      "step": 19616
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 5.10980815717739e-06,
+      "loss": 0.5892,
+      "step": 19617
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 5.1072365375713626e-06,
+      "loss": 0.7778,
+      "step": 19618
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 5.104664917965335e-06,
+      "loss": 0.7234,
+      "step": 19619
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 5.102093298359307e-06,
+      "loss": 0.7228,
+      "step": 19620
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 5.099521678753278e-06,
+      "loss": 0.9666,
+      "step": 19621
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 5.0969500591472515e-06,
+      "loss": 0.4711,
+      "step": 19622
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 5.094378439541223e-06,
+      "loss": 0.8187,
+      "step": 19623
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 5.091806819935195e-06,
+      "loss": 1.0335,
+      "step": 19624
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 5.089235200329167e-06,
+      "loss": 0.52,
+      "step": 19625
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 5.0866635807231395e-06,
+      "loss": 0.5388,
+      "step": 19626
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 5.084091961117112e-06,
+      "loss": 0.6268,
+      "step": 19627
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 5.081520341511084e-06,
+      "loss": 0.8042,
+      "step": 19628
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 5.078948721905056e-06,
+      "loss": 0.9148,
+      "step": 19629
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 5.076377102299028e-06,
+      "loss": 0.5396,
+      "step": 19630
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 5.073805482693001e-06,
+      "loss": 0.5228,
+      "step": 19631
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 5.071233863086972e-06,
+      "loss": 0.6195,
+      "step": 19632
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 5.068662243480945e-06,
+      "loss": 0.5548,
+      "step": 19633
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 5.0660906238749165e-06,
+      "loss": 0.6598,
+      "step": 19634
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 5.063519004268889e-06,
+      "loss": 0.7226,
+      "step": 19635
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 5.060947384662861e-06,
+      "loss": 0.546,
+      "step": 19636
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 5.058375765056833e-06,
+      "loss": 0.4455,
+      "step": 19637
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 5.055804145450805e-06,
+      "loss": 0.5451,
+      "step": 19638
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 5.053232525844778e-06,
+      "loss": 0.5266,
+      "step": 19639
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 5.05066090623875e-06,
+      "loss": 0.5584,
+      "step": 19640
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 5.048089286632721e-06,
+      "loss": 0.7483,
+      "step": 19641
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 5.045517667026694e-06,
+      "loss": 0.4897,
+      "step": 19642
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 5.042946047420666e-06,
+      "loss": 0.5629,
+      "step": 19643
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 5.040374427814638e-06,
+      "loss": 0.7831,
+      "step": 19644
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 5.03780280820861e-06,
+      "loss": 0.7756,
+      "step": 19645
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 5.035231188602582e-06,
+      "loss": 0.4725,
+      "step": 19646
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 5.032659568996554e-06,
+      "loss": 0.5851,
+      "step": 19647
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 5.030087949390527e-06,
+      "loss": 0.9748,
+      "step": 19648
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 5.027516329784498e-06,
+      "loss": 0.3655,
+      "step": 19649
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 5.02494471017847e-06,
+      "loss": 0.5327,
+      "step": 19650
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 5.022373090572443e-06,
+      "loss": 0.832,
+      "step": 19651
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 5.019801470966415e-06,
+      "loss": 0.5046,
+      "step": 19652
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 5.017229851360387e-06,
+      "loss": 0.6174,
+      "step": 19653
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 5.014658231754359e-06,
+      "loss": 0.8993,
+      "step": 19654
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 5.0120866121483315e-06,
+      "loss": 0.5923,
+      "step": 19655
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 5.009514992542303e-06,
+      "loss": 0.5726,
+      "step": 19656
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 5.006943372936276e-06,
+      "loss": 0.6665,
+      "step": 19657
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 5.004371753330247e-06,
+      "loss": 0.8592,
+      "step": 19658
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 5.00180013372422e-06,
+      "loss": 0.8124,
+      "step": 19659
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 4.999228514118192e-06,
+      "loss": 0.5503,
+      "step": 19660
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 4.996656894512164e-06,
+      "loss": 0.7075,
+      "step": 19661
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 4.9940852749061354e-06,
+      "loss": 0.55,
+      "step": 19662
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 4.9915136553001085e-06,
+      "loss": 0.3839,
+      "step": 19663
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 4.98894203569408e-06,
+      "loss": 0.6258,
+      "step": 19664
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 4.986370416088053e-06,
+      "loss": 0.7217,
+      "step": 19665
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 4.983798796482024e-06,
+      "loss": 0.6321,
+      "step": 19666
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 4.9812271768759966e-06,
+      "loss": 0.6208,
+      "step": 19667
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 4.978655557269969e-06,
+      "loss": 0.5778,
+      "step": 19668
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 4.976083937663941e-06,
+      "loss": 0.5547,
+      "step": 19669
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 4.973512318057913e-06,
+      "loss": 0.8807,
+      "step": 19670
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 4.9709406984518855e-06,
+      "loss": 0.9526,
+      "step": 19671
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 4.968369078845858e-06,
+      "loss": 0.3334,
+      "step": 19672
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 4.965797459239829e-06,
+      "loss": 0.5702,
+      "step": 19673
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 4.963225839633802e-06,
+      "loss": 0.701,
+      "step": 19674
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 4.9606542200277735e-06,
+      "loss": 0.7741,
+      "step": 19675
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 4.958082600421746e-06,
+      "loss": 0.4921,
+      "step": 19676
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 4.955510980815718e-06,
+      "loss": 0.7546,
+      "step": 19677
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 4.95293936120969e-06,
+      "loss": 0.5938,
+      "step": 19678
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 4.9503677416036624e-06,
+      "loss": 0.6769,
+      "step": 19679
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 4.947796121997635e-06,
+      "loss": 0.4472,
+      "step": 19680
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 4.945224502391607e-06,
+      "loss": 0.6392,
+      "step": 19681
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 4.942652882785578e-06,
+      "loss": 0.6701,
+      "step": 19682
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 4.940081263179551e-06,
+      "loss": 0.3819,
+      "step": 19683
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 4.937509643573523e-06,
+      "loss": 0.7375,
+      "step": 19684
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 4.934938023967495e-06,
+      "loss": 0.5327,
+      "step": 19685
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 4.932366404361467e-06,
+      "loss": 0.6183,
+      "step": 19686
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 4.929794784755439e-06,
+      "loss": 0.5295,
+      "step": 19687
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 4.927223165149411e-06,
+      "loss": 0.3591,
+      "step": 19688
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 4.924651545543384e-06,
+      "loss": 0.5667,
+      "step": 19689
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 4.922079925937355e-06,
+      "loss": 0.2748,
+      "step": 19690
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 4.9195083063313275e-06,
+      "loss": 0.4484,
+      "step": 19691
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 4.9169366867253e-06,
+      "loss": 0.7,
+      "step": 19692
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 4.914365067119272e-06,
+      "loss": 0.5918,
+      "step": 19693
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 4.911793447513244e-06,
+      "loss": 0.5261,
+      "step": 19694
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 4.909221827907216e-06,
+      "loss": 0.67,
+      "step": 19695
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 4.906650208301189e-06,
+      "loss": 0.7117,
+      "step": 19696
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 4.90407858869516e-06,
+      "loss": 0.7283,
+      "step": 19697
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 4.901506969089133e-06,
+      "loss": 0.4258,
+      "step": 19698
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 4.898935349483104e-06,
+      "loss": 0.3715,
+      "step": 19699
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 4.896363729877077e-06,
+      "loss": 0.4579,
+      "step": 19700
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 4.893792110271049e-06,
+      "loss": 0.564,
+      "step": 19701
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 4.891220490665021e-06,
+      "loss": 0.5438,
+      "step": 19702
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 4.888648871058993e-06,
+      "loss": 0.3815,
+      "step": 19703
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 4.8860772514529655e-06,
+      "loss": 0.6012,
+      "step": 19704
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 4.883505631846937e-06,
+      "loss": 0.3561,
+      "step": 19705
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 4.88093401224091e-06,
+      "loss": 0.6311,
+      "step": 19706
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 4.878362392634881e-06,
+      "loss": 0.4964,
+      "step": 19707
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 4.875790773028854e-06,
+      "loss": 0.4486,
+      "step": 19708
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 4.873219153422826e-06,
+      "loss": 0.5352,
+      "step": 19709
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 4.870647533816798e-06,
+      "loss": 0.6639,
+      "step": 19710
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 4.86807591421077e-06,
+      "loss": 0.3292,
+      "step": 19711
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 4.8655042946047425e-06,
+      "loss": 0.8664,
+      "step": 19712
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 4.862932674998715e-06,
+      "loss": 0.7004,
+      "step": 19713
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 4.860361055392686e-06,
+      "loss": 0.4866,
+      "step": 19714
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 4.857789435786659e-06,
+      "loss": 0.4066,
+      "step": 19715
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 4.8552178161806306e-06,
+      "loss": 0.514,
+      "step": 19716
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 4.852646196574603e-06,
+      "loss": 0.9536,
+      "step": 19717
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 4.850074576968575e-06,
+      "loss": 0.5313,
+      "step": 19718
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 4.847502957362547e-06,
+      "loss": 0.5013,
+      "step": 19719
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 4.8449313377565195e-06,
+      "loss": 0.4503,
+      "step": 19720
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 4.842359718150492e-06,
+      "loss": 0.5527,
+      "step": 19721
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 4.839788098544464e-06,
+      "loss": 0.5258,
+      "step": 19722
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 4.837216478938435e-06,
+      "loss": 0.8258,
+      "step": 19723
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 4.834644859332408e-06,
+      "loss": 0.3895,
+      "step": 19724
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 4.83207323972638e-06,
+      "loss": 0.4055,
+      "step": 19725
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 4.829501620120352e-06,
+      "loss": 0.4241,
+      "step": 19726
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 4.826930000514324e-06,
+      "loss": 0.4415,
+      "step": 19727
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 4.8243583809082964e-06,
+      "loss": 0.5675,
+      "step": 19728
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 4.821786761302268e-06,
+      "loss": 0.5581,
+      "step": 19729
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 4.819215141696241e-06,
+      "loss": 0.8312,
+      "step": 19730
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 4.816643522090212e-06,
+      "loss": 0.4959,
+      "step": 19731
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 4.8140719024841845e-06,
+      "loss": 0.4473,
+      "step": 19732
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 4.811500282878157e-06,
+      "loss": 0.4486,
+      "step": 19733
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 4.808928663272129e-06,
+      "loss": 0.4815,
+      "step": 19734
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 4.806357043666101e-06,
+      "loss": 0.4019,
+      "step": 19735
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 4.803785424060073e-06,
+      "loss": 0.6755,
+      "step": 19736
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 4.801213804454046e-06,
+      "loss": 0.4523,
+      "step": 19737
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 4.798642184848017e-06,
+      "loss": 0.5446,
+      "step": 19738
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 4.79607056524199e-06,
+      "loss": 0.604,
+      "step": 19739
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 4.7934989456359615e-06,
+      "loss": 0.753,
+      "step": 19740
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 4.7909273260299345e-06,
+      "loss": 0.6886,
+      "step": 19741
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 4.788355706423906e-06,
+      "loss": 0.5808,
+      "step": 19742
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 4.785784086817878e-06,
+      "loss": 0.5138,
+      "step": 19743
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 4.78321246721185e-06,
+      "loss": 0.6753,
+      "step": 19744
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 4.780640847605823e-06,
+      "loss": 0.711,
+      "step": 19745
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 4.778069227999794e-06,
+      "loss": 0.6601,
+      "step": 19746
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 4.775497608393767e-06,
+      "loss": 0.7964,
+      "step": 19747
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 4.7729259887877384e-06,
+      "loss": 0.5432,
+      "step": 19748
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 4.770354369181711e-06,
+      "loss": 0.6391,
+      "step": 19749
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 4.767782749575683e-06,
+      "loss": 0.4138,
+      "step": 19750
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 4.765211129969655e-06,
+      "loss": 0.4512,
+      "step": 19751
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 4.762639510363627e-06,
+      "loss": 0.6333,
+      "step": 19752
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 4.7600678907575996e-06,
+      "loss": 0.3394,
+      "step": 19753
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 4.757496271151572e-06,
+      "loss": 0.6157,
+      "step": 19754
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 4.754924651545543e-06,
+      "loss": 0.8325,
+      "step": 19755
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 4.752353031939516e-06,
+      "loss": 0.3311,
+      "step": 19756
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 4.749781412333488e-06,
+      "loss": 0.1552,
+      "step": 19757
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 4.74720979272746e-06,
+      "loss": 0.5878,
+      "step": 19758
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 4.744638173121432e-06,
+      "loss": 0.7951,
+      "step": 19759
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 4.742066553515404e-06,
+      "loss": 0.6384,
+      "step": 19760
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 4.7394949339093765e-06,
+      "loss": 0.4307,
+      "step": 19761
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 4.736923314303349e-06,
+      "loss": 0.6122,
+      "step": 19762
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 4.734351694697321e-06,
+      "loss": 0.6638,
+      "step": 19763
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 4.731780075091292e-06,
+      "loss": 0.6155,
+      "step": 19764
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 4.7292084554852654e-06,
+      "loss": 0.2792,
+      "step": 19765
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 4.726636835879237e-06,
+      "loss": 0.6912,
+      "step": 19766
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 4.724065216273209e-06,
+      "loss": 0.4497,
+      "step": 19767
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 4.721493596667181e-06,
+      "loss": 0.618,
+      "step": 19768
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 4.7189219770611535e-06,
+      "loss": 0.4405,
+      "step": 19769
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 4.716350357455125e-06,
+      "loss": 1.1657,
+      "step": 19770
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 4.713778737849098e-06,
+      "loss": 0.8052,
+      "step": 19771
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 4.711207118243069e-06,
+      "loss": 0.6783,
+      "step": 19772
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 4.7086354986370415e-06,
+      "loss": 0.2832,
+      "step": 19773
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 4.706063879031014e-06,
+      "loss": 0.6496,
+      "step": 19774
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 4.703492259424986e-06,
+      "loss": 0.443,
+      "step": 19775
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 4.700920639818958e-06,
+      "loss": 0.6103,
+      "step": 19776
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 4.6983490202129305e-06,
+      "loss": 0.3607,
+      "step": 19777
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 4.695777400606903e-06,
+      "loss": 0.5427,
+      "step": 19778
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 4.693205781000875e-06,
+      "loss": 0.3751,
+      "step": 19779
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 4.690634161394847e-06,
+      "loss": 0.5193,
+      "step": 19780
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 4.6880625417888185e-06,
+      "loss": 0.5708,
+      "step": 19781
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 4.685490922182792e-06,
+      "loss": 0.4602,
+      "step": 19782
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 4.682919302576763e-06,
+      "loss": 0.4132,
+      "step": 19783
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 4.680347682970735e-06,
+      "loss": 0.381,
+      "step": 19784
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 4.677776063364707e-06,
+      "loss": 0.4379,
+      "step": 19785
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 4.67520444375868e-06,
+      "loss": 0.5779,
+      "step": 19786
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 4.672632824152651e-06,
+      "loss": 0.5722,
+      "step": 19787
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 4.670061204546624e-06,
+      "loss": 0.4755,
+      "step": 19788
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 4.6674895849405955e-06,
+      "loss": 0.6769,
+      "step": 19789
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 4.664917965334568e-06,
+      "loss": 0.6388,
+      "step": 19790
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 4.66234634572854e-06,
+      "loss": 0.8135,
+      "step": 19791
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 4.659774726122512e-06,
+      "loss": 0.3587,
+      "step": 19792
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 4.657203106516484e-06,
+      "loss": 0.5423,
+      "step": 19793
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 4.654631486910457e-06,
+      "loss": 0.4818,
+      "step": 19794
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 4.652059867304429e-06,
+      "loss": 0.6261,
+      "step": 19795
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 4.6494882476984e-06,
+      "loss": 0.3633,
+      "step": 19796
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 4.646916628092373e-06,
+      "loss": 0.7109,
+      "step": 19797
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 4.644345008486345e-06,
+      "loss": 0.8427,
+      "step": 19798
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 4.641773388880317e-06,
+      "loss": 0.5773,
+      "step": 19799
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 4.639201769274289e-06,
+      "loss": 0.7945,
+      "step": 19800
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 4.636630149668261e-06,
+      "loss": 0.7596,
+      "step": 19801
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 4.634058530062233e-06,
+      "loss": 0.4431,
+      "step": 19802
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 4.631486910456206e-06,
+      "loss": 0.4848,
+      "step": 19803
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 4.628915290850178e-06,
+      "loss": 0.7638,
+      "step": 19804
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 4.626343671244149e-06,
+      "loss": 0.3359,
+      "step": 19805
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 4.6237720516381225e-06,
+      "loss": 0.7437,
+      "step": 19806
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 4.621200432032094e-06,
+      "loss": 0.349,
+      "step": 19807
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 4.618628812426066e-06,
+      "loss": 0.2973,
+      "step": 19808
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 4.616057192820038e-06,
+      "loss": 0.5685,
+      "step": 19809
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 4.6134855732140105e-06,
+      "loss": 0.603,
+      "step": 19810
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 4.610913953607982e-06,
+      "loss": 0.664,
+      "step": 19811
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 4.608342334001955e-06,
+      "loss": 0.9385,
+      "step": 19812
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 4.605770714395926e-06,
+      "loss": 0.4002,
+      "step": 19813
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 4.603199094789899e-06,
+      "loss": 0.6872,
+      "step": 19814
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 4.600627475183871e-06,
+      "loss": 0.6593,
+      "step": 19815
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 4.598055855577843e-06,
+      "loss": 0.3768,
+      "step": 19816
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 4.595484235971815e-06,
+      "loss": 0.6623,
+      "step": 19817
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 4.5929126163657875e-06,
+      "loss": 0.6352,
+      "step": 19818
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 4.59034099675976e-06,
+      "loss": 0.5279,
+      "step": 19819
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 4.587769377153732e-06,
+      "loss": 0.7346,
+      "step": 19820
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 4.585197757547704e-06,
+      "loss": 0.5412,
+      "step": 19821
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 4.5826261379416756e-06,
+      "loss": 0.63,
+      "step": 19822
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 4.580054518335649e-06,
+      "loss": 0.8832,
+      "step": 19823
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 4.57748289872962e-06,
+      "loss": 0.4062,
+      "step": 19824
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 4.574911279123592e-06,
+      "loss": 0.5064,
+      "step": 19825
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 4.5723396595175645e-06,
+      "loss": 0.7461,
+      "step": 19826
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 4.569768039911537e-06,
+      "loss": 1.0289,
+      "step": 19827
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 4.567196420305508e-06,
+      "loss": 0.7236,
+      "step": 19828
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 4.564624800699481e-06,
+      "loss": 0.5981,
+      "step": 19829
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 4.5620531810934525e-06,
+      "loss": 0.5652,
+      "step": 19830
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 4.559481561487425e-06,
+      "loss": 1.0439,
+      "step": 19831
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 4.556909941881397e-06,
+      "loss": 0.3827,
+      "step": 19832
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 4.554338322275369e-06,
+      "loss": 0.8119,
+      "step": 19833
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 4.5517667026693414e-06,
+      "loss": 1.0362,
+      "step": 19834
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 4.549195083063314e-06,
+      "loss": 0.4948,
+      "step": 19835
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 4.546623463457286e-06,
+      "loss": 0.473,
+      "step": 19836
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 4.544051843851257e-06,
+      "loss": 0.3248,
+      "step": 19837
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 4.54148022424523e-06,
+      "loss": 0.5642,
+      "step": 19838
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 4.538908604639202e-06,
+      "loss": 0.3673,
+      "step": 19839
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 4.536336985033174e-06,
+      "loss": 0.4344,
+      "step": 19840
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 4.533765365427146e-06,
+      "loss": 0.8425,
+      "step": 19841
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 4.531193745821118e-06,
+      "loss": 0.7369,
+      "step": 19842
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 4.52862212621509e-06,
+      "loss": 0.4447,
+      "step": 19843
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 4.526050506609063e-06,
+      "loss": 0.6276,
+      "step": 19844
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 4.523478887003035e-06,
+      "loss": 0.4749,
+      "step": 19845
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 4.5209072673970064e-06,
+      "loss": 0.5364,
+      "step": 19846
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 4.5183356477909795e-06,
+      "loss": 0.3869,
+      "step": 19847
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 4.515764028184951e-06,
+      "loss": 0.6024,
+      "step": 19848
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 4.513192408578923e-06,
+      "loss": 0.3752,
+      "step": 19849
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 4.510620788972895e-06,
+      "loss": 0.489,
+      "step": 19850
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 4.508049169366868e-06,
+      "loss": 0.2429,
+      "step": 19851
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 4.50547754976084e-06,
+      "loss": 0.5549,
+      "step": 19852
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 4.502905930154812e-06,
+      "loss": 0.6831,
+      "step": 19853
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 4.500334310548783e-06,
+      "loss": 0.3357,
+      "step": 19854
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 4.4977626909427565e-06,
+      "loss": 0.5543,
+      "step": 19855
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 4.495191071336728e-06,
+      "loss": 0.3333,
+      "step": 19856
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 4.4926194517307e-06,
+      "loss": 0.5136,
+      "step": 19857
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 4.490047832124672e-06,
+      "loss": 0.4826,
+      "step": 19858
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 4.4874762125186445e-06,
+      "loss": 0.4622,
+      "step": 19859
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 4.484904592912617e-06,
+      "loss": 0.4975,
+      "step": 19860
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 4.482332973306589e-06,
+      "loss": 0.5065,
+      "step": 19861
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 4.479761353700561e-06,
+      "loss": 0.6739,
+      "step": 19862
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 4.477189734094533e-06,
+      "loss": 0.8329,
+      "step": 19863
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 4.474618114488506e-06,
+      "loss": 0.5748,
+      "step": 19864
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 4.472046494882477e-06,
+      "loss": 0.4353,
+      "step": 19865
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 4.469474875276449e-06,
+      "loss": 0.6396,
+      "step": 19866
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 4.4669032556704215e-06,
+      "loss": 0.6435,
+      "step": 19867
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 4.464331636064394e-06,
+      "loss": 0.9647,
+      "step": 19868
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 4.461760016458365e-06,
+      "loss": 0.4688,
+      "step": 19869
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 4.459188396852338e-06,
+      "loss": 0.7875,
+      "step": 19870
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 4.4566167772463096e-06,
+      "loss": 0.8021,
+      "step": 19871
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 4.454045157640282e-06,
+      "loss": 0.234,
+      "step": 19872
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 4.451473538034254e-06,
+      "loss": 0.4493,
+      "step": 19873
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 4.448901918428226e-06,
+      "loss": 0.4762,
+      "step": 19874
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 4.4463302988221985e-06,
+      "loss": 0.4923,
+      "step": 19875
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 4.443758679216171e-06,
+      "loss": 0.7248,
+      "step": 19876
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 4.441187059610143e-06,
+      "loss": 0.3704,
+      "step": 19877
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 4.438615440004114e-06,
+      "loss": 0.497,
+      "step": 19878
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 4.436043820398087e-06,
+      "loss": 0.5714,
+      "step": 19879
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 4.433472200792059e-06,
+      "loss": 0.6146,
+      "step": 19880
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 4.430900581186031e-06,
+      "loss": 0.6192,
+      "step": 19881
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 4.428328961580003e-06,
+      "loss": 0.7208,
+      "step": 19882
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 4.4257573419739754e-06,
+      "loss": 0.8106,
+      "step": 19883
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 4.423185722367947e-06,
+      "loss": 0.3238,
+      "step": 19884
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 4.42061410276192e-06,
+      "loss": 0.3657,
+      "step": 19885
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 4.418042483155892e-06,
+      "loss": 0.352,
+      "step": 19886
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 4.4154708635498635e-06,
+      "loss": 0.6878,
+      "step": 19887
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 4.4128992439438366e-06,
+      "loss": 0.6333,
+      "step": 19888
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 4.410327624337808e-06,
+      "loss": 0.6435,
+      "step": 19889
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 4.407756004731781e-06,
+      "loss": 0.5199,
+      "step": 19890
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 4.405184385125752e-06,
+      "loss": 0.7305,
+      "step": 19891
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 4.402612765519725e-06,
+      "loss": 0.6635,
+      "step": 19892
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 4.400041145913697e-06,
+      "loss": 0.5534,
+      "step": 19893
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 4.397469526307669e-06,
+      "loss": 0.7666,
+      "step": 19894
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 4.3948979067016405e-06,
+      "loss": 0.8984,
+      "step": 19895
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 4.3923262870956135e-06,
+      "loss": 0.7465,
+      "step": 19896
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 4.389754667489585e-06,
+      "loss": 0.6629,
+      "step": 19897
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 4.387183047883557e-06,
+      "loss": 0.5126,
+      "step": 19898
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 4.384611428277529e-06,
+      "loss": 0.5052,
+      "step": 19899
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 4.382039808671502e-06,
+      "loss": 0.5856,
+      "step": 19900
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 4.379468189065474e-06,
+      "loss": 0.5064,
+      "step": 19901
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 4.376896569459446e-06,
+      "loss": 0.4511,
+      "step": 19902
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 4.374324949853418e-06,
+      "loss": 0.6336,
+      "step": 19903
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 4.37175333024739e-06,
+      "loss": 0.5386,
+      "step": 19904
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 4.369181710641363e-06,
+      "loss": 0.5947,
+      "step": 19905
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 4.366610091035334e-06,
+      "loss": 0.2995,
+      "step": 19906
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 4.364038471429306e-06,
+      "loss": 0.5216,
+      "step": 19907
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 4.3614668518232786e-06,
+      "loss": 0.7529,
+      "step": 19908
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 4.358895232217251e-06,
+      "loss": 0.6987,
+      "step": 19909
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 4.356323612611222e-06,
+      "loss": 0.7926,
+      "step": 19910
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 4.353751993005195e-06,
+      "loss": 0.2228,
+      "step": 19911
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 4.351180373399167e-06,
+      "loss": 0.5776,
+      "step": 19912
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 4.348608753793139e-06,
+      "loss": 0.6428,
+      "step": 19913
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 4.346037134187111e-06,
+      "loss": 0.6785,
+      "step": 19914
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 4.343465514581083e-06,
+      "loss": 0.4105,
+      "step": 19915
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 4.3408938949750555e-06,
+      "loss": 0.9819,
+      "step": 19916
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 4.338322275369028e-06,
+      "loss": 0.3343,
+      "step": 19917
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 4.335750655763e-06,
+      "loss": 0.5971,
+      "step": 19918
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 4.333179036156971e-06,
+      "loss": 0.8395,
+      "step": 19919
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 4.330607416550944e-06,
+      "loss": 0.5635,
+      "step": 19920
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 4.328035796944916e-06,
+      "loss": 0.3841,
+      "step": 19921
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 4.325464177338888e-06,
+      "loss": 0.6291,
+      "step": 19922
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 4.32289255773286e-06,
+      "loss": 0.615,
+      "step": 19923
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 4.3203209381268325e-06,
+      "loss": 0.7747,
+      "step": 19924
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 4.317749318520804e-06,
+      "loss": 0.5684,
+      "step": 19925
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 4.315177698914777e-06,
+      "loss": 0.4569,
+      "step": 19926
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 4.312606079308749e-06,
+      "loss": 0.5031,
+      "step": 19927
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 4.310034459702721e-06,
+      "loss": 0.7599,
+      "step": 19928
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 4.307462840096694e-06,
+      "loss": 0.4942,
+      "step": 19929
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 4.304891220490665e-06,
+      "loss": 0.6284,
+      "step": 19930
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 4.302319600884638e-06,
+      "loss": 0.8482,
+      "step": 19931
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 4.2997479812786094e-06,
+      "loss": 0.7826,
+      "step": 19932
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 4.297176361672582e-06,
+      "loss": 0.7475,
+      "step": 19933
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 4.294604742066554e-06,
+      "loss": 0.4192,
+      "step": 19934
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 4.292033122460526e-06,
+      "loss": 0.6944,
+      "step": 19935
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 4.2894615028544975e-06,
+      "loss": 0.6897,
+      "step": 19936
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 4.286889883248471e-06,
+      "loss": 0.8948,
+      "step": 19937
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 4.284318263642442e-06,
+      "loss": 0.5211,
+      "step": 19938
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 4.281746644036414e-06,
+      "loss": 0.7891,
+      "step": 19939
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 4.279175024430386e-06,
+      "loss": 0.6535,
+      "step": 19940
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 4.276603404824359e-06,
+      "loss": 0.6633,
+      "step": 19941
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 4.274031785218331e-06,
+      "loss": 0.3501,
+      "step": 19942
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 4.271460165612303e-06,
+      "loss": 0.694,
+      "step": 19943
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 4.268888546006275e-06,
+      "loss": 0.5184,
+      "step": 19944
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 4.266316926400247e-06,
+      "loss": 0.5456,
+      "step": 19945
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 4.26374530679422e-06,
+      "loss": 0.3999,
+      "step": 19946
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 4.261173687188191e-06,
+      "loss": 0.795,
+      "step": 19947
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 4.258602067582163e-06,
+      "loss": 0.6319,
+      "step": 19948
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 4.256030447976136e-06,
+      "loss": 0.5989,
+      "step": 19949
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 4.253458828370108e-06,
+      "loss": 0.8626,
+      "step": 19950
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 4.250887208764079e-06,
+      "loss": 0.4788,
+      "step": 19951
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 4.248315589158052e-06,
+      "loss": 0.2837,
+      "step": 19952
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 4.245743969552024e-06,
+      "loss": 0.4477,
+      "step": 19953
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 4.243172349945996e-06,
+      "loss": 0.6662,
+      "step": 19954
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 4.240600730339968e-06,
+      "loss": 0.3161,
+      "step": 19955
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 4.23802911073394e-06,
+      "loss": 0.8433,
+      "step": 19956
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 4.2354574911279126e-06,
+      "loss": 0.7609,
+      "step": 19957
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 4.232885871521885e-06,
+      "loss": 0.5331,
+      "step": 19958
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 4.230314251915857e-06,
+      "loss": 0.5185,
+      "step": 19959
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 4.227742632309828e-06,
+      "loss": 0.5729,
+      "step": 19960
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 4.2251710127038015e-06,
+      "loss": 0.3962,
+      "step": 19961
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 4.222599393097773e-06,
+      "loss": 0.5927,
+      "step": 19962
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 4.220027773491745e-06,
+      "loss": 0.6091,
+      "step": 19963
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 4.217456153885717e-06,
+      "loss": 0.7168,
+      "step": 19964
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 4.2148845342796895e-06,
+      "loss": 0.9046,
+      "step": 19965
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 4.212312914673662e-06,
+      "loss": 0.3198,
+      "step": 19966
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 4.209741295067634e-06,
+      "loss": 0.6615,
+      "step": 19967
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 4.207169675461606e-06,
+      "loss": 0.7291,
+      "step": 19968
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 4.2045980558555784e-06,
+      "loss": 0.6341,
+      "step": 19969
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 4.202026436249551e-06,
+      "loss": 0.7562,
+      "step": 19970
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 4.199454816643522e-06,
+      "loss": 0.6993,
+      "step": 19971
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 4.196883197037495e-06,
+      "loss": 0.5338,
+      "step": 19972
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 4.1943115774314665e-06,
+      "loss": 0.3993,
+      "step": 19973
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 4.191739957825439e-06,
+      "loss": 0.8132,
+      "step": 19974
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 4.189168338219411e-06,
+      "loss": 0.3195,
+      "step": 19975
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 4.186596718613383e-06,
+      "loss": 0.6142,
+      "step": 19976
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 4.1840250990073546e-06,
+      "loss": 0.865,
+      "step": 19977
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 4.181453479401328e-06,
+      "loss": 0.6525,
+      "step": 19978
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 4.178881859795299e-06,
+      "loss": 0.4527,
+      "step": 19979
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 4.176310240189271e-06,
+      "loss": 0.9277,
+      "step": 19980
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 4.1737386205832435e-06,
+      "loss": 0.7008,
+      "step": 19981
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 4.171167000977216e-06,
+      "loss": 0.5242,
+      "step": 19982
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 4.168595381371188e-06,
+      "loss": 0.524,
+      "step": 19983
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 4.16602376176516e-06,
+      "loss": 0.4338,
+      "step": 19984
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 4.163452142159132e-06,
+      "loss": 0.7318,
+      "step": 19985
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 4.160880522553104e-06,
+      "loss": 0.3939,
+      "step": 19986
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 4.158308902947077e-06,
+      "loss": 0.6912,
+      "step": 19987
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 4.155737283341048e-06,
+      "loss": 0.546,
+      "step": 19988
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 4.15316566373502e-06,
+      "loss": 0.6269,
+      "step": 19989
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 4.150594044128993e-06,
+      "loss": 0.5022,
+      "step": 19990
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 4.148022424522965e-06,
+      "loss": 0.6811,
+      "step": 19991
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 4.145450804916936e-06,
+      "loss": 0.5998,
+      "step": 19992
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 4.142879185310909e-06,
+      "loss": 0.5192,
+      "step": 19993
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 4.140307565704881e-06,
+      "loss": 0.6319,
+      "step": 19994
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 4.137735946098853e-06,
+      "loss": 0.7508,
+      "step": 19995
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 4.135164326492825e-06,
+      "loss": 0.9658,
+      "step": 19996
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 4.132592706886797e-06,
+      "loss": 0.5464,
+      "step": 19997
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 4.13002108728077e-06,
+      "loss": 0.6022,
+      "step": 19998
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 4.127449467674742e-06,
+      "loss": 0.9479,
+      "step": 19999
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 4.124877848068714e-06,
+      "loss": 0.4357,
+      "step": 20000
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 4.1223062284626854e-06,
+      "loss": 0.6407,
+      "step": 20001
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 4.1197346088566585e-06,
+      "loss": 0.4063,
+      "step": 20002
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 4.11716298925063e-06,
+      "loss": 0.6731,
+      "step": 20003
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 4.114591369644603e-06,
+      "loss": 0.7296,
+      "step": 20004
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 4.112019750038574e-06,
+      "loss": 0.4882,
+      "step": 20005
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 4.1094481304325466e-06,
+      "loss": 0.265,
+      "step": 20006
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 4.106876510826519e-06,
+      "loss": 0.2582,
+      "step": 20007
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 4.104304891220491e-06,
+      "loss": 0.6909,
+      "step": 20008
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 4.101733271614463e-06,
+      "loss": 0.5486,
+      "step": 20009
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 4.0991616520084355e-06,
+      "loss": 0.5568,
+      "step": 20010
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 4.096590032402408e-06,
+      "loss": 0.5523,
+      "step": 20011
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 4.094018412796379e-06,
+      "loss": 0.7793,
+      "step": 20012
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 4.091446793190352e-06,
+      "loss": 0.4067,
+      "step": 20013
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 4.0888751735843235e-06,
+      "loss": 0.5238,
+      "step": 20014
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 4.086303553978296e-06,
+      "loss": 0.9659,
+      "step": 20015
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 4.083731934372268e-06,
+      "loss": 0.607,
+      "step": 20016
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 4.08116031476624e-06,
+      "loss": 0.505,
+      "step": 20017
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 4.078588695160212e-06,
+      "loss": 0.4118,
+      "step": 20018
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 4.076017075554185e-06,
+      "loss": 0.5845,
+      "step": 20019
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 4.073445455948156e-06,
+      "loss": 0.5243,
+      "step": 20020
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 4.070873836342128e-06,
+      "loss": 0.4816,
+      "step": 20021
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 4.0683022167361005e-06,
+      "loss": 0.722,
+      "step": 20022
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 4.065730597130073e-06,
+      "loss": 0.6376,
+      "step": 20023
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 4.063158977524045e-06,
+      "loss": 0.5159,
+      "step": 20024
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 4.060587357918017e-06,
+      "loss": 0.3348,
+      "step": 20025
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 4.058015738311989e-06,
+      "loss": 0.4934,
+      "step": 20026
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 4.055444118705961e-06,
+      "loss": 0.4049,
+      "step": 20027
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 4.052872499099934e-06,
+      "loss": 0.6518,
+      "step": 20028
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 4.050300879493905e-06,
+      "loss": 0.752,
+      "step": 20029
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 4.0477292598878775e-06,
+      "loss": 0.6969,
+      "step": 20030
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 4.04515764028185e-06,
+      "loss": 0.3448,
+      "step": 20031
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 4.042586020675822e-06,
+      "loss": 0.184,
+      "step": 20032
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 4.040014401069793e-06,
+      "loss": 0.944,
+      "step": 20033
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 4.037442781463766e-06,
+      "loss": 0.4302,
+      "step": 20034
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 4.034871161857738e-06,
+      "loss": 0.6561,
+      "step": 20035
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 4.03229954225171e-06,
+      "loss": 0.2983,
+      "step": 20036
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 4.029727922645682e-06,
+      "loss": 0.5197,
+      "step": 20037
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 4.0271563030396544e-06,
+      "loss": 0.5202,
+      "step": 20038
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 4.024584683433627e-06,
+      "loss": 0.5563,
+      "step": 20039
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 4.022013063827599e-06,
+      "loss": 0.3986,
+      "step": 20040
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 4.019441444221571e-06,
+      "loss": 0.95,
+      "step": 20041
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 4.016869824615543e-06,
+      "loss": 0.8772,
+      "step": 20042
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 4.0142982050095156e-06,
+      "loss": 0.4088,
+      "step": 20043
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 4.011726585403487e-06,
+      "loss": 0.7815,
+      "step": 20044
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 4.00915496579746e-06,
+      "loss": 0.6296,
+      "step": 20045
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 4.006583346191431e-06,
+      "loss": 0.7629,
+      "step": 20046
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 4.004011726585404e-06,
+      "loss": 0.2962,
+      "step": 20047
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 4.001440106979376e-06,
+      "loss": 0.643,
+      "step": 20048
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 3.998868487373348e-06,
+      "loss": 0.5195,
+      "step": 20049
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 3.9962968677673195e-06,
+      "loss": 0.6816,
+      "step": 20050
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 3.9937252481612925e-06,
+      "loss": 0.6633,
+      "step": 20051
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 3.991153628555265e-06,
+      "loss": 0.3852,
+      "step": 20052
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 3.988582008949236e-06,
+      "loss": 0.6318,
+      "step": 20053
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 3.986010389343209e-06,
+      "loss": 0.4713,
+      "step": 20054
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 3.983438769737181e-06,
+      "loss": 0.4883,
+      "step": 20055
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 3.980867150131153e-06,
+      "loss": 0.4848,
+      "step": 20056
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 3.978295530525125e-06,
+      "loss": 0.4675,
+      "step": 20057
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 3.975723910919097e-06,
+      "loss": 0.5609,
+      "step": 20058
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 3.973152291313069e-06,
+      "loss": 0.6488,
+      "step": 20059
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 3.970580671707042e-06,
+      "loss": 0.5265,
+      "step": 20060
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 3.968009052101013e-06,
+      "loss": 0.5018,
+      "step": 20061
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 3.965437432494985e-06,
+      "loss": 0.7083,
+      "step": 20062
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 3.9628658128889575e-06,
+      "loss": 0.4323,
+      "step": 20063
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 3.96029419328293e-06,
+      "loss": 0.641,
+      "step": 20064
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 3.957722573676902e-06,
+      "loss": 0.6741,
+      "step": 20065
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 3.955150954070874e-06,
+      "loss": 0.3823,
+      "step": 20066
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 3.9525793344648465e-06,
+      "loss": 0.3077,
+      "step": 20067
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 3.950007714858818e-06,
+      "loss": 0.5107,
+      "step": 20068
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 3.947436095252791e-06,
+      "loss": 0.5733,
+      "step": 20069
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 3.944864475646762e-06,
+      "loss": 0.5461,
+      "step": 20070
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 3.9422928560407345e-06,
+      "loss": 0.2415,
+      "step": 20071
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 3.939721236434707e-06,
+      "loss": 0.7261,
+      "step": 20072
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 3.937149616828679e-06,
+      "loss": 0.5304,
+      "step": 20073
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 3.93457799722265e-06,
+      "loss": 0.573,
+      "step": 20074
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 3.932006377616623e-06,
+      "loss": 0.8212,
+      "step": 20075
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 3.929434758010595e-06,
+      "loss": 0.3913,
+      "step": 20076
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 3.926863138404567e-06,
+      "loss": 0.7367,
+      "step": 20077
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 3.924291518798539e-06,
+      "loss": 0.5299,
+      "step": 20078
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 3.9217198991925115e-06,
+      "loss": 0.4238,
+      "step": 20079
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 3.919148279586484e-06,
+      "loss": 0.4913,
+      "step": 20080
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 3.916576659980456e-06,
+      "loss": 0.6152,
+      "step": 20081
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 3.914005040374428e-06,
+      "loss": 0.4113,
+      "step": 20082
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 3.9114334207684e-06,
+      "loss": 0.5155,
+      "step": 20083
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 3.908861801162373e-06,
+      "loss": 0.5199,
+      "step": 20084
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 3.906290181556344e-06,
+      "loss": 0.7851,
+      "step": 20085
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 3.903718561950317e-06,
+      "loss": 0.8477,
+      "step": 20086
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 3.9011469423442884e-06,
+      "loss": 0.4756,
+      "step": 20087
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 3.898575322738261e-06,
+      "loss": 0.4838,
+      "step": 20088
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 3.896003703132233e-06,
+      "loss": 0.7754,
+      "step": 20089
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 3.893432083526205e-06,
+      "loss": 0.3214,
+      "step": 20090
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 3.8908604639201765e-06,
+      "loss": 0.5313,
+      "step": 20091
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 3.8882888443141496e-06,
+      "loss": 0.4682,
+      "step": 20092
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 3.885717224708122e-06,
+      "loss": 0.4616,
+      "step": 20093
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 3.883145605102093e-06,
+      "loss": 0.4997,
+      "step": 20094
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 3.880573985496066e-06,
+      "loss": 0.6047,
+      "step": 20095
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 3.878002365890038e-06,
+      "loss": 0.842,
+      "step": 20096
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 3.87543074628401e-06,
+      "loss": 0.6871,
+      "step": 20097
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 3.872859126677982e-06,
+      "loss": 0.5834,
+      "step": 20098
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 3.870287507071954e-06,
+      "loss": 0.6498,
+      "step": 20099
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 3.867715887465926e-06,
+      "loss": 0.626,
+      "step": 20100
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 3.865144267859899e-06,
+      "loss": 0.326,
+      "step": 20101
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 3.86257264825387e-06,
+      "loss": 0.3309,
+      "step": 20102
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 3.860001028647842e-06,
+      "loss": 0.4294,
+      "step": 20103
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 3.857429409041815e-06,
+      "loss": 0.4341,
+      "step": 20104
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 3.854857789435787e-06,
+      "loss": 0.6268,
+      "step": 20105
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 3.852286169829759e-06,
+      "loss": 0.5666,
+      "step": 20106
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 3.849714550223731e-06,
+      "loss": 0.4955,
+      "step": 20107
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 3.8471429306177035e-06,
+      "loss": 0.7335,
+      "step": 20108
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 3.844571311011675e-06,
+      "loss": 0.4686,
+      "step": 20109
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 3.841999691405648e-06,
+      "loss": 0.4739,
+      "step": 20110
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 3.839428071799619e-06,
+      "loss": 0.6568,
+      "step": 20111
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 3.8368564521935916e-06,
+      "loss": 0.6453,
+      "step": 20112
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 3.834284832587564e-06,
+      "loss": 0.5244,
+      "step": 20113
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 3.831713212981536e-06,
+      "loss": 0.6615,
+      "step": 20114
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 3.829141593375507e-06,
+      "loss": 0.6127,
+      "step": 20115
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 3.8265699737694805e-06,
+      "loss": 0.4979,
+      "step": 20116
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 3.823998354163452e-06,
+      "loss": 0.6788,
+      "step": 20117
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 3.821426734557425e-06,
+      "loss": 0.5687,
+      "step": 20118
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 3.818855114951396e-06,
+      "loss": 0.5063,
+      "step": 20119
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 3.8162834953453685e-06,
+      "loss": 0.935,
+      "step": 20120
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 3.813711875739341e-06,
+      "loss": 0.6607,
+      "step": 20121
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 3.811140256133313e-06,
+      "loss": 0.879,
+      "step": 20122
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 3.8085686365272848e-06,
+      "loss": 0.4095,
+      "step": 20123
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 3.8059970169212574e-06,
+      "loss": 0.7103,
+      "step": 20124
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 3.8034253973152292e-06,
+      "loss": 0.5996,
+      "step": 20125
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 3.8008537777092015e-06,
+      "loss": 0.7601,
+      "step": 20126
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 3.7982821581031737e-06,
+      "loss": 0.5926,
+      "step": 20127
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 3.795710538497146e-06,
+      "loss": 1.0,
+      "step": 20128
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 3.7931389188911177e-06,
+      "loss": 0.363,
+      "step": 20129
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 3.7905672992850904e-06,
+      "loss": 0.6469,
+      "step": 20130
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 3.787995679679062e-06,
+      "loss": 0.743,
+      "step": 20131
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 3.785424060073034e-06,
+      "loss": 0.4001,
+      "step": 20132
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 3.7828524404670066e-06,
+      "loss": 0.6069,
+      "step": 20133
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 3.7802808208609784e-06,
+      "loss": 0.4766,
+      "step": 20134
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 3.7777092012549502e-06,
+      "loss": 0.6011,
+      "step": 20135
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 3.775137581648923e-06,
+      "loss": 0.7908,
+      "step": 20136
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 3.7725659620428947e-06,
+      "loss": 0.5338,
+      "step": 20137
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 3.7699943424368665e-06,
+      "loss": 0.6407,
+      "step": 20138
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 3.767422722830839e-06,
+      "loss": 0.3767,
+      "step": 20139
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 3.764851103224811e-06,
+      "loss": 0.87,
+      "step": 20140
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 3.762279483618783e-06,
+      "loss": 0.7334,
+      "step": 20141
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 3.759707864012756e-06,
+      "loss": 0.5578,
+      "step": 20142
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 3.7571362444067276e-06,
+      "loss": 0.5777,
+      "step": 20143
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 3.7545646248006994e-06,
+      "loss": 0.4934,
+      "step": 20144
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 3.751993005194672e-06,
+      "loss": 0.4975,
+      "step": 20145
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 3.749421385588644e-06,
+      "loss": 0.5234,
+      "step": 20146
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 3.7468497659826157e-06,
+      "loss": 0.7868,
+      "step": 20147
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 3.7442781463765883e-06,
+      "loss": 0.3654,
+      "step": 20148
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 3.74170652677056e-06,
+      "loss": 0.8631,
+      "step": 20149
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 3.739134907164532e-06,
+      "loss": 0.7303,
+      "step": 20150
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 3.7365632875585046e-06,
+      "loss": 0.5393,
+      "step": 20151
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 3.7339916679524764e-06,
+      "loss": 0.7181,
+      "step": 20152
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 3.7314200483464486e-06,
+      "loss": 0.3981,
+      "step": 20153
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 3.728848428740421e-06,
+      "loss": 0.4507,
+      "step": 20154
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 3.726276809134393e-06,
+      "loss": 0.4468,
+      "step": 20155
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 3.7237051895283653e-06,
+      "loss": 0.6767,
+      "step": 20156
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 3.7211335699223375e-06,
+      "loss": 0.7346,
+      "step": 20157
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 3.7185619503163093e-06,
+      "loss": 0.6952,
+      "step": 20158
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 3.715990330710282e-06,
+      "loss": 0.637,
+      "step": 20159
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 3.7134187111042538e-06,
+      "loss": 0.7812,
+      "step": 20160
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 3.7108470914982256e-06,
+      "loss": 0.5294,
+      "step": 20161
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 3.7082754718921982e-06,
+      "loss": 0.5304,
+      "step": 20162
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 3.70570385228617e-06,
+      "loss": 0.7809,
+      "step": 20163
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 3.703132232680142e-06,
+      "loss": 0.7087,
+      "step": 20164
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 3.7005606130741145e-06,
+      "loss": 0.7003,
+      "step": 20165
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 3.6979889934680863e-06,
+      "loss": 0.5883,
+      "step": 20166
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 3.6954173738620585e-06,
+      "loss": 0.4754,
+      "step": 20167
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 3.6928457542560307e-06,
+      "loss": 0.6138,
+      "step": 20168
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 3.690274134650003e-06,
+      "loss": 0.3127,
+      "step": 20169
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 3.6877025150439748e-06,
+      "loss": 0.9128,
+      "step": 20170
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 3.6851308954379474e-06,
+      "loss": 0.5577,
+      "step": 20171
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 3.6825592758319192e-06,
+      "loss": 0.6092,
+      "step": 20172
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 3.679987656225891e-06,
+      "loss": 0.4717,
+      "step": 20173
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 3.6774160366198637e-06,
+      "loss": 0.6456,
+      "step": 20174
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 3.6748444170138355e-06,
+      "loss": 0.6208,
+      "step": 20175
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 3.6722727974078073e-06,
+      "loss": 0.4027,
+      "step": 20176
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 3.66970117780178e-06,
+      "loss": 0.7996,
+      "step": 20177
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 3.6671295581957517e-06,
+      "loss": 0.8242,
+      "step": 20178
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 3.6645579385897235e-06,
+      "loss": 0.5713,
+      "step": 20179
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 3.661986318983696e-06,
+      "loss": 0.5289,
+      "step": 20180
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 3.659414699377668e-06,
+      "loss": 0.4915,
+      "step": 20181
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 3.65684307977164e-06,
+      "loss": 0.4877,
+      "step": 20182
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 3.6542714601656124e-06,
+      "loss": 0.4111,
+      "step": 20183
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 3.6516998405595847e-06,
+      "loss": 0.6722,
+      "step": 20184
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 3.6491282209535565e-06,
+      "loss": 0.3373,
+      "step": 20185
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 3.646556601347529e-06,
+      "loss": 0.6656,
+      "step": 20186
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 3.643984981741501e-06,
+      "loss": 0.3321,
+      "step": 20187
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 3.6414133621354727e-06,
+      "loss": 0.7399,
+      "step": 20188
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 3.6388417425294454e-06,
+      "loss": 0.2611,
+      "step": 20189
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 3.636270122923417e-06,
+      "loss": 0.4289,
+      "step": 20190
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 3.63369850331739e-06,
+      "loss": 0.5778,
+      "step": 20191
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 3.6311268837113616e-06,
+      "loss": 0.3417,
+      "step": 20192
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 3.6285552641053334e-06,
+      "loss": 0.749,
+      "step": 20193
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 3.625983644499306e-06,
+      "loss": 0.4201,
+      "step": 20194
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 3.623412024893278e-06,
+      "loss": 0.6445,
+      "step": 20195
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 3.62084040528725e-06,
+      "loss": 0.3871,
+      "step": 20196
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 3.6182687856812223e-06,
+      "loss": 0.472,
+      "step": 20197
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 3.6156971660751946e-06,
+      "loss": 0.5982,
+      "step": 20198
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 3.6131255464691664e-06,
+      "loss": 0.8377,
+      "step": 20199
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 3.610553926863139e-06,
+      "loss": 0.6222,
+      "step": 20200
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 3.607982307257111e-06,
+      "loss": 0.2921,
+      "step": 20201
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 3.6054106876510826e-06,
+      "loss": 0.4927,
+      "step": 20202
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 3.6028390680450553e-06,
+      "loss": 0.5101,
+      "step": 20203
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 3.600267448439027e-06,
+      "loss": 0.7188,
+      "step": 20204
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 3.597695828832999e-06,
+      "loss": 0.7103,
+      "step": 20205
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 3.5951242092269715e-06,
+      "loss": 0.5889,
+      "step": 20206
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 3.5925525896209433e-06,
+      "loss": 0.5702,
+      "step": 20207
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 3.5899809700149155e-06,
+      "loss": 0.7561,
+      "step": 20208
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 3.5874093504088878e-06,
+      "loss": 0.6539,
+      "step": 20209
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 3.58483773080286e-06,
+      "loss": 0.668,
+      "step": 20210
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 3.582266111196832e-06,
+      "loss": 0.4732,
+      "step": 20211
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 3.5796944915908045e-06,
+      "loss": 0.3587,
+      "step": 20212
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 3.5771228719847763e-06,
+      "loss": 0.5659,
+      "step": 20213
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 3.574551252378748e-06,
+      "loss": 0.3676,
+      "step": 20214
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 3.5719796327727207e-06,
+      "loss": 0.6481,
+      "step": 20215
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 3.5694080131666925e-06,
+      "loss": 0.5766,
+      "step": 20216
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 3.5668363935606643e-06,
+      "loss": 0.7121,
+      "step": 20217
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 3.564264773954637e-06,
+      "loss": 0.8591,
+      "step": 20218
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 3.5616931543486088e-06,
+      "loss": 0.4478,
+      "step": 20219
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 3.5591215347425806e-06,
+      "loss": 0.9174,
+      "step": 20220
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 3.5565499151365532e-06,
+      "loss": 0.6673,
+      "step": 20221
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 3.553978295530525e-06,
+      "loss": 0.7992,
+      "step": 20222
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 3.5514066759244973e-06,
+      "loss": 0.6623,
+      "step": 20223
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 3.5488350563184695e-06,
+      "loss": 0.4581,
+      "step": 20224
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 3.5462634367124417e-06,
+      "loss": 0.4069,
+      "step": 20225
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 3.5436918171064135e-06,
+      "loss": 0.6614,
+      "step": 20226
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 3.541120197500386e-06,
+      "loss": 0.6013,
+      "step": 20227
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 3.538548577894358e-06,
+      "loss": 0.8814,
+      "step": 20228
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 3.5359769582883306e-06,
+      "loss": 0.6007,
+      "step": 20229
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 3.5334053386823024e-06,
+      "loss": 0.3772,
+      "step": 20230
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 3.5308337190762742e-06,
+      "loss": 0.7662,
+      "step": 20231
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 3.528262099470247e-06,
+      "loss": 0.7819,
+      "step": 20232
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 3.5256904798642187e-06,
+      "loss": 0.615,
+      "step": 20233
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 3.5231188602581905e-06,
+      "loss": 0.5788,
+      "step": 20234
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 3.520547240652163e-06,
+      "loss": 0.7748,
+      "step": 20235
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 3.517975621046135e-06,
+      "loss": 0.3504,
+      "step": 20236
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 3.515404001440107e-06,
+      "loss": 0.59,
+      "step": 20237
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 3.5128323818340794e-06,
+      "loss": 0.4969,
+      "step": 20238
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 3.5102607622280516e-06,
+      "loss": 0.4934,
+      "step": 20239
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 3.5076891426220234e-06,
+      "loss": 0.8374,
+      "step": 20240
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 3.505117523015996e-06,
+      "loss": 0.4894,
+      "step": 20241
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 3.502545903409968e-06,
+      "loss": 0.5285,
+      "step": 20242
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 3.4999742838039397e-06,
+      "loss": 0.971,
+      "step": 20243
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 3.4974026641979123e-06,
+      "loss": 0.3888,
+      "step": 20244
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 3.494831044591884e-06,
+      "loss": 0.4418,
+      "step": 20245
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 3.492259424985856e-06,
+      "loss": 0.5248,
+      "step": 20246
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 3.4896878053798286e-06,
+      "loss": 0.9335,
+      "step": 20247
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 3.4871161857738004e-06,
+      "loss": 0.5675,
+      "step": 20248
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 3.4845445661677726e-06,
+      "loss": 0.3933,
+      "step": 20249
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 3.481972946561745e-06,
+      "loss": 0.5724,
+      "step": 20250
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 3.479401326955717e-06,
+      "loss": 0.4624,
+      "step": 20251
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 3.476829707349689e-06,
+      "loss": 0.9325,
+      "step": 20252
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 3.4742580877436615e-06,
+      "loss": 0.7743,
+      "step": 20253
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 3.4716864681376333e-06,
+      "loss": 0.6028,
+      "step": 20254
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 3.469114848531605e-06,
+      "loss": 0.7974,
+      "step": 20255
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 3.4665432289255778e-06,
+      "loss": 0.8734,
+      "step": 20256
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 3.4639716093195496e-06,
+      "loss": 0.6062,
+      "step": 20257
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 3.4613999897135214e-06,
+      "loss": 0.4989,
+      "step": 20258
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 3.458828370107494e-06,
+      "loss": 0.6675,
+      "step": 20259
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 3.456256750501466e-06,
+      "loss": 0.7736,
+      "step": 20260
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 3.4536851308954376e-06,
+      "loss": 0.691,
+      "step": 20261
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 3.4511135112894103e-06,
+      "loss": 0.6197,
+      "step": 20262
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 3.448541891683382e-06,
+      "loss": 0.4741,
+      "step": 20263
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 3.4459702720773543e-06,
+      "loss": 0.55,
+      "step": 20264
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 3.4433986524713265e-06,
+      "loss": 0.7629,
+      "step": 20265
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 3.4408270328652988e-06,
+      "loss": 0.6051,
+      "step": 20266
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 3.4382554132592714e-06,
+      "loss": 0.4969,
+      "step": 20267
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 3.435683793653243e-06,
+      "loss": 0.5038,
+      "step": 20268
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 3.433112174047215e-06,
+      "loss": 0.4866,
+      "step": 20269
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 3.4305405544411877e-06,
+      "loss": 0.4332,
+      "step": 20270
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 3.4279689348351595e-06,
+      "loss": 0.6566,
+      "step": 20271
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 3.4253973152291313e-06,
+      "loss": 0.5794,
+      "step": 20272
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 3.422825695623104e-06,
+      "loss": 0.6004,
+      "step": 20273
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 3.4202540760170757e-06,
+      "loss": 0.9579,
+      "step": 20274
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 3.4176824564110475e-06,
+      "loss": 0.672,
+      "step": 20275
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 3.41511083680502e-06,
+      "loss": 0.4819,
+      "step": 20276
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 3.412539217198992e-06,
+      "loss": 0.4129,
+      "step": 20277
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 3.409967597592964e-06,
+      "loss": 0.3156,
+      "step": 20278
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 3.4073959779869364e-06,
+      "loss": 0.6434,
+      "step": 20279
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 3.4048243583809086e-06,
+      "loss": 0.3666,
+      "step": 20280
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 3.4022527387748805e-06,
+      "loss": 0.8857,
+      "step": 20281
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 3.399681119168853e-06,
+      "loss": 0.5824,
+      "step": 20282
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 3.397109499562825e-06,
+      "loss": 0.334,
+      "step": 20283
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 3.3945378799567967e-06,
+      "loss": 0.4595,
+      "step": 20284
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 3.3919662603507694e-06,
+      "loss": 0.5472,
+      "step": 20285
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 3.389394640744741e-06,
+      "loss": 0.5895,
+      "step": 20286
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 3.386823021138713e-06,
+      "loss": 0.5233,
+      "step": 20287
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 3.3842514015326856e-06,
+      "loss": 0.9755,
+      "step": 20288
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 3.3816797819266574e-06,
+      "loss": 0.642,
+      "step": 20289
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 3.3791081623206296e-06,
+      "loss": 0.5579,
+      "step": 20290
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 3.376536542714602e-06,
+      "loss": 0.4862,
+      "step": 20291
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 3.373964923108574e-06,
+      "loss": 0.3412,
+      "step": 20292
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 3.371393303502546e-06,
+      "loss": 0.582,
+      "step": 20293
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 3.3688216838965185e-06,
+      "loss": 0.5637,
+      "step": 20294
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 3.3662500642904904e-06,
+      "loss": 0.473,
+      "step": 20295
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 3.363678444684462e-06,
+      "loss": 0.828,
+      "step": 20296
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 3.361106825078435e-06,
+      "loss": 0.4218,
+      "step": 20297
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 3.3585352054724066e-06,
+      "loss": 0.6915,
+      "step": 20298
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 3.3559635858663784e-06,
+      "loss": 0.5885,
+      "step": 20299
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 3.353391966260351e-06,
+      "loss": 0.3497,
+      "step": 20300
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 3.350820346654323e-06,
+      "loss": 0.2806,
+      "step": 20301
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 3.3482487270482947e-06,
+      "loss": 0.7382,
+      "step": 20302
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 3.3456771074422673e-06,
+      "loss": 0.6726,
+      "step": 20303
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 3.343105487836239e-06,
+      "loss": 0.5261,
+      "step": 20304
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 3.3405338682302118e-06,
+      "loss": 0.4347,
+      "step": 20305
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 3.3379622486241836e-06,
+      "loss": 0.3096,
+      "step": 20306
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 3.335390629018156e-06,
+      "loss": 0.543,
+      "step": 20307
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 3.3328190094121284e-06,
+      "loss": 0.5764,
+      "step": 20308
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 3.3302473898061002e-06,
+      "loss": 0.6083,
+      "step": 20309
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 3.327675770200072e-06,
+      "loss": 0.3315,
+      "step": 20310
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 3.3251041505940447e-06,
+      "loss": 0.706,
+      "step": 20311
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 3.3225325309880165e-06,
+      "loss": 0.5521,
+      "step": 20312
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 3.3199609113819883e-06,
+      "loss": 0.5684,
+      "step": 20313
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 3.317389291775961e-06,
+      "loss": 0.5977,
+      "step": 20314
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 3.3148176721699328e-06,
+      "loss": 0.4007,
+      "step": 20315
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 3.3122460525639046e-06,
+      "loss": 0.4541,
+      "step": 20316
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 3.3096744329578772e-06,
+      "loss": 0.3872,
+      "step": 20317
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 3.307102813351849e-06,
+      "loss": 0.8483,
+      "step": 20318
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 3.3045311937458212e-06,
+      "loss": 0.7332,
+      "step": 20319
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 3.3019595741397935e-06,
+      "loss": 0.4021,
+      "step": 20320
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 3.2993879545337657e-06,
+      "loss": 0.4063,
+      "step": 20321
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 3.2968163349277375e-06,
+      "loss": 0.5148,
+      "step": 20322
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 3.29424471532171e-06,
+      "loss": 0.5314,
+      "step": 20323
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 3.291673095715682e-06,
+      "loss": 0.5044,
+      "step": 20324
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 3.2891014761096538e-06,
+      "loss": 0.2578,
+      "step": 20325
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 3.2865298565036264e-06,
+      "loss": 0.6732,
+      "step": 20326
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 3.283958236897598e-06,
+      "loss": 0.6882,
+      "step": 20327
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 3.28138661729157e-06,
+      "loss": 0.5488,
+      "step": 20328
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 3.2788149976855427e-06,
+      "loss": 0.6817,
+      "step": 20329
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 3.2762433780795145e-06,
+      "loss": 0.6489,
+      "step": 20330
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 3.2736717584734867e-06,
+      "loss": 0.8543,
+      "step": 20331
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 3.271100138867459e-06,
+      "loss": 0.4437,
+      "step": 20332
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 3.268528519261431e-06,
+      "loss": 0.6968,
+      "step": 20333
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 3.265956899655403e-06,
+      "loss": 0.4341,
+      "step": 20334
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 3.2633852800493756e-06,
+      "loss": 1.012,
+      "step": 20335
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 3.2608136604433474e-06,
+      "loss": 0.7235,
+      "step": 20336
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 3.258242040837319e-06,
+      "loss": 0.6563,
+      "step": 20337
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 3.255670421231292e-06,
+      "loss": 0.326,
+      "step": 20338
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 3.2530988016252637e-06,
+      "loss": 0.6438,
+      "step": 20339
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 3.2505271820192355e-06,
+      "loss": 0.7567,
+      "step": 20340
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 3.247955562413208e-06,
+      "loss": 0.3916,
+      "step": 20341
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 3.24538394280718e-06,
+      "loss": 0.4468,
+      "step": 20342
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 3.2428123232011526e-06,
+      "loss": 0.7917,
+      "step": 20343
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 3.2402407035951244e-06,
+      "loss": 0.7127,
+      "step": 20344
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 3.237669083989096e-06,
+      "loss": 0.6456,
+      "step": 20345
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 3.235097464383069e-06,
+      "loss": 0.6639,
+      "step": 20346
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 3.2325258447770406e-06,
+      "loss": 0.3432,
+      "step": 20347
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 3.229954225171013e-06,
+      "loss": 0.5205,
+      "step": 20348
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 3.2273826055649855e-06,
+      "loss": 0.416,
+      "step": 20349
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 3.2248109859589573e-06,
+      "loss": 0.6116,
+      "step": 20350
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 3.222239366352929e-06,
+      "loss": 0.7271,
+      "step": 20351
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 3.2196677467469017e-06,
+      "loss": 0.4743,
+      "step": 20352
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 3.2170961271408736e-06,
+      "loss": 0.4517,
+      "step": 20353
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 3.2145245075348454e-06,
+      "loss": 0.7756,
+      "step": 20354
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 3.211952887928818e-06,
+      "loss": 0.6218,
+      "step": 20355
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 3.20938126832279e-06,
+      "loss": 0.5853,
+      "step": 20356
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 3.2068096487167616e-06,
+      "loss": 0.4811,
+      "step": 20357
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 3.2042380291107343e-06,
+      "loss": 0.59,
+      "step": 20358
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 3.201666409504706e-06,
+      "loss": 0.6146,
+      "step": 20359
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 3.1990947898986783e-06,
+      "loss": 0.6828,
+      "step": 20360
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 3.1965231702926505e-06,
+      "loss": 0.5337,
+      "step": 20361
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 3.1939515506866227e-06,
+      "loss": 0.3941,
+      "step": 20362
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 3.1913799310805945e-06,
+      "loss": 0.568,
+      "step": 20363
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 3.188808311474567e-06,
+      "loss": 0.3968,
+      "step": 20364
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 3.186236691868539e-06,
+      "loss": 0.3683,
+      "step": 20365
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 3.183665072262511e-06,
+      "loss": 0.5763,
+      "step": 20366
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 3.1810934526564835e-06,
+      "loss": 0.4879,
+      "step": 20367
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 3.1785218330504553e-06,
+      "loss": 0.4002,
+      "step": 20368
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 3.175950213444427e-06,
+      "loss": 0.7134,
+      "step": 20369
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 3.1733785938383997e-06,
+      "loss": 0.3598,
+      "step": 20370
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 3.1708069742323715e-06,
+      "loss": 0.3956,
+      "step": 20371
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 3.1682353546263433e-06,
+      "loss": 0.7478,
+      "step": 20372
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 3.165663735020316e-06,
+      "loss": 0.3737,
+      "step": 20373
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 3.163092115414288e-06,
+      "loss": 0.5221,
+      "step": 20374
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 3.16052049580826e-06,
+      "loss": 0.5256,
+      "step": 20375
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 3.1579488762022326e-06,
+      "loss": 0.4728,
+      "step": 20376
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 3.1553772565962044e-06,
+      "loss": 0.6503,
+      "step": 20377
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 3.1528056369901762e-06,
+      "loss": 0.5077,
+      "step": 20378
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 3.150234017384149e-06,
+      "loss": 0.6356,
+      "step": 20379
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 3.1476623977781207e-06,
+      "loss": 0.6146,
+      "step": 20380
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 3.1450907781720933e-06,
+      "loss": 0.8738,
+      "step": 20381
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 3.142519158566065e-06,
+      "loss": 0.3241,
+      "step": 20382
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 3.139947538960037e-06,
+      "loss": 0.7603,
+      "step": 20383
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 3.1373759193540096e-06,
+      "loss": 0.6957,
+      "step": 20384
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 3.1348042997479814e-06,
+      "loss": 0.8215,
+      "step": 20385
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 3.132232680141953e-06,
+      "loss": 0.7306,
+      "step": 20386
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 3.129661060535926e-06,
+      "loss": 0.528,
+      "step": 20387
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 3.1270894409298977e-06,
+      "loss": 0.7591,
+      "step": 20388
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 3.12451782132387e-06,
+      "loss": 0.5122,
+      "step": 20389
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 3.121946201717842e-06,
+      "loss": 0.7011,
+      "step": 20390
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 3.1193745821118143e-06,
+      "loss": 0.5712,
+      "step": 20391
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 3.1168029625057866e-06,
+      "loss": 0.5544,
+      "step": 20392
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 3.1142313428997584e-06,
+      "loss": 0.6393,
+      "step": 20393
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 3.1116597232937306e-06,
+      "loss": 0.7888,
+      "step": 20394
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 3.109088103687703e-06,
+      "loss": 0.7122,
+      "step": 20395
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 3.1065164840816746e-06,
+      "loss": 0.57,
+      "step": 20396
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 3.103944864475647e-06,
+      "loss": 0.3199,
+      "step": 20397
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 3.101373244869619e-06,
+      "loss": 0.4628,
+      "step": 20398
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 3.098801625263591e-06,
+      "loss": 0.572,
+      "step": 20399
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 3.096230005657563e-06,
+      "loss": 0.6245,
+      "step": 20400
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 3.0936583860515353e-06,
+      "loss": 0.6598,
+      "step": 20401
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 3.0910867664455076e-06,
+      "loss": 0.592,
+      "step": 20402
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 3.0885151468394798e-06,
+      "loss": 0.6446,
+      "step": 20403
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 3.085943527233452e-06,
+      "loss": 0.8293,
+      "step": 20404
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 3.083371907627424e-06,
+      "loss": 0.9937,
+      "step": 20405
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 3.080800288021396e-06,
+      "loss": 0.405,
+      "step": 20406
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 3.0782286684153683e-06,
+      "loss": 0.5832,
+      "step": 20407
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 3.0756570488093405e-06,
+      "loss": 0.3465,
+      "step": 20408
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 3.0730854292033123e-06,
+      "loss": 0.4676,
+      "step": 20409
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 3.0705138095972845e-06,
+      "loss": 0.4227,
+      "step": 20410
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 3.0679421899912568e-06,
+      "loss": 0.5839,
+      "step": 20411
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 3.0653705703852286e-06,
+      "loss": 0.5993,
+      "step": 20412
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 3.0627989507792008e-06,
+      "loss": 0.5548,
+      "step": 20413
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 3.060227331173173e-06,
+      "loss": 0.3405,
+      "step": 20414
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 3.0576557115671452e-06,
+      "loss": 0.5521,
+      "step": 20415
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 3.0550840919611175e-06,
+      "loss": 0.1486,
+      "step": 20416
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 3.0525124723550897e-06,
+      "loss": 0.5739,
+      "step": 20417
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 3.0499408527490615e-06,
+      "loss": 0.8524,
+      "step": 20418
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 3.0473692331430337e-06,
+      "loss": 0.486,
+      "step": 20419
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 3.044797613537006e-06,
+      "loss": 0.3121,
+      "step": 20420
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 3.0422259939309777e-06,
+      "loss": 0.5943,
+      "step": 20421
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 3.03965437432495e-06,
+      "loss": 0.6178,
+      "step": 20422
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 3.037082754718922e-06,
+      "loss": 1.0377,
+      "step": 20423
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 3.034511135112894e-06,
+      "loss": 0.509,
+      "step": 20424
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 3.0319395155068662e-06,
+      "loss": 0.4173,
+      "step": 20425
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 3.0293678959008385e-06,
+      "loss": 0.5226,
+      "step": 20426
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 3.0267962762948107e-06,
+      "loss": 0.4836,
+      "step": 20427
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 3.0242246566887825e-06,
+      "loss": 0.6492,
+      "step": 20428
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 3.0216530370827547e-06,
+      "loss": 0.9394,
+      "step": 20429
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 3.019081417476727e-06,
+      "loss": 0.729,
+      "step": 20430
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 3.016509797870699e-06,
+      "loss": 0.6241,
+      "step": 20431
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 3.0139381782646714e-06,
+      "loss": 0.6658,
+      "step": 20432
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 3.0113665586586436e-06,
+      "loss": 0.2517,
+      "step": 20433
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 3.0087949390526154e-06,
+      "loss": 0.7902,
+      "step": 20434
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 3.0062233194465876e-06,
+      "loss": 0.3902,
+      "step": 20435
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 3.00365169984056e-06,
+      "loss": 0.4336,
+      "step": 20436
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 3.0010800802345317e-06,
+      "loss": 0.5228,
+      "step": 20437
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 2.998508460628504e-06,
+      "loss": 0.3914,
+      "step": 20438
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 2.995936841022476e-06,
+      "loss": 0.3839,
+      "step": 20439
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 2.993365221416448e-06,
+      "loss": 0.5153,
+      "step": 20440
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 2.99079360181042e-06,
+      "loss": 0.4058,
+      "step": 20441
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 2.9882219822043924e-06,
+      "loss": 0.5867,
+      "step": 20442
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 2.9856503625983646e-06,
+      "loss": 0.5767,
+      "step": 20443
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 2.983078742992337e-06,
+      "loss": 0.2448,
+      "step": 20444
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 2.980507123386309e-06,
+      "loss": 0.3178,
+      "step": 20445
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 2.9779355037802813e-06,
+      "loss": 0.6026,
+      "step": 20446
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 2.975363884174253e-06,
+      "loss": 0.7293,
+      "step": 20447
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 2.9727922645682253e-06,
+      "loss": 0.356,
+      "step": 20448
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 2.9702206449621975e-06,
+      "loss": 0.3028,
+      "step": 20449
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 2.9676490253561693e-06,
+      "loss": 0.7103,
+      "step": 20450
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 2.9650774057501416e-06,
+      "loss": 0.4058,
+      "step": 20451
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 2.962505786144114e-06,
+      "loss": 0.4496,
+      "step": 20452
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 2.9599341665380856e-06,
+      "loss": 0.586,
+      "step": 20453
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 2.957362546932058e-06,
+      "loss": 0.4551,
+      "step": 20454
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 2.95479092732603e-06,
+      "loss": 0.3866,
+      "step": 20455
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 2.9522193077200023e-06,
+      "loss": 0.6606,
+      "step": 20456
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 2.9496476881139745e-06,
+      "loss": 0.603,
+      "step": 20457
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 2.9470760685079467e-06,
+      "loss": 0.5618,
+      "step": 20458
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 2.9445044489019185e-06,
+      "loss": 0.2983,
+      "step": 20459
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 2.9419328292958908e-06,
+      "loss": 0.495,
+      "step": 20460
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 2.939361209689863e-06,
+      "loss": 0.4344,
+      "step": 20461
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 2.936789590083835e-06,
+      "loss": 0.3941,
+      "step": 20462
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 2.934217970477807e-06,
+      "loss": 0.646,
+      "step": 20463
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 2.9316463508717792e-06,
+      "loss": 0.6454,
+      "step": 20464
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 2.9290747312657515e-06,
+      "loss": 0.47,
+      "step": 20465
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 2.9265031116597233e-06,
+      "loss": 0.3859,
+      "step": 20466
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 2.9239314920536955e-06,
+      "loss": 0.4397,
+      "step": 20467
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 2.9213598724476677e-06,
+      "loss": 0.5366,
+      "step": 20468
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 2.9187882528416395e-06,
+      "loss": 0.2875,
+      "step": 20469
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 2.9162166332356118e-06,
+      "loss": 0.692,
+      "step": 20470
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 2.913645013629584e-06,
+      "loss": 0.5877,
+      "step": 20471
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 2.911073394023556e-06,
+      "loss": 0.7068,
+      "step": 20472
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 2.9085017744175284e-06,
+      "loss": 0.5392,
+      "step": 20473
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 2.9059301548115007e-06,
+      "loss": 0.7631,
+      "step": 20474
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 2.9033585352054725e-06,
+      "loss": 0.5548,
+      "step": 20475
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 2.9007869155994447e-06,
+      "loss": 0.5346,
+      "step": 20476
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 2.898215295993417e-06,
+      "loss": 0.4154,
+      "step": 20477
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 2.8956436763873887e-06,
+      "loss": 0.7034,
+      "step": 20478
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 2.893072056781361e-06,
+      "loss": 0.5455,
+      "step": 20479
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 2.890500437175333e-06,
+      "loss": 0.9394,
+      "step": 20480
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 2.887928817569305e-06,
+      "loss": 0.9058,
+      "step": 20481
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 2.885357197963277e-06,
+      "loss": 0.7172,
+      "step": 20482
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 2.8827855783572494e-06,
+      "loss": 0.7947,
+      "step": 20483
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 2.8802139587512217e-06,
+      "loss": 0.7012,
+      "step": 20484
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 2.877642339145194e-06,
+      "loss": 0.5646,
+      "step": 20485
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 2.875070719539166e-06,
+      "loss": 0.5937,
+      "step": 20486
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 2.8724990999331383e-06,
+      "loss": 0.439,
+      "step": 20487
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 2.86992748032711e-06,
+      "loss": 0.7475,
+      "step": 20488
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 2.8673558607210824e-06,
+      "loss": 0.8816,
+      "step": 20489
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 2.8647842411150546e-06,
+      "loss": 0.7577,
+      "step": 20490
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 2.8622126215090264e-06,
+      "loss": 0.435,
+      "step": 20491
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 2.8596410019029986e-06,
+      "loss": 0.6726,
+      "step": 20492
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 2.857069382296971e-06,
+      "loss": 0.5265,
+      "step": 20493
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 2.8544977626909426e-06,
+      "loss": 0.8618,
+      "step": 20494
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 2.851926143084915e-06,
+      "loss": 0.5926,
+      "step": 20495
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 2.849354523478887e-06,
+      "loss": 0.5591,
+      "step": 20496
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 2.8467829038728593e-06,
+      "loss": 0.9345,
+      "step": 20497
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 2.8442112842668316e-06,
+      "loss": 0.5607,
+      "step": 20498
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 2.8416396646608038e-06,
+      "loss": 0.5323,
+      "step": 20499
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 2.8390680450547756e-06,
+      "loss": 0.4323,
+      "step": 20500
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 2.836496425448748e-06,
+      "loss": 1.0191,
+      "step": 20501
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 2.83392480584272e-06,
+      "loss": 0.8285,
+      "step": 20502
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 2.8313531862366923e-06,
+      "loss": 0.603,
+      "step": 20503
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 2.828781566630664e-06,
+      "loss": 0.5719,
+      "step": 20504
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 2.8262099470246363e-06,
+      "loss": 0.6564,
+      "step": 20505
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 2.8236383274186085e-06,
+      "loss": 0.5092,
+      "step": 20506
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 2.8210667078125803e-06,
+      "loss": 0.7123,
+      "step": 20507
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 2.8184950882065525e-06,
+      "loss": 0.5048,
+      "step": 20508
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 2.8159234686005248e-06,
+      "loss": 0.7264,
+      "step": 20509
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 2.8133518489944966e-06,
+      "loss": 0.6102,
+      "step": 20510
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 2.810780229388469e-06,
+      "loss": 0.6815,
+      "step": 20511
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 2.808208609782441e-06,
+      "loss": 0.8275,
+      "step": 20512
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 2.8056369901764133e-06,
+      "loss": 0.5579,
+      "step": 20513
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 2.8030653705703855e-06,
+      "loss": 0.6486,
+      "step": 20514
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 2.8004937509643577e-06,
+      "loss": 0.5019,
+      "step": 20515
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 2.7979221313583295e-06,
+      "loss": 0.6505,
+      "step": 20516
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 2.7953505117523017e-06,
+      "loss": 0.4114,
+      "step": 20517
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 2.792778892146274e-06,
+      "loss": 0.4111,
+      "step": 20518
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 2.7902072725402458e-06,
+      "loss": 0.645,
+      "step": 20519
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 2.787635652934218e-06,
+      "loss": 0.5606,
+      "step": 20520
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 2.7850640333281902e-06,
+      "loss": 0.815,
+      "step": 20521
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 2.7824924137221624e-06,
+      "loss": 0.3655,
+      "step": 20522
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 2.7799207941161342e-06,
+      "loss": 0.3983,
+      "step": 20523
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 2.7773491745101065e-06,
+      "loss": 0.4508,
+      "step": 20524
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 2.7747775549040787e-06,
+      "loss": 0.5252,
+      "step": 20525
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 2.772205935298051e-06,
+      "loss": 0.706,
+      "step": 20526
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 2.769634315692023e-06,
+      "loss": 0.4913,
+      "step": 20527
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 2.7670626960859954e-06,
+      "loss": 0.3522,
+      "step": 20528
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 2.764491076479967e-06,
+      "loss": 0.7161,
+      "step": 20529
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 2.7619194568739394e-06,
+      "loss": 0.7866,
+      "step": 20530
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 2.7593478372679116e-06,
+      "loss": 0.7036,
+      "step": 20531
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 2.7567762176618834e-06,
+      "loss": 0.9057,
+      "step": 20532
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 2.7542045980558557e-06,
+      "loss": 0.7093,
+      "step": 20533
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 2.751632978449828e-06,
+      "loss": 0.7885,
+      "step": 20534
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 2.7490613588437997e-06,
+      "loss": 0.4839,
+      "step": 20535
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 2.746489739237772e-06,
+      "loss": 0.4869,
+      "step": 20536
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 2.743918119631744e-06,
+      "loss": 0.5962,
+      "step": 20537
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 2.7413465000257164e-06,
+      "loss": 0.6127,
+      "step": 20538
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 2.7387748804196886e-06,
+      "loss": 0.4806,
+      "step": 20539
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 2.736203260813661e-06,
+      "loss": 0.5841,
+      "step": 20540
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 2.733631641207633e-06,
+      "loss": 0.2793,
+      "step": 20541
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 2.731060021601605e-06,
+      "loss": 0.3334,
+      "step": 20542
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 2.728488401995577e-06,
+      "loss": 0.5425,
+      "step": 20543
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 2.7259167823895493e-06,
+      "loss": 0.5149,
+      "step": 20544
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 2.723345162783521e-06,
+      "loss": 0.6234,
+      "step": 20545
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 2.7207735431774933e-06,
+      "loss": 0.8114,
+      "step": 20546
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 2.7182019235714656e-06,
+      "loss": 0.4919,
+      "step": 20547
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 2.7156303039654374e-06,
+      "loss": 0.3002,
+      "step": 20548
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 2.7130586843594096e-06,
+      "loss": 0.546,
+      "step": 20549
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 2.710487064753382e-06,
+      "loss": 0.4011,
+      "step": 20550
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 2.7079154451473536e-06,
+      "loss": 0.8284,
+      "step": 20551
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 2.705343825541326e-06,
+      "loss": 0.4805,
+      "step": 20552
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 2.702772205935298e-06,
+      "loss": 0.4468,
+      "step": 20553
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 2.7002005863292703e-06,
+      "loss": 0.4149,
+      "step": 20554
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 2.6976289667232425e-06,
+      "loss": 0.7719,
+      "step": 20555
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 2.6950573471172148e-06,
+      "loss": 0.6765,
+      "step": 20556
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 2.6924857275111866e-06,
+      "loss": 0.5003,
+      "step": 20557
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 2.6899141079051588e-06,
+      "loss": 0.621,
+      "step": 20558
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 2.687342488299131e-06,
+      "loss": 0.6271,
+      "step": 20559
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 2.6847708686931032e-06,
+      "loss": 0.4205,
+      "step": 20560
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 2.682199249087075e-06,
+      "loss": 0.7305,
+      "step": 20561
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 2.6796276294810473e-06,
+      "loss": 0.316,
+      "step": 20562
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 2.6770560098750195e-06,
+      "loss": 0.8164,
+      "step": 20563
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 2.6744843902689913e-06,
+      "loss": 0.4826,
+      "step": 20564
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 2.6719127706629635e-06,
+      "loss": 0.3699,
+      "step": 20565
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 2.6693411510569357e-06,
+      "loss": 0.756,
+      "step": 20566
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 2.666769531450908e-06,
+      "loss": 0.5791,
+      "step": 20567
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 2.66419791184488e-06,
+      "loss": 0.6592,
+      "step": 20568
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 2.6616262922388524e-06,
+      "loss": 0.8198,
+      "step": 20569
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 2.6590546726328242e-06,
+      "loss": 0.7367,
+      "step": 20570
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 2.6564830530267965e-06,
+      "loss": 0.522,
+      "step": 20571
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 2.6539114334207687e-06,
+      "loss": 0.6778,
+      "step": 20572
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 2.6513398138147405e-06,
+      "loss": 0.6938,
+      "step": 20573
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 2.6487681942087127e-06,
+      "loss": 0.5338,
+      "step": 20574
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 2.646196574602685e-06,
+      "loss": 0.7169,
+      "step": 20575
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 2.6436249549966567e-06,
+      "loss": 0.5817,
+      "step": 20576
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 2.641053335390629e-06,
+      "loss": 0.7954,
+      "step": 20577
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 2.638481715784601e-06,
+      "loss": 0.7432,
+      "step": 20578
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 2.6359100961785734e-06,
+      "loss": 0.5838,
+      "step": 20579
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 2.6333384765725456e-06,
+      "loss": 0.8735,
+      "step": 20580
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 2.630766856966518e-06,
+      "loss": 0.818,
+      "step": 20581
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 2.62819523736049e-06,
+      "loss": 0.6128,
+      "step": 20582
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 2.625623617754462e-06,
+      "loss": 0.7191,
+      "step": 20583
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 2.623051998148434e-06,
+      "loss": 0.6449,
+      "step": 20584
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 2.6204803785424064e-06,
+      "loss": 0.4983,
+      "step": 20585
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 2.617908758936378e-06,
+      "loss": 0.5248,
+      "step": 20586
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 2.6153371393303504e-06,
+      "loss": 0.5474,
+      "step": 20587
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 2.6127655197243226e-06,
+      "loss": 0.6234,
+      "step": 20588
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 2.6101939001182944e-06,
+      "loss": 0.347,
+      "step": 20589
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 2.6076222805122666e-06,
+      "loss": 0.6011,
+      "step": 20590
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 2.605050660906239e-06,
+      "loss": 0.7278,
+      "step": 20591
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 2.6024790413002107e-06,
+      "loss": 0.3826,
+      "step": 20592
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 2.599907421694183e-06,
+      "loss": 0.6671,
+      "step": 20593
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 2.597335802088155e-06,
+      "loss": 0.3838,
+      "step": 20594
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 2.5947641824821273e-06,
+      "loss": 0.4698,
+      "step": 20595
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 2.5921925628760996e-06,
+      "loss": 0.3685,
+      "step": 20596
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 2.589620943270072e-06,
+      "loss": 0.5265,
+      "step": 20597
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 2.587049323664044e-06,
+      "loss": 0.502,
+      "step": 20598
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 2.584477704058016e-06,
+      "loss": 0.5679,
+      "step": 20599
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 2.581906084451988e-06,
+      "loss": 0.7323,
+      "step": 20600
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 2.5793344648459603e-06,
+      "loss": 0.4987,
+      "step": 20601
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 2.576762845239932e-06,
+      "loss": 0.7666,
+      "step": 20602
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 2.5741912256339043e-06,
+      "loss": 0.564,
+      "step": 20603
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 2.5716196060278765e-06,
+      "loss": 0.7036,
+      "step": 20604
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 2.5690479864218483e-06,
+      "loss": 0.7254,
+      "step": 20605
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 2.5664763668158206e-06,
+      "loss": 0.6802,
+      "step": 20606
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 2.563904747209793e-06,
+      "loss": 0.5381,
+      "step": 20607
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 2.561333127603765e-06,
+      "loss": 0.9508,
+      "step": 20608
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 2.5587615079977372e-06,
+      "loss": 0.8193,
+      "step": 20609
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 2.5561898883917095e-06,
+      "loss": 0.5716,
+      "step": 20610
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 2.5536182687856813e-06,
+      "loss": 0.6742,
+      "step": 20611
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 2.5510466491796535e-06,
+      "loss": 0.4951,
+      "step": 20612
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 2.5484750295736257e-06,
+      "loss": 0.6543,
+      "step": 20613
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 2.5459034099675975e-06,
+      "loss": 0.3684,
+      "step": 20614
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 2.5433317903615698e-06,
+      "loss": 0.5982,
+      "step": 20615
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 2.540760170755542e-06,
+      "loss": 0.4596,
+      "step": 20616
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 2.538188551149514e-06,
+      "loss": 0.8431,
+      "step": 20617
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 2.535616931543486e-06,
+      "loss": 0.6998,
+      "step": 20618
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 2.5330453119374582e-06,
+      "loss": 0.508,
+      "step": 20619
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 2.5304736923314305e-06,
+      "loss": 0.8337,
+      "step": 20620
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 2.5279020727254027e-06,
+      "loss": 0.2839,
+      "step": 20621
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 2.525330453119375e-06,
+      "loss": 0.5338,
+      "step": 20622
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 2.522758833513347e-06,
+      "loss": 0.5047,
+      "step": 20623
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 2.520187213907319e-06,
+      "loss": 0.7777,
+      "step": 20624
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 2.517615594301291e-06,
+      "loss": 0.7938,
+      "step": 20625
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 2.5150439746952634e-06,
+      "loss": 0.4464,
+      "step": 20626
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 2.512472355089235e-06,
+      "loss": 0.5738,
+      "step": 20627
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 2.5099007354832074e-06,
+      "loss": 0.5493,
+      "step": 20628
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 2.5073291158771797e-06,
+      "loss": 0.5123,
+      "step": 20629
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 2.5047574962711515e-06,
+      "loss": 0.7736,
+      "step": 20630
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 2.5021858766651237e-06,
+      "loss": 0.5473,
+      "step": 20631
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 2.499614257059096e-06,
+      "loss": 0.564,
+      "step": 20632
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 2.4970426374530677e-06,
+      "loss": 0.8775,
+      "step": 20633
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 2.49447101784704e-06,
+      "loss": 0.5779,
+      "step": 20634
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 2.491899398241012e-06,
+      "loss": 0.6046,
+      "step": 20635
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 2.4893277786349844e-06,
+      "loss": 0.6623,
+      "step": 20636
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 2.4867561590289566e-06,
+      "loss": 0.569,
+      "step": 20637
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 2.484184539422929e-06,
+      "loss": 0.3532,
+      "step": 20638
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 2.481612919816901e-06,
+      "loss": 0.4656,
+      "step": 20639
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 2.479041300210873e-06,
+      "loss": 0.3734,
+      "step": 20640
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 2.476469680604845e-06,
+      "loss": 0.4971,
+      "step": 20641
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 2.4738980609988173e-06,
+      "loss": 0.5738,
+      "step": 20642
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 2.471326441392789e-06,
+      "loss": 0.4946,
+      "step": 20643
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 2.4687548217867614e-06,
+      "loss": 0.8363,
+      "step": 20644
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 2.4661832021807336e-06,
+      "loss": 0.7865,
+      "step": 20645
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 2.4636115825747054e-06,
+      "loss": 0.5171,
+      "step": 20646
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 2.4610399629686776e-06,
+      "loss": 0.6673,
+      "step": 20647
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 2.45846834336265e-06,
+      "loss": 0.6569,
+      "step": 20648
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 2.455896723756622e-06,
+      "loss": 0.3369,
+      "step": 20649
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 2.4533251041505943e-06,
+      "loss": 0.5189,
+      "step": 20650
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 2.4507534845445665e-06,
+      "loss": 0.5799,
+      "step": 20651
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 2.4481818649385383e-06,
+      "loss": 0.619,
+      "step": 20652
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 2.4456102453325105e-06,
+      "loss": 0.4048,
+      "step": 20653
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 2.4430386257264828e-06,
+      "loss": 0.7386,
+      "step": 20654
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 2.440467006120455e-06,
+      "loss": 0.3929,
+      "step": 20655
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 2.437895386514427e-06,
+      "loss": 0.4952,
+      "step": 20656
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 2.435323766908399e-06,
+      "loss": 0.5269,
+      "step": 20657
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 2.4327521473023713e-06,
+      "loss": 0.4106,
+      "step": 20658
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 2.430180527696343e-06,
+      "loss": 0.6763,
+      "step": 20659
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 2.4276089080903153e-06,
+      "loss": 0.6668,
+      "step": 20660
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 2.4250372884842875e-06,
+      "loss": 0.6009,
+      "step": 20661
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 2.4224656688782597e-06,
+      "loss": 0.7411,
+      "step": 20662
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 2.419894049272232e-06,
+      "loss": 0.6682,
+      "step": 20663
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 2.417322429666204e-06,
+      "loss": 0.6105,
+      "step": 20664
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 2.414750810060176e-06,
+      "loss": 0.5953,
+      "step": 20665
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 2.4121791904541482e-06,
+      "loss": 0.5176,
+      "step": 20666
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 2.4096075708481204e-06,
+      "loss": 0.7002,
+      "step": 20667
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 2.4070359512420922e-06,
+      "loss": 0.639,
+      "step": 20668
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 2.4044643316360645e-06,
+      "loss": 0.5907,
+      "step": 20669
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 2.4018927120300367e-06,
+      "loss": 0.7216,
+      "step": 20670
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 2.3993210924240085e-06,
+      "loss": 0.7155,
+      "step": 20671
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 2.3967494728179807e-06,
+      "loss": 0.4186,
+      "step": 20672
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 2.394177853211953e-06,
+      "loss": 0.4744,
+      "step": 20673
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 2.391606233605925e-06,
+      "loss": 0.7064,
+      "step": 20674
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 2.389034613999897e-06,
+      "loss": 0.8466,
+      "step": 20675
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 2.3864629943938692e-06,
+      "loss": 0.8228,
+      "step": 20676
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 2.3838913747878414e-06,
+      "loss": 0.5156,
+      "step": 20677
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 2.3813197551818137e-06,
+      "loss": 0.6049,
+      "step": 20678
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 2.378748135575786e-06,
+      "loss": 0.831,
+      "step": 20679
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 2.376176515969758e-06,
+      "loss": 0.5186,
+      "step": 20680
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 2.37360489636373e-06,
+      "loss": 0.89,
+      "step": 20681
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 2.371033276757702e-06,
+      "loss": 0.4275,
+      "step": 20682
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 2.3684616571516744e-06,
+      "loss": 0.3824,
+      "step": 20683
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 2.365890037545646e-06,
+      "loss": 0.7493,
+      "step": 20684
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 2.3633184179396184e-06,
+      "loss": 0.6403,
+      "step": 20685
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 2.3607467983335906e-06,
+      "loss": 0.57,
+      "step": 20686
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 2.3581751787275624e-06,
+      "loss": 0.5538,
+      "step": 20687
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 2.3556035591215347e-06,
+      "loss": 0.4723,
+      "step": 20688
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 2.353031939515507e-06,
+      "loss": 0.3261,
+      "step": 20689
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 2.350460319909479e-06,
+      "loss": 0.5,
+      "step": 20690
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 2.3478887003034513e-06,
+      "loss": 1.3352,
+      "step": 20691
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 2.3453170806974236e-06,
+      "loss": 0.5995,
+      "step": 20692
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 2.342745461091396e-06,
+      "loss": 0.3807,
+      "step": 20693
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 2.3401738414853676e-06,
+      "loss": 0.5937,
+      "step": 20694
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 2.33760222187934e-06,
+      "loss": 0.5998,
+      "step": 20695
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 2.335030602273312e-06,
+      "loss": 0.618,
+      "step": 20696
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 2.332458982667284e-06,
+      "loss": 0.5421,
+      "step": 20697
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 2.329887363061256e-06,
+      "loss": 0.6639,
+      "step": 20698
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 2.3273157434552283e-06,
+      "loss": 0.557,
+      "step": 20699
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 2.3247441238492e-06,
+      "loss": 0.4324,
+      "step": 20700
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 2.3221725042431723e-06,
+      "loss": 0.4656,
+      "step": 20701
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 2.3196008846371446e-06,
+      "loss": 0.4754,
+      "step": 20702
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 2.3170292650311164e-06,
+      "loss": 0.5005,
+      "step": 20703
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 2.314457645425089e-06,
+      "loss": 0.7056,
+      "step": 20704
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 2.3118860258190612e-06,
+      "loss": 0.7414,
+      "step": 20705
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 2.309314406213033e-06,
+      "loss": 0.2623,
+      "step": 20706
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 2.3067427866070053e-06,
+      "loss": 0.4789,
+      "step": 20707
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 2.3041711670009775e-06,
+      "loss": 0.6992,
+      "step": 20708
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 2.3015995473949493e-06,
+      "loss": 0.5691,
+      "step": 20709
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 2.2990279277889215e-06,
+      "loss": 0.414,
+      "step": 20710
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 2.2964563081828937e-06,
+      "loss": 0.3756,
+      "step": 20711
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 2.293884688576866e-06,
+      "loss": 0.7639,
+      "step": 20712
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 2.2913130689708378e-06,
+      "loss": 0.4562,
+      "step": 20713
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 2.28874144936481e-06,
+      "loss": 0.9065,
+      "step": 20714
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 2.2861698297587822e-06,
+      "loss": 0.5534,
+      "step": 20715
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 2.283598210152754e-06,
+      "loss": 0.6334,
+      "step": 20716
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 2.2810265905467263e-06,
+      "loss": 0.6677,
+      "step": 20717
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 2.2784549709406985e-06,
+      "loss": 0.3402,
+      "step": 20718
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 2.2758833513346707e-06,
+      "loss": 0.7846,
+      "step": 20719
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 2.273311731728643e-06,
+      "loss": 0.6156,
+      "step": 20720
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 2.270740112122615e-06,
+      "loss": 0.564,
+      "step": 20721
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 2.268168492516587e-06,
+      "loss": 0.4657,
+      "step": 20722
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 2.265596872910559e-06,
+      "loss": 0.4928,
+      "step": 20723
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 2.2630252533045314e-06,
+      "loss": 0.5026,
+      "step": 20724
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 2.2604536336985032e-06,
+      "loss": 0.7209,
+      "step": 20725
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 2.2578820140924754e-06,
+      "loss": 0.7686,
+      "step": 20726
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 2.2553103944864477e-06,
+      "loss": 0.6245,
+      "step": 20727
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 2.25273877488042e-06,
+      "loss": 0.7012,
+      "step": 20728
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 2.2501671552743917e-06,
+      "loss": 0.6918,
+      "step": 20729
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 2.247595535668364e-06,
+      "loss": 0.7206,
+      "step": 20730
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 2.245023916062336e-06,
+      "loss": 0.8893,
+      "step": 20731
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 2.2424522964563084e-06,
+      "loss": 0.6348,
+      "step": 20732
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 2.2398806768502806e-06,
+      "loss": 0.7352,
+      "step": 20733
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 2.237309057244253e-06,
+      "loss": 0.5587,
+      "step": 20734
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 2.2347374376382246e-06,
+      "loss": 0.825,
+      "step": 20735
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 2.232165818032197e-06,
+      "loss": 0.6255,
+      "step": 20736
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 2.229594198426169e-06,
+      "loss": 0.5994,
+      "step": 20737
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 2.227022578820141e-06,
+      "loss": 1.0677,
+      "step": 20738
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 2.224450959214113e-06,
+      "loss": 0.3949,
+      "step": 20739
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 2.2218793396080853e-06,
+      "loss": 0.5027,
+      "step": 20740
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 2.219307720002057e-06,
+      "loss": 0.6379,
+      "step": 20741
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 2.2167361003960294e-06,
+      "loss": 0.5627,
+      "step": 20742
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 2.2141644807900016e-06,
+      "loss": 0.2949,
+      "step": 20743
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 2.2115928611839734e-06,
+      "loss": 0.719,
+      "step": 20744
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 2.209021241577946e-06,
+      "loss": 0.399,
+      "step": 20745
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 2.2064496219719183e-06,
+      "loss": 0.2969,
+      "step": 20746
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 2.2038780023658905e-06,
+      "loss": 0.5246,
+      "step": 20747
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 2.2013063827598623e-06,
+      "loss": 0.4674,
+      "step": 20748
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 2.1987347631538345e-06,
+      "loss": 0.3201,
+      "step": 20749
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 2.1961631435478068e-06,
+      "loss": 0.5634,
+      "step": 20750
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 2.1935915239417786e-06,
+      "loss": 0.4453,
+      "step": 20751
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 2.191019904335751e-06,
+      "loss": 0.5859,
+      "step": 20752
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 2.188448284729723e-06,
+      "loss": 0.4369,
+      "step": 20753
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 2.185876665123695e-06,
+      "loss": 0.3457,
+      "step": 20754
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 2.183305045517667e-06,
+      "loss": 0.4575,
+      "step": 20755
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 2.1807334259116393e-06,
+      "loss": 0.3704,
+      "step": 20756
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 2.178161806305611e-06,
+      "loss": 0.6506,
+      "step": 20757
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 2.1755901866995833e-06,
+      "loss": 0.8417,
+      "step": 20758
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 2.1730185670935555e-06,
+      "loss": 0.8377,
+      "step": 20759
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 2.1704469474875278e-06,
+      "loss": 0.4928,
+      "step": 20760
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 2.1678753278815e-06,
+      "loss": 0.4593,
+      "step": 20761
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 2.165303708275472e-06,
+      "loss": 0.8155,
+      "step": 20762
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 2.162732088669444e-06,
+      "loss": 0.5297,
+      "step": 20763
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 2.1601604690634162e-06,
+      "loss": 0.6777,
+      "step": 20764
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 2.1575888494573885e-06,
+      "loss": 0.5577,
+      "step": 20765
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 2.1550172298513607e-06,
+      "loss": 0.6443,
+      "step": 20766
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 2.1524456102453325e-06,
+      "loss": 0.5318,
+      "step": 20767
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 2.1498739906393047e-06,
+      "loss": 0.5284,
+      "step": 20768
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 2.147302371033277e-06,
+      "loss": 0.6391,
+      "step": 20769
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 2.1447307514272488e-06,
+      "loss": 0.6825,
+      "step": 20770
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 2.142159131821221e-06,
+      "loss": 0.492,
+      "step": 20771
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 2.139587512215193e-06,
+      "loss": 0.9549,
+      "step": 20772
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 2.1370158926091654e-06,
+      "loss": 0.6549,
+      "step": 20773
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 2.1344442730031377e-06,
+      "loss": 0.5952,
+      "step": 20774
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 2.13187265339711e-06,
+      "loss": 0.744,
+      "step": 20775
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 2.1293010337910817e-06,
+      "loss": 0.7504,
+      "step": 20776
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 2.126729414185054e-06,
+      "loss": 0.5061,
+      "step": 20777
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 2.124157794579026e-06,
+      "loss": 0.6654,
+      "step": 20778
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 2.121586174972998e-06,
+      "loss": 0.536,
+      "step": 20779
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 2.11901455536697e-06,
+      "loss": 0.6072,
+      "step": 20780
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 2.1164429357609424e-06,
+      "loss": 0.2884,
+      "step": 20781
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 2.113871316154914e-06,
+      "loss": 0.5031,
+      "step": 20782
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 2.1112996965488864e-06,
+      "loss": 0.5293,
+      "step": 20783
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 2.1087280769428587e-06,
+      "loss": 0.8528,
+      "step": 20784
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 2.106156457336831e-06,
+      "loss": 0.5607,
+      "step": 20785
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 2.103584837730803e-06,
+      "loss": 0.3142,
+      "step": 20786
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 2.1010132181247753e-06,
+      "loss": 0.5452,
+      "step": 20787
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 2.0984415985187476e-06,
+      "loss": 0.5305,
+      "step": 20788
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 2.0958699789127194e-06,
+      "loss": 0.6106,
+      "step": 20789
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 2.0932983593066916e-06,
+      "loss": 0.7336,
+      "step": 20790
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 2.090726739700664e-06,
+      "loss": 0.7714,
+      "step": 20791
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 2.0881551200946356e-06,
+      "loss": 0.4939,
+      "step": 20792
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 2.085583500488608e-06,
+      "loss": 0.3306,
+      "step": 20793
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 2.08301188088258e-06,
+      "loss": 0.5544,
+      "step": 20794
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 2.080440261276552e-06,
+      "loss": 0.7643,
+      "step": 20795
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 2.077868641670524e-06,
+      "loss": 0.874,
+      "step": 20796
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 2.0752970220644963e-06,
+      "loss": 1.153,
+      "step": 20797
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 2.072725402458468e-06,
+      "loss": 0.6219,
+      "step": 20798
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 2.0701537828524404e-06,
+      "loss": 0.6631,
+      "step": 20799
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 2.0675821632464126e-06,
+      "loss": 0.3477,
+      "step": 20800
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 2.065010543640385e-06,
+      "loss": 0.7209,
+      "step": 20801
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 2.062438924034357e-06,
+      "loss": 0.5067,
+      "step": 20802
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 2.0598673044283293e-06,
+      "loss": 0.5143,
+      "step": 20803
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 2.0572956848223015e-06,
+      "loss": 0.8318,
+      "step": 20804
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 2.0547240652162733e-06,
+      "loss": 0.5823,
+      "step": 20805
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 2.0521524456102455e-06,
+      "loss": 1.1105,
+      "step": 20806
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 2.0495808260042177e-06,
+      "loss": 0.5413,
+      "step": 20807
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 2.0470092063981895e-06,
+      "loss": 0.4915,
+      "step": 20808
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 2.0444375867921618e-06,
+      "loss": 0.5663,
+      "step": 20809
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 2.041865967186134e-06,
+      "loss": 0.6548,
+      "step": 20810
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 2.039294347580106e-06,
+      "loss": 0.805,
+      "step": 20811
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 2.036722727974078e-06,
+      "loss": 0.5508,
+      "step": 20812
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 2.0341511083680503e-06,
+      "loss": 0.5226,
+      "step": 20813
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 2.0315794887620225e-06,
+      "loss": 0.4568,
+      "step": 20814
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 2.0290078691559947e-06,
+      "loss": 0.2645,
+      "step": 20815
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 2.026436249549967e-06,
+      "loss": 0.6194,
+      "step": 20816
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 2.0238646299439387e-06,
+      "loss": 0.7856,
+      "step": 20817
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 2.021293010337911e-06,
+      "loss": 0.5497,
+      "step": 20818
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 2.018721390731883e-06,
+      "loss": 1.0804,
+      "step": 20819
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 2.016149771125855e-06,
+      "loss": 0.5393,
+      "step": 20820
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 2.0135781515198272e-06,
+      "loss": 0.4784,
+      "step": 20821
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 2.0110065319137994e-06,
+      "loss": 0.3845,
+      "step": 20822
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 2.0084349123077717e-06,
+      "loss": 0.5494,
+      "step": 20823
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 2.0058632927017435e-06,
+      "loss": 0.3882,
+      "step": 20824
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 2.0032916730957157e-06,
+      "loss": 0.5654,
+      "step": 20825
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 2.000720053489688e-06,
+      "loss": 0.5108,
+      "step": 20826
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 1.9981484338836597e-06,
+      "loss": 0.5499,
+      "step": 20827
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 1.9955768142776324e-06,
+      "loss": 0.5041,
+      "step": 20828
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 1.9930051946716046e-06,
+      "loss": 0.5181,
+      "step": 20829
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 1.9904335750655764e-06,
+      "loss": 0.7129,
+      "step": 20830
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 1.9878619554595486e-06,
+      "loss": 0.4978,
+      "step": 20831
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 1.985290335853521e-06,
+      "loss": 0.3632,
+      "step": 20832
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 1.9827187162474927e-06,
+      "loss": 0.4956,
+      "step": 20833
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 1.980147096641465e-06,
+      "loss": 0.8084,
+      "step": 20834
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 1.977575477035437e-06,
+      "loss": 0.6998,
+      "step": 20835
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 1.975003857429409e-06,
+      "loss": 0.4286,
+      "step": 20836
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 1.972432237823381e-06,
+      "loss": 0.4837,
+      "step": 20837
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 1.9698606182173534e-06,
+      "loss": 0.4039,
+      "step": 20838
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 1.967288998611325e-06,
+      "loss": 0.4422,
+      "step": 20839
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 1.9647173790052974e-06,
+      "loss": 0.5372,
+      "step": 20840
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 1.9621457593992696e-06,
+      "loss": 0.5194,
+      "step": 20841
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 1.959574139793242e-06,
+      "loss": 0.7531,
+      "step": 20842
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 1.957002520187214e-06,
+      "loss": 0.6547,
+      "step": 20843
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 1.9544309005811863e-06,
+      "loss": 0.5816,
+      "step": 20844
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 1.9518592809751585e-06,
+      "loss": 0.7068,
+      "step": 20845
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 1.9492876613691303e-06,
+      "loss": 0.7174,
+      "step": 20846
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 1.9467160417631026e-06,
+      "loss": 0.4295,
+      "step": 20847
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 1.9441444221570748e-06,
+      "loss": 0.6573,
+      "step": 20848
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 1.9415728025510466e-06,
+      "loss": 0.6291,
+      "step": 20849
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 1.939001182945019e-06,
+      "loss": 0.3599,
+      "step": 20850
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 1.936429563338991e-06,
+      "loss": 0.5608,
+      "step": 20851
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 1.933857943732963e-06,
+      "loss": 0.4389,
+      "step": 20852
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 1.931286324126935e-06,
+      "loss": 0.6309,
+      "step": 20853
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 1.9287147045209073e-06,
+      "loss": 0.4181,
+      "step": 20854
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 1.9261430849148795e-06,
+      "loss": 0.5236,
+      "step": 20855
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 1.9235714653088517e-06,
+      "loss": 0.624,
+      "step": 20856
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 1.920999845702824e-06,
+      "loss": 0.474,
+      "step": 20857
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 1.9184282260967958e-06,
+      "loss": 0.7492,
+      "step": 20858
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 1.915856606490768e-06,
+      "loss": 0.3865,
+      "step": 20859
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 1.9132849868847402e-06,
+      "loss": 0.3495,
+      "step": 20860
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 1.9107133672787125e-06,
+      "loss": 0.6909,
+      "step": 20861
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 1.9081417476726843e-06,
+      "loss": 0.7143,
+      "step": 20862
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 1.9055701280666565e-06,
+      "loss": 0.3984,
+      "step": 20863
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 1.9029985084606287e-06,
+      "loss": 0.2777,
+      "step": 20864
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 1.9004268888546007e-06,
+      "loss": 0.4448,
+      "step": 20865
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 1.897855269248573e-06,
+      "loss": 0.8224,
+      "step": 20866
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 1.8952836496425452e-06,
+      "loss": 0.8518,
+      "step": 20867
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 1.892712030036517e-06,
+      "loss": 0.4884,
+      "step": 20868
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 1.8901404104304892e-06,
+      "loss": 0.8154,
+      "step": 20869
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 1.8875687908244614e-06,
+      "loss": 0.4754,
+      "step": 20870
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 1.8849971712184332e-06,
+      "loss": 0.5175,
+      "step": 20871
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 1.8824255516124055e-06,
+      "loss": 0.6231,
+      "step": 20872
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 1.879853932006378e-06,
+      "loss": 0.5564,
+      "step": 20873
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 1.8772823124003497e-06,
+      "loss": 0.6351,
+      "step": 20874
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 1.874710692794322e-06,
+      "loss": 0.4363,
+      "step": 20875
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 1.8721390731882942e-06,
+      "loss": 0.5034,
+      "step": 20876
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 1.869567453582266e-06,
+      "loss": 0.7986,
+      "step": 20877
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 1.8669958339762382e-06,
+      "loss": 0.4183,
+      "step": 20878
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 1.8644242143702104e-06,
+      "loss": 0.5234,
+      "step": 20879
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 1.8618525947641826e-06,
+      "loss": 0.3408,
+      "step": 20880
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 1.8592809751581547e-06,
+      "loss": 0.6922,
+      "step": 20881
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 1.8567093555521269e-06,
+      "loss": 0.7356,
+      "step": 20882
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 1.8541377359460991e-06,
+      "loss": 0.9963,
+      "step": 20883
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 1.851566116340071e-06,
+      "loss": 0.6139,
+      "step": 20884
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 1.8489944967340431e-06,
+      "loss": 0.3062,
+      "step": 20885
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 1.8464228771280154e-06,
+      "loss": 0.4929,
+      "step": 20886
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 1.8438512575219874e-06,
+      "loss": 0.7386,
+      "step": 20887
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 1.8412796379159596e-06,
+      "loss": 0.7865,
+      "step": 20888
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 1.8387080183099318e-06,
+      "loss": 0.273,
+      "step": 20889
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 1.8361363987039036e-06,
+      "loss": 0.3173,
+      "step": 20890
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 1.8335647790978759e-06,
+      "loss": 0.3947,
+      "step": 20891
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 1.830993159491848e-06,
+      "loss": 0.4781,
+      "step": 20892
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 1.82842153988582e-06,
+      "loss": 0.6802,
+      "step": 20893
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 1.8258499202797923e-06,
+      "loss": 0.4618,
+      "step": 20894
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 1.8232783006737646e-06,
+      "loss": 0.64,
+      "step": 20895
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 1.8207066810677364e-06,
+      "loss": 0.4872,
+      "step": 20896
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 1.8181350614617086e-06,
+      "loss": 0.4871,
+      "step": 20897
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 1.8155634418556808e-06,
+      "loss": 0.6393,
+      "step": 20898
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 1.812991822249653e-06,
+      "loss": 0.5594,
+      "step": 20899
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 1.810420202643625e-06,
+      "loss": 0.6914,
+      "step": 20900
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 1.8078485830375973e-06,
+      "loss": 0.4832,
+      "step": 20901
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 1.8052769634315695e-06,
+      "loss": 0.6746,
+      "step": 20902
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 1.8027053438255413e-06,
+      "loss": 0.6901,
+      "step": 20903
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 1.8001337242195135e-06,
+      "loss": 0.6701,
+      "step": 20904
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 1.7975621046134858e-06,
+      "loss": 0.6002,
+      "step": 20905
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 1.7949904850074578e-06,
+      "loss": 0.6437,
+      "step": 20906
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 1.79241886540143e-06,
+      "loss": 0.6407,
+      "step": 20907
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 1.7898472457954022e-06,
+      "loss": 0.4615,
+      "step": 20908
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 1.787275626189374e-06,
+      "loss": 0.7982,
+      "step": 20909
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 1.7847040065833463e-06,
+      "loss": 0.5956,
+      "step": 20910
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 1.7821323869773185e-06,
+      "loss": 0.5257,
+      "step": 20911
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 1.7795607673712903e-06,
+      "loss": 0.4748,
+      "step": 20912
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 1.7769891477652625e-06,
+      "loss": 0.6295,
+      "step": 20913
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 1.7744175281592347e-06,
+      "loss": 0.4798,
+      "step": 20914
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 1.7718459085532068e-06,
+      "loss": 0.4981,
+      "step": 20915
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 1.769274288947179e-06,
+      "loss": 0.6026,
+      "step": 20916
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 1.7667026693411512e-06,
+      "loss": 0.7864,
+      "step": 20917
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 1.7641310497351234e-06,
+      "loss": 0.7349,
+      "step": 20918
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 1.7615594301290952e-06,
+      "loss": 0.665,
+      "step": 20919
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 1.7589878105230675e-06,
+      "loss": 0.7499,
+      "step": 20920
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 1.7564161909170397e-06,
+      "loss": 0.6533,
+      "step": 20921
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 1.7538445713110117e-06,
+      "loss": 0.6587,
+      "step": 20922
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 1.751272951704984e-06,
+      "loss": 0.6783,
+      "step": 20923
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 1.7487013320989562e-06,
+      "loss": 0.4659,
+      "step": 20924
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 1.746129712492928e-06,
+      "loss": 0.6203,
+      "step": 20925
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 1.7435580928869002e-06,
+      "loss": 0.6304,
+      "step": 20926
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 1.7409864732808724e-06,
+      "loss": 0.4843,
+      "step": 20927
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 1.7384148536748444e-06,
+      "loss": 0.3433,
+      "step": 20928
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 1.7358432340688167e-06,
+      "loss": 0.5439,
+      "step": 20929
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 1.7332716144627889e-06,
+      "loss": 1.0202,
+      "step": 20930
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 1.7306999948567607e-06,
+      "loss": 0.5445,
+      "step": 20931
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 1.728128375250733e-06,
+      "loss": 0.5093,
+      "step": 20932
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 1.7255567556447051e-06,
+      "loss": 0.4782,
+      "step": 20933
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 1.7229851360386771e-06,
+      "loss": 0.4504,
+      "step": 20934
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 1.7204135164326494e-06,
+      "loss": 0.5202,
+      "step": 20935
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 1.7178418968266216e-06,
+      "loss": 0.8984,
+      "step": 20936
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 1.7152702772205938e-06,
+      "loss": 0.3196,
+      "step": 20937
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 1.7126986576145656e-06,
+      "loss": 0.6524,
+      "step": 20938
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 1.7101270380085379e-06,
+      "loss": 0.9311,
+      "step": 20939
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 1.70755541840251e-06,
+      "loss": 0.7232,
+      "step": 20940
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 1.704983798796482e-06,
+      "loss": 0.7047,
+      "step": 20941
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 1.7024121791904543e-06,
+      "loss": 0.4757,
+      "step": 20942
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 1.6998405595844266e-06,
+      "loss": 0.597,
+      "step": 20943
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 1.6972689399783984e-06,
+      "loss": 0.5287,
+      "step": 20944
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 1.6946973203723706e-06,
+      "loss": 0.8233,
+      "step": 20945
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 1.6921257007663428e-06,
+      "loss": 0.3899,
+      "step": 20946
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 1.6895540811603148e-06,
+      "loss": 0.4582,
+      "step": 20947
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 1.686982461554287e-06,
+      "loss": 0.3187,
+      "step": 20948
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 1.6844108419482593e-06,
+      "loss": 0.322,
+      "step": 20949
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 1.681839222342231e-06,
+      "loss": 0.4832,
+      "step": 20950
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 1.6792676027362033e-06,
+      "loss": 0.5611,
+      "step": 20951
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 1.6766959831301755e-06,
+      "loss": 0.4825,
+      "step": 20952
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 1.6741243635241473e-06,
+      "loss": 0.9066,
+      "step": 20953
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 1.6715527439181196e-06,
+      "loss": 0.7415,
+      "step": 20954
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 1.6689811243120918e-06,
+      "loss": 0.5404,
+      "step": 20955
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 1.6664095047060642e-06,
+      "loss": 0.6516,
+      "step": 20956
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 1.663837885100036e-06,
+      "loss": 0.7853,
+      "step": 20957
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 1.6612662654940083e-06,
+      "loss": 0.4379,
+      "step": 20958
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 1.6586946458879805e-06,
+      "loss": 0.3268,
+      "step": 20959
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 1.6561230262819523e-06,
+      "loss": 0.6931,
+      "step": 20960
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 1.6535514066759245e-06,
+      "loss": 0.1677,
+      "step": 20961
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 1.6509797870698967e-06,
+      "loss": 0.8946,
+      "step": 20962
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 1.6484081674638687e-06,
+      "loss": 0.5582,
+      "step": 20963
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 1.645836547857841e-06,
+      "loss": 0.7922,
+      "step": 20964
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 1.6432649282518132e-06,
+      "loss": 0.6436,
+      "step": 20965
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 1.640693308645785e-06,
+      "loss": 0.5945,
+      "step": 20966
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 1.6381216890397572e-06,
+      "loss": 0.7273,
+      "step": 20967
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 1.6355500694337295e-06,
+      "loss": 0.4736,
+      "step": 20968
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 1.6329784498277015e-06,
+      "loss": 0.692,
+      "step": 20969
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 1.6304068302216737e-06,
+      "loss": 0.7055,
+      "step": 20970
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 1.627835210615646e-06,
+      "loss": 0.4575,
+      "step": 20971
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 1.6252635910096177e-06,
+      "loss": 0.6972,
+      "step": 20972
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 1.62269197140359e-06,
+      "loss": 0.4337,
+      "step": 20973
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 1.6201203517975622e-06,
+      "loss": 0.62,
+      "step": 20974
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 1.6175487321915344e-06,
+      "loss": 0.3904,
+      "step": 20975
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 1.6149771125855064e-06,
+      "loss": 0.5534,
+      "step": 20976
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 1.6124054929794786e-06,
+      "loss": 0.3688,
+      "step": 20977
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 1.6098338733734509e-06,
+      "loss": 0.872,
+      "step": 20978
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 1.6072622537674227e-06,
+      "loss": 0.7276,
+      "step": 20979
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 1.604690634161395e-06,
+      "loss": 0.7487,
+      "step": 20980
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 1.6021190145553671e-06,
+      "loss": 0.9393,
+      "step": 20981
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 1.5995473949493391e-06,
+      "loss": 0.5942,
+      "step": 20982
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 1.5969757753433114e-06,
+      "loss": 0.4947,
+      "step": 20983
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 1.5944041557372836e-06,
+      "loss": 0.7724,
+      "step": 20984
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 1.5918325361312554e-06,
+      "loss": 0.4093,
+      "step": 20985
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 1.5892609165252276e-06,
+      "loss": 0.5194,
+      "step": 20986
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 1.5866892969191999e-06,
+      "loss": 0.556,
+      "step": 20987
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 1.5841176773131717e-06,
+      "loss": 0.7395,
+      "step": 20988
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 1.581546057707144e-06,
+      "loss": 0.4685,
+      "step": 20989
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 1.5789744381011163e-06,
+      "loss": 0.5927,
+      "step": 20990
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 1.5764028184950881e-06,
+      "loss": 0.317,
+      "step": 20991
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 1.5738311988890603e-06,
+      "loss": 0.7859,
+      "step": 20992
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 1.5712595792830326e-06,
+      "loss": 0.9209,
+      "step": 20993
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 1.5686879596770048e-06,
+      "loss": 0.694,
+      "step": 20994
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 1.5661163400709766e-06,
+      "loss": 0.6699,
+      "step": 20995
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 1.5635447204649488e-06,
+      "loss": 0.7562,
+      "step": 20996
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 1.560973100858921e-06,
+      "loss": 0.4768,
+      "step": 20997
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 1.5584014812528933e-06,
+      "loss": 0.6443,
+      "step": 20998
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 1.5558298616468653e-06,
+      "loss": 0.6067,
+      "step": 20999
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 1.5532582420408373e-06,
+      "loss": 0.6751,
+      "step": 21000
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 1.5506866224348095e-06,
+      "loss": 0.5916,
+      "step": 21001
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 1.5481150028287816e-06,
+      "loss": 0.6902,
+      "step": 21002
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 1.5455433832227538e-06,
+      "loss": 0.7286,
+      "step": 21003
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 1.542971763616726e-06,
+      "loss": 0.669,
+      "step": 21004
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 1.540400144010698e-06,
+      "loss": 0.6105,
+      "step": 21005
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 1.5378285244046702e-06,
+      "loss": 0.7494,
+      "step": 21006
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 1.5352569047986423e-06,
+      "loss": 0.4263,
+      "step": 21007
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 1.5326852851926143e-06,
+      "loss": 0.9551,
+      "step": 21008
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 1.5301136655865865e-06,
+      "loss": 0.3959,
+      "step": 21009
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 1.5275420459805587e-06,
+      "loss": 0.5469,
+      "step": 21010
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 1.5249704263745307e-06,
+      "loss": 0.6676,
+      "step": 21011
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 1.522398806768503e-06,
+      "loss": 0.526,
+      "step": 21012
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 1.519827187162475e-06,
+      "loss": 0.6601,
+      "step": 21013
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 1.517255567556447e-06,
+      "loss": 0.4081,
+      "step": 21014
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 1.5146839479504192e-06,
+      "loss": 0.3876,
+      "step": 21015
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 1.5121123283443912e-06,
+      "loss": 0.4664,
+      "step": 21016
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 1.5095407087383635e-06,
+      "loss": 0.4253,
+      "step": 21017
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 1.5069690891323357e-06,
+      "loss": 0.4871,
+      "step": 21018
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 1.5043974695263077e-06,
+      "loss": 0.6522,
+      "step": 21019
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 1.50182584992028e-06,
+      "loss": 0.7045,
+      "step": 21020
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 1.499254230314252e-06,
+      "loss": 0.7206,
+      "step": 21021
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 1.496682610708224e-06,
+      "loss": 0.4224,
+      "step": 21022
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 1.4941109911021962e-06,
+      "loss": 0.6698,
+      "step": 21023
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 1.4915393714961684e-06,
+      "loss": 0.5028,
+      "step": 21024
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 1.4889677518901406e-06,
+      "loss": 0.5569,
+      "step": 21025
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 1.4863961322841127e-06,
+      "loss": 0.6798,
+      "step": 21026
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 1.4838245126780847e-06,
+      "loss": 0.6045,
+      "step": 21027
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 1.481252893072057e-06,
+      "loss": 0.7138,
+      "step": 21028
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 1.478681273466029e-06,
+      "loss": 0.6402,
+      "step": 21029
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 1.4761096538600011e-06,
+      "loss": 0.7694,
+      "step": 21030
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 1.4735380342539734e-06,
+      "loss": 0.4927,
+      "step": 21031
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 1.4709664146479454e-06,
+      "loss": 0.8274,
+      "step": 21032
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 1.4683947950419174e-06,
+      "loss": 0.5784,
+      "step": 21033
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 1.4658231754358896e-06,
+      "loss": 0.6058,
+      "step": 21034
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 1.4632515558298616e-06,
+      "loss": 0.2743,
+      "step": 21035
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 1.4606799362238339e-06,
+      "loss": 0.429,
+      "step": 21036
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 1.4581083166178059e-06,
+      "loss": 0.832,
+      "step": 21037
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 1.455536697011778e-06,
+      "loss": 0.8358,
+      "step": 21038
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 1.4529650774057503e-06,
+      "loss": 0.5241,
+      "step": 21039
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 1.4503934577997223e-06,
+      "loss": 0.7034,
+      "step": 21040
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 1.4478218381936944e-06,
+      "loss": 0.6319,
+      "step": 21041
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 1.4452502185876666e-06,
+      "loss": 0.4325,
+      "step": 21042
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 1.4426785989816386e-06,
+      "loss": 0.6943,
+      "step": 21043
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 1.4401069793756108e-06,
+      "loss": 0.5688,
+      "step": 21044
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 1.437535359769583e-06,
+      "loss": 0.2903,
+      "step": 21045
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 1.434963740163555e-06,
+      "loss": 0.4795,
+      "step": 21046
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 1.4323921205575273e-06,
+      "loss": 0.6208,
+      "step": 21047
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 1.4298205009514993e-06,
+      "loss": 0.4519,
+      "step": 21048
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 1.4272488813454713e-06,
+      "loss": 0.3282,
+      "step": 21049
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 1.4246772617394436e-06,
+      "loss": 0.6735,
+      "step": 21050
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 1.4221056421334158e-06,
+      "loss": 0.5164,
+      "step": 21051
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 1.4195340225273878e-06,
+      "loss": 0.4755,
+      "step": 21052
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 1.41696240292136e-06,
+      "loss": 0.6879,
+      "step": 21053
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 1.414390783315332e-06,
+      "loss": 0.3543,
+      "step": 21054
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 1.4118191637093043e-06,
+      "loss": 0.5083,
+      "step": 21055
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 1.4092475441032763e-06,
+      "loss": 0.8173,
+      "step": 21056
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 1.4066759244972483e-06,
+      "loss": 0.5613,
+      "step": 21057
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 1.4041043048912205e-06,
+      "loss": 0.6964,
+      "step": 21058
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 1.4015326852851927e-06,
+      "loss": 0.7861,
+      "step": 21059
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 1.3989610656791648e-06,
+      "loss": 0.4837,
+      "step": 21060
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 1.396389446073137e-06,
+      "loss": 0.6275,
+      "step": 21061
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 1.393817826467109e-06,
+      "loss": 0.4801,
+      "step": 21062
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 1.3912462068610812e-06,
+      "loss": 0.913,
+      "step": 21063
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 1.3886745872550532e-06,
+      "loss": 0.4278,
+      "step": 21064
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 1.3861029676490255e-06,
+      "loss": 0.554,
+      "step": 21065
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 1.3835313480429977e-06,
+      "loss": 0.5263,
+      "step": 21066
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 1.3809597284369697e-06,
+      "loss": 0.5607,
+      "step": 21067
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 1.3783881088309417e-06,
+      "loss": 0.5272,
+      "step": 21068
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 1.375816489224914e-06,
+      "loss": 0.4485,
+      "step": 21069
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 1.373244869618886e-06,
+      "loss": 0.4735,
+      "step": 21070
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 1.3706732500128582e-06,
+      "loss": 0.3844,
+      "step": 21071
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 1.3681016304068304e-06,
+      "loss": 0.5761,
+      "step": 21072
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 1.3655300108008024e-06,
+      "loss": 0.5159,
+      "step": 21073
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 1.3629583911947747e-06,
+      "loss": 0.661,
+      "step": 21074
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 1.3603867715887467e-06,
+      "loss": 0.5203,
+      "step": 21075
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 1.3578151519827187e-06,
+      "loss": 0.5924,
+      "step": 21076
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 1.355243532376691e-06,
+      "loss": 0.4885,
+      "step": 21077
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 1.352671912770663e-06,
+      "loss": 0.8141,
+      "step": 21078
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 1.3501002931646352e-06,
+      "loss": 0.5892,
+      "step": 21079
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 1.3475286735586074e-06,
+      "loss": 0.5988,
+      "step": 21080
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 1.3449570539525794e-06,
+      "loss": 0.3626,
+      "step": 21081
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 1.3423854343465516e-06,
+      "loss": 0.4191,
+      "step": 21082
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 1.3398138147405236e-06,
+      "loss": 0.4483,
+      "step": 21083
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 1.3372421951344956e-06,
+      "loss": 0.7559,
+      "step": 21084
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 1.3346705755284679e-06,
+      "loss": 0.7993,
+      "step": 21085
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 1.33209895592244e-06,
+      "loss": 0.7022,
+      "step": 21086
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 1.3295273363164121e-06,
+      "loss": 0.6597,
+      "step": 21087
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 1.3269557167103843e-06,
+      "loss": 0.6032,
+      "step": 21088
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 1.3243840971043564e-06,
+      "loss": 0.6941,
+      "step": 21089
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 1.3218124774983284e-06,
+      "loss": 0.6814,
+      "step": 21090
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 1.3192408578923006e-06,
+      "loss": 0.4229,
+      "step": 21091
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 1.3166692382862728e-06,
+      "loss": 0.3546,
+      "step": 21092
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 1.314097618680245e-06,
+      "loss": 0.6117,
+      "step": 21093
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 1.311525999074217e-06,
+      "loss": 0.6056,
+      "step": 21094
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 1.308954379468189e-06,
+      "loss": 0.5858,
+      "step": 21095
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 1.3063827598621613e-06,
+      "loss": 0.6566,
+      "step": 21096
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 1.3038111402561333e-06,
+      "loss": 0.4167,
+      "step": 21097
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 1.3012395206501053e-06,
+      "loss": 0.7453,
+      "step": 21098
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 1.2986679010440776e-06,
+      "loss": 0.5122,
+      "step": 21099
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 1.2960962814380498e-06,
+      "loss": 0.7856,
+      "step": 21100
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 1.293524661832022e-06,
+      "loss": 0.5876,
+      "step": 21101
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 1.290953042225994e-06,
+      "loss": 0.7597,
+      "step": 21102
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 1.288381422619966e-06,
+      "loss": 0.5215,
+      "step": 21103
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 1.2858098030139383e-06,
+      "loss": 0.4853,
+      "step": 21104
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 1.2832381834079103e-06,
+      "loss": 0.5929,
+      "step": 21105
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 1.2806665638018825e-06,
+      "loss": 0.703,
+      "step": 21106
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 1.2780949441958547e-06,
+      "loss": 0.6357,
+      "step": 21107
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 1.2755233245898268e-06,
+      "loss": 0.7134,
+      "step": 21108
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 1.2729517049837988e-06,
+      "loss": 0.6773,
+      "step": 21109
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 1.270380085377771e-06,
+      "loss": 1.0048,
+      "step": 21110
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 1.267808465771743e-06,
+      "loss": 0.5472,
+      "step": 21111
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 1.2652368461657152e-06,
+      "loss": 0.6679,
+      "step": 21112
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 1.2626652265596875e-06,
+      "loss": 0.4073,
+      "step": 21113
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 1.2600936069536595e-06,
+      "loss": 0.3859,
+      "step": 21114
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 1.2575219873476317e-06,
+      "loss": 0.5038,
+      "step": 21115
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 1.2549503677416037e-06,
+      "loss": 0.7575,
+      "step": 21116
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 1.2523787481355757e-06,
+      "loss": 0.6366,
+      "step": 21117
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 1.249807128529548e-06,
+      "loss": 0.5343,
+      "step": 21118
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 1.24723550892352e-06,
+      "loss": 0.8091,
+      "step": 21119
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 1.2446638893174922e-06,
+      "loss": 0.6348,
+      "step": 21120
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 1.2420922697114644e-06,
+      "loss": 0.8287,
+      "step": 21121
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 1.2395206501054364e-06,
+      "loss": 0.6638,
+      "step": 21122
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 1.2369490304994087e-06,
+      "loss": 0.6156,
+      "step": 21123
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 1.2343774108933807e-06,
+      "loss": 0.4885,
+      "step": 21124
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 1.2318057912873527e-06,
+      "loss": 0.8851,
+      "step": 21125
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 1.229234171681325e-06,
+      "loss": 0.651,
+      "step": 21126
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 1.2266625520752971e-06,
+      "loss": 0.5185,
+      "step": 21127
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 1.2240909324692692e-06,
+      "loss": 0.7269,
+      "step": 21128
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 1.2215193128632414e-06,
+      "loss": 0.4239,
+      "step": 21129
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 1.2189476932572134e-06,
+      "loss": 0.7057,
+      "step": 21130
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 1.2163760736511856e-06,
+      "loss": 0.6653,
+      "step": 21131
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 1.2138044540451576e-06,
+      "loss": 0.5831,
+      "step": 21132
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 1.2112328344391299e-06,
+      "loss": 0.6508,
+      "step": 21133
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 1.208661214833102e-06,
+      "loss": 0.8049,
+      "step": 21134
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 1.2060895952270741e-06,
+      "loss": 0.459,
+      "step": 21135
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 1.2035179756210461e-06,
+      "loss": 0.2975,
+      "step": 21136
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 1.2009463560150184e-06,
+      "loss": 0.3789,
+      "step": 21137
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 1.1983747364089904e-06,
+      "loss": 0.6989,
+      "step": 21138
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 1.1958031168029626e-06,
+      "loss": 0.5721,
+      "step": 21139
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 1.1932314971969346e-06,
+      "loss": 0.4743,
+      "step": 21140
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 1.1906598775909068e-06,
+      "loss": 0.6432,
+      "step": 21141
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 1.188088257984879e-06,
+      "loss": 0.6514,
+      "step": 21142
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 1.185516638378851e-06,
+      "loss": 0.6684,
+      "step": 21143
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 1.182945018772823e-06,
+      "loss": 0.6582,
+      "step": 21144
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 1.1803733991667953e-06,
+      "loss": 0.5723,
+      "step": 21145
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 1.1778017795607673e-06,
+      "loss": 0.2368,
+      "step": 21146
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 1.1752301599547396e-06,
+      "loss": 0.6779,
+      "step": 21147
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 1.1726585403487118e-06,
+      "loss": 0.6187,
+      "step": 21148
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 1.1700869207426838e-06,
+      "loss": 0.5534,
+      "step": 21149
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 1.167515301136656e-06,
+      "loss": 0.486,
+      "step": 21150
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 1.164943681530628e-06,
+      "loss": 0.6404,
+      "step": 21151
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 1.1623720619246e-06,
+      "loss": 0.4532,
+      "step": 21152
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 1.1598004423185723e-06,
+      "loss": 0.6214,
+      "step": 21153
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 1.1572288227125445e-06,
+      "loss": 0.9826,
+      "step": 21154
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 1.1546572031065165e-06,
+      "loss": 0.5297,
+      "step": 21155
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 1.1520855835004887e-06,
+      "loss": 0.8512,
+      "step": 21156
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 1.1495139638944608e-06,
+      "loss": 0.6221,
+      "step": 21157
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 1.146942344288433e-06,
+      "loss": 0.4984,
+      "step": 21158
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 1.144370724682405e-06,
+      "loss": 0.7791,
+      "step": 21159
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 1.141799105076377e-06,
+      "loss": 0.9034,
+      "step": 21160
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 1.1392274854703492e-06,
+      "loss": 0.904,
+      "step": 21161
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 1.1366558658643215e-06,
+      "loss": 0.4733,
+      "step": 21162
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 1.1340842462582935e-06,
+      "loss": 0.543,
+      "step": 21163
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 1.1315126266522657e-06,
+      "loss": 0.3434,
+      "step": 21164
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 1.1289410070462377e-06,
+      "loss": 0.4307,
+      "step": 21165
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 1.12636938744021e-06,
+      "loss": 0.5868,
+      "step": 21166
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 1.123797767834182e-06,
+      "loss": 0.6442,
+      "step": 21167
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 1.1212261482281542e-06,
+      "loss": 0.3879,
+      "step": 21168
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 1.1186545286221264e-06,
+      "loss": 0.5013,
+      "step": 21169
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 1.1160829090160984e-06,
+      "loss": 0.1472,
+      "step": 21170
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 1.1135112894100704e-06,
+      "loss": 0.5066,
+      "step": 21171
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 1.1109396698040427e-06,
+      "loss": 0.4796,
+      "step": 21172
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 1.1083680501980147e-06,
+      "loss": 0.2501,
+      "step": 21173
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 1.1057964305919867e-06,
+      "loss": 0.6116,
+      "step": 21174
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 1.1032248109859591e-06,
+      "loss": 0.4047,
+      "step": 21175
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 1.1006531913799312e-06,
+      "loss": 0.8551,
+      "step": 21176
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 1.0980815717739034e-06,
+      "loss": 0.6592,
+      "step": 21177
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 1.0955099521678754e-06,
+      "loss": 0.6968,
+      "step": 21178
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 1.0929383325618474e-06,
+      "loss": 0.7515,
+      "step": 21179
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 1.0903667129558196e-06,
+      "loss": 0.5182,
+      "step": 21180
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 1.0877950933497917e-06,
+      "loss": 0.6737,
+      "step": 21181
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 1.0852234737437639e-06,
+      "loss": 0.7628,
+      "step": 21182
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 1.082651854137736e-06,
+      "loss": 0.5958,
+      "step": 21183
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 1.0800802345317081e-06,
+      "loss": 0.5473,
+      "step": 21184
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 1.0775086149256803e-06,
+      "loss": 0.6059,
+      "step": 21185
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 1.0749369953196524e-06,
+      "loss": 0.5013,
+      "step": 21186
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 1.0723653757136244e-06,
+      "loss": 0.5326,
+      "step": 21187
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 1.0697937561075966e-06,
+      "loss": 0.7486,
+      "step": 21188
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 1.0672221365015688e-06,
+      "loss": 0.4572,
+      "step": 21189
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 1.0646505168955408e-06,
+      "loss": 0.5465,
+      "step": 21190
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 1.062078897289513e-06,
+      "loss": 0.6578,
+      "step": 21191
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 1.059507277683485e-06,
+      "loss": 0.3896,
+      "step": 21192
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 1.056935658077457e-06,
+      "loss": 0.5665,
+      "step": 21193
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 1.0543640384714293e-06,
+      "loss": 0.4543,
+      "step": 21194
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 1.0517924188654016e-06,
+      "loss": 0.5841,
+      "step": 21195
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 1.0492207992593738e-06,
+      "loss": 0.617,
+      "step": 21196
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 1.0466491796533458e-06,
+      "loss": 0.8686,
+      "step": 21197
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 1.0440775600473178e-06,
+      "loss": 0.6835,
+      "step": 21198
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 1.04150594044129e-06,
+      "loss": 0.452,
+      "step": 21199
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 1.038934320835262e-06,
+      "loss": 0.8606,
+      "step": 21200
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 1.036362701229234e-06,
+      "loss": 0.4826,
+      "step": 21201
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 1.0337910816232063e-06,
+      "loss": 0.8278,
+      "step": 21202
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 1.0312194620171785e-06,
+      "loss": 0.9481,
+      "step": 21203
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 1.0286478424111507e-06,
+      "loss": 0.4498,
+      "step": 21204
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 1.0260762228051228e-06,
+      "loss": 0.3689,
+      "step": 21205
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 1.0235046031990948e-06,
+      "loss": 0.5369,
+      "step": 21206
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 1.020932983593067e-06,
+      "loss": 0.6348,
+      "step": 21207
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 1.018361363987039e-06,
+      "loss": 0.7045,
+      "step": 21208
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 1.0157897443810112e-06,
+      "loss": 0.7217,
+      "step": 21209
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 1.0132181247749835e-06,
+      "loss": 0.542,
+      "step": 21210
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 1.0106465051689555e-06,
+      "loss": 0.5715,
+      "step": 21211
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 1.0080748855629275e-06,
+      "loss": 0.4232,
+      "step": 21212
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 1.0055032659568997e-06,
+      "loss": 0.5017,
+      "step": 21213
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 1.0029316463508717e-06,
+      "loss": 0.6599,
+      "step": 21214
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 1.000360026744844e-06,
+      "loss": 0.5594,
+      "step": 21215
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 9.977884071388162e-07,
+      "loss": 0.4216,
+      "step": 21216
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 9.952167875327882e-07,
+      "loss": 0.6443,
+      "step": 21217
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 9.926451679267604e-07,
+      "loss": 0.4821,
+      "step": 21218
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 9.900735483207324e-07,
+      "loss": 0.9138,
+      "step": 21219
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 9.875019287147045e-07,
+      "loss": 0.5039,
+      "step": 21220
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 9.849303091086767e-07,
+      "loss": 0.819,
+      "step": 21221
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 9.823586895026487e-07,
+      "loss": 0.4416,
+      "step": 21222
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 9.79787069896621e-07,
+      "loss": 0.5011,
+      "step": 21223
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 9.772154502905932e-07,
+      "loss": 0.869,
+      "step": 21224
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 9.746438306845652e-07,
+      "loss": 0.4885,
+      "step": 21225
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 9.720722110785374e-07,
+      "loss": 0.324,
+      "step": 21226
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 9.695005914725094e-07,
+      "loss": 0.3248,
+      "step": 21227
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 9.669289718664814e-07,
+      "loss": 0.5764,
+      "step": 21228
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 9.643573522604536e-07,
+      "loss": 0.5141,
+      "step": 21229
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 9.617857326544259e-07,
+      "loss": 0.7352,
+      "step": 21230
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 9.592141130483979e-07,
+      "loss": 0.4797,
+      "step": 21231
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 9.566424934423701e-07,
+      "loss": 0.9419,
+      "step": 21232
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 9.540708738363421e-07,
+      "loss": 0.2882,
+      "step": 21233
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 9.514992542303144e-07,
+      "loss": 0.7664,
+      "step": 21234
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 9.489276346242865e-07,
+      "loss": 0.6403,
+      "step": 21235
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 9.463560150182585e-07,
+      "loss": 0.7861,
+      "step": 21236
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 9.437843954122307e-07,
+      "loss": 0.633,
+      "step": 21237
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 9.412127758062027e-07,
+      "loss": 0.462,
+      "step": 21238
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 9.386411562001749e-07,
+      "loss": 0.6936,
+      "step": 21239
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 9.360695365941471e-07,
+      "loss": 0.4677,
+      "step": 21240
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 9.334979169881191e-07,
+      "loss": 0.5492,
+      "step": 21241
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 9.309262973820913e-07,
+      "loss": 0.7708,
+      "step": 21242
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 9.283546777760634e-07,
+      "loss": 0.5449,
+      "step": 21243
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 9.257830581700355e-07,
+      "loss": 0.4901,
+      "step": 21244
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 9.232114385640077e-07,
+      "loss": 0.4041,
+      "step": 21245
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 9.206398189579798e-07,
+      "loss": 0.7829,
+      "step": 21246
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 9.180681993519518e-07,
+      "loss": 0.7697,
+      "step": 21247
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 9.15496579745924e-07,
+      "loss": 0.4882,
+      "step": 21248
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 9.129249601398962e-07,
+      "loss": 0.7336,
+      "step": 21249
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 9.103533405338682e-07,
+      "loss": 0.9065,
+      "step": 21250
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 9.077817209278404e-07,
+      "loss": 0.456,
+      "step": 21251
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 9.052101013218125e-07,
+      "loss": 0.5446,
+      "step": 21252
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 9.026384817157848e-07,
+      "loss": 0.9979,
+      "step": 21253
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 9.000668621097568e-07,
+      "loss": 0.3132,
+      "step": 21254
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 8.974952425037289e-07,
+      "loss": 0.8062,
+      "step": 21255
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 8.949236228977011e-07,
+      "loss": 0.5679,
+      "step": 21256
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 8.923520032916731e-07,
+      "loss": 0.5004,
+      "step": 21257
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 8.897803836856451e-07,
+      "loss": 0.3723,
+      "step": 21258
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 8.872087640796174e-07,
+      "loss": 0.3667,
+      "step": 21259
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 8.846371444735895e-07,
+      "loss": 0.3549,
+      "step": 21260
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 8.820655248675617e-07,
+      "loss": 0.477,
+      "step": 21261
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 8.794939052615337e-07,
+      "loss": 0.9084,
+      "step": 21262
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 8.769222856555059e-07,
+      "loss": 0.5367,
+      "step": 21263
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 8.743506660494781e-07,
+      "loss": 0.368,
+      "step": 21264
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 8.717790464434501e-07,
+      "loss": 0.4666,
+      "step": 21265
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 8.692074268374222e-07,
+      "loss": 0.349,
+      "step": 21266
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 8.666358072313944e-07,
+      "loss": 0.6555,
+      "step": 21267
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 8.640641876253665e-07,
+      "loss": 0.6876,
+      "step": 21268
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 8.614925680193386e-07,
+      "loss": 0.4813,
+      "step": 21269
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 8.589209484133108e-07,
+      "loss": 0.5873,
+      "step": 21270
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 8.563493288072828e-07,
+      "loss": 0.5589,
+      "step": 21271
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 8.53777709201255e-07,
+      "loss": 0.4421,
+      "step": 21272
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 8.512060895952272e-07,
+      "loss": 0.5696,
+      "step": 21273
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 8.486344699891992e-07,
+      "loss": 0.5404,
+      "step": 21274
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 8.460628503831714e-07,
+      "loss": 0.5526,
+      "step": 21275
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 8.434912307771435e-07,
+      "loss": 0.6034,
+      "step": 21276
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 8.409196111711155e-07,
+      "loss": 0.6418,
+      "step": 21277
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 8.383479915650878e-07,
+      "loss": 0.3794,
+      "step": 21278
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 8.357763719590598e-07,
+      "loss": 0.7257,
+      "step": 21279
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 8.332047523530321e-07,
+      "loss": 0.3343,
+      "step": 21280
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 8.306331327470041e-07,
+      "loss": 0.8591,
+      "step": 21281
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 8.280615131409761e-07,
+      "loss": 0.6837,
+      "step": 21282
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 8.254898935349484e-07,
+      "loss": 0.4065,
+      "step": 21283
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 8.229182739289205e-07,
+      "loss": 0.8024,
+      "step": 21284
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 8.203466543228925e-07,
+      "loss": 0.6759,
+      "step": 21285
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 8.177750347168647e-07,
+      "loss": 0.6272,
+      "step": 21286
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 8.152034151108368e-07,
+      "loss": 0.5376,
+      "step": 21287
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 8.126317955048089e-07,
+      "loss": 0.7823,
+      "step": 21288
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 8.100601758987811e-07,
+      "loss": 0.6767,
+      "step": 21289
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 8.074885562927532e-07,
+      "loss": 0.4461,
+      "step": 21290
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 8.049169366867254e-07,
+      "loss": 0.55,
+      "step": 21291
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 8.023453170806975e-07,
+      "loss": 0.7656,
+      "step": 21292
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 7.997736974746696e-07,
+      "loss": 0.6533,
+      "step": 21293
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 7.972020778686418e-07,
+      "loss": 0.8066,
+      "step": 21294
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 7.946304582626138e-07,
+      "loss": 0.6503,
+      "step": 21295
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 7.920588386565858e-07,
+      "loss": 0.6639,
+      "step": 21296
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 7.894872190505582e-07,
+      "loss": 0.7116,
+      "step": 21297
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 7.869155994445302e-07,
+      "loss": 0.4484,
+      "step": 21298
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 7.843439798385024e-07,
+      "loss": 0.555,
+      "step": 21299
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 7.817723602324744e-07,
+      "loss": 0.4105,
+      "step": 21300
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 7.792007406264466e-07,
+      "loss": 0.685,
+      "step": 21301
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 7.766291210204187e-07,
+      "loss": 0.7258,
+      "step": 21302
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 7.740575014143908e-07,
+      "loss": 0.637,
+      "step": 21303
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 7.71485881808363e-07,
+      "loss": 0.3359,
+      "step": 21304
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 7.689142622023351e-07,
+      "loss": 0.386,
+      "step": 21305
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 7.663426425963071e-07,
+      "loss": 0.6052,
+      "step": 21306
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 7.637710229902794e-07,
+      "loss": 0.4463,
+      "step": 21307
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 7.611994033842515e-07,
+      "loss": 0.6743,
+      "step": 21308
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 7.586277837782235e-07,
+      "loss": 0.4279,
+      "step": 21309
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 7.560561641721956e-07,
+      "loss": 0.3929,
+      "step": 21310
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 7.534845445661678e-07,
+      "loss": 0.9861,
+      "step": 21311
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 7.5091292496014e-07,
+      "loss": 0.4612,
+      "step": 21312
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 7.48341305354112e-07,
+      "loss": 0.5776,
+      "step": 21313
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 7.457696857480842e-07,
+      "loss": 0.5209,
+      "step": 21314
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 7.431980661420563e-07,
+      "loss": 0.4434,
+      "step": 21315
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 7.406264465360284e-07,
+      "loss": 0.5573,
+      "step": 21316
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 7.380548269300006e-07,
+      "loss": 0.4597,
+      "step": 21317
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 7.354832073239727e-07,
+      "loss": 0.5869,
+      "step": 21318
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 7.329115877179448e-07,
+      "loss": 0.6693,
+      "step": 21319
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 7.303399681119169e-07,
+      "loss": 0.4254,
+      "step": 21320
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 7.27768348505889e-07,
+      "loss": 0.2942,
+      "step": 21321
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 7.251967288998612e-07,
+      "loss": 0.4709,
+      "step": 21322
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 7.226251092938333e-07,
+      "loss": 0.2348,
+      "step": 21323
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 7.200534896878054e-07,
+      "loss": 0.4611,
+      "step": 21324
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 7.174818700817775e-07,
+      "loss": 0.5999,
+      "step": 21325
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 7.149102504757497e-07,
+      "loss": 0.2808,
+      "step": 21326
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 7.123386308697218e-07,
+      "loss": 0.4195,
+      "step": 21327
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 7.097670112636939e-07,
+      "loss": 0.5504,
+      "step": 21328
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 7.07195391657666e-07,
+      "loss": 0.2929,
+      "step": 21329
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 7.046237720516381e-07,
+      "loss": 0.4517,
+      "step": 21330
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 7.020521524456103e-07,
+      "loss": 0.4358,
+      "step": 21331
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 6.994805328395824e-07,
+      "loss": 0.6542,
+      "step": 21332
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 6.969089132335545e-07,
+      "loss": 0.402,
+      "step": 21333
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 6.943372936275266e-07,
+      "loss": 0.6395,
+      "step": 21334
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 6.917656740214988e-07,
+      "loss": 0.8046,
+      "step": 21335
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 6.891940544154709e-07,
+      "loss": 0.748,
+      "step": 21336
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 6.86622434809443e-07,
+      "loss": 0.5657,
+      "step": 21337
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 6.840508152034152e-07,
+      "loss": 0.557,
+      "step": 21338
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 6.814791955973873e-07,
+      "loss": 0.5681,
+      "step": 21339
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 6.789075759913593e-07,
+      "loss": 0.7166,
+      "step": 21340
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 6.763359563853315e-07,
+      "loss": 0.5086,
+      "step": 21341
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 6.737643367793037e-07,
+      "loss": 0.6998,
+      "step": 21342
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 6.711927171732758e-07,
+      "loss": 0.4122,
+      "step": 21343
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 6.686210975672478e-07,
+      "loss": 0.4201,
+      "step": 21344
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 6.6604947796122e-07,
+      "loss": 0.5971,
+      "step": 21345
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 6.634778583551922e-07,
+      "loss": 0.7194,
+      "step": 21346
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 6.609062387491642e-07,
+      "loss": 0.6588,
+      "step": 21347
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 6.583346191431364e-07,
+      "loss": 0.5642,
+      "step": 21348
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 6.557629995371085e-07,
+      "loss": 0.2656,
+      "step": 21349
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 6.531913799310807e-07,
+      "loss": 0.7566,
+      "step": 21350
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 6.506197603250527e-07,
+      "loss": 0.3832,
+      "step": 21351
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 6.480481407190249e-07,
+      "loss": 0.6297,
+      "step": 21352
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 6.45476521112997e-07,
+      "loss": 0.4353,
+      "step": 21353
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 6.429049015069691e-07,
+      "loss": 0.5029,
+      "step": 21354
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 6.403332819009413e-07,
+      "loss": 0.4096,
+      "step": 21355
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 6.377616622949134e-07,
+      "loss": 0.3367,
+      "step": 21356
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 6.351900426888855e-07,
+      "loss": 0.6128,
+      "step": 21357
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 6.326184230828576e-07,
+      "loss": 0.3487,
+      "step": 21358
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 6.300468034768297e-07,
+      "loss": 0.3465,
+      "step": 21359
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 6.274751838708019e-07,
+      "loss": 0.7752,
+      "step": 21360
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 6.24903564264774e-07,
+      "loss": 0.6173,
+      "step": 21361
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 6.223319446587461e-07,
+      "loss": 0.9498,
+      "step": 21362
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 6.197603250527182e-07,
+      "loss": 0.5118,
+      "step": 21363
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 6.171887054466903e-07,
+      "loss": 0.6312,
+      "step": 21364
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 6.146170858406625e-07,
+      "loss": 0.7218,
+      "step": 21365
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 6.120454662346346e-07,
+      "loss": 0.7698,
+      "step": 21366
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 6.094738466286067e-07,
+      "loss": 0.4168,
+      "step": 21367
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 6.069022270225788e-07,
+      "loss": 0.7584,
+      "step": 21368
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 6.04330607416551e-07,
+      "loss": 0.8446,
+      "step": 21369
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 6.017589878105231e-07,
+      "loss": 0.5757,
+      "step": 21370
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 5.991873682044952e-07,
+      "loss": 0.696,
+      "step": 21371
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 5.966157485984673e-07,
+      "loss": 0.5896,
+      "step": 21372
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 5.940441289924395e-07,
+      "loss": 0.4058,
+      "step": 21373
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 5.914725093864115e-07,
+      "loss": 0.3963,
+      "step": 21374
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 5.889008897803837e-07,
+      "loss": 0.7165,
+      "step": 21375
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 5.863292701743559e-07,
+      "loss": 0.2501,
+      "step": 21376
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 5.83757650568328e-07,
+      "loss": 0.6889,
+      "step": 21377
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 5.811860309623e-07,
+      "loss": 0.6551,
+      "step": 21378
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 5.786144113562723e-07,
+      "loss": 0.4721,
+      "step": 21379
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 5.760427917502444e-07,
+      "loss": 0.7122,
+      "step": 21380
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 5.734711721442165e-07,
+      "loss": 0.6196,
+      "step": 21381
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 5.708995525381885e-07,
+      "loss": 0.3756,
+      "step": 21382
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 5.683279329321607e-07,
+      "loss": 0.4151,
+      "step": 21383
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 5.657563133261329e-07,
+      "loss": 0.7395,
+      "step": 21384
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 5.63184693720105e-07,
+      "loss": 0.7391,
+      "step": 21385
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 5.606130741140771e-07,
+      "loss": 0.3118,
+      "step": 21386
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 5.580414545080492e-07,
+      "loss": 0.8243,
+      "step": 21387
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 5.554698349020213e-07,
+      "loss": 0.5498,
+      "step": 21388
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 5.528982152959934e-07,
+      "loss": 0.4816,
+      "step": 21389
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 5.503265956899656e-07,
+      "loss": 0.4289,
+      "step": 21390
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 5.477549760839377e-07,
+      "loss": 0.6205,
+      "step": 21391
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 5.451833564779098e-07,
+      "loss": 0.743,
+      "step": 21392
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 5.426117368718819e-07,
+      "loss": 0.9798,
+      "step": 21393
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 5.400401172658541e-07,
+      "loss": 0.6427,
+      "step": 21394
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 5.374684976598262e-07,
+      "loss": 0.7828,
+      "step": 21395
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 5.348968780537983e-07,
+      "loss": 0.6175,
+      "step": 21396
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 5.323252584477704e-07,
+      "loss": 0.3248,
+      "step": 21397
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 5.297536388417425e-07,
+      "loss": 0.4736,
+      "step": 21398
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 5.271820192357147e-07,
+      "loss": 0.4327,
+      "step": 21399
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 5.246103996296869e-07,
+      "loss": 0.5648,
+      "step": 21400
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 5.220387800236589e-07,
+      "loss": 0.5641,
+      "step": 21401
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 5.19467160417631e-07,
+      "loss": 0.4835,
+      "step": 21402
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 5.168955408116031e-07,
+      "loss": 0.5108,
+      "step": 21403
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 5.143239212055754e-07,
+      "loss": 0.6879,
+      "step": 21404
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 5.117523015995474e-07,
+      "loss": 0.7795,
+      "step": 21405
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 5.091806819935195e-07,
+      "loss": 0.6675,
+      "step": 21406
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 5.066090623874917e-07,
+      "loss": 0.6725,
+      "step": 21407
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 5.040374427814637e-07,
+      "loss": 0.7407,
+      "step": 21408
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 5.014658231754359e-07,
+      "loss": 0.6175,
+      "step": 21409
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 4.988942035694081e-07,
+      "loss": 0.7416,
+      "step": 21410
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 4.963225839633802e-07,
+      "loss": 0.7252,
+      "step": 21411
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 4.937509643573522e-07,
+      "loss": 0.6352,
+      "step": 21412
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 4.911793447513243e-07,
+      "loss": 0.5302,
+      "step": 21413
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 4.886077251452966e-07,
+      "loss": 0.436,
+      "step": 21414
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 4.860361055392687e-07,
+      "loss": 0.7367,
+      "step": 21415
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 4.834644859332407e-07,
+      "loss": 0.6174,
+      "step": 21416
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 4.808928663272129e-07,
+      "loss": 0.4065,
+      "step": 21417
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 4.783212467211851e-07,
+      "loss": 0.4319,
+      "step": 21418
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 4.757496271151572e-07,
+      "loss": 0.6896,
+      "step": 21419
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 4.7317800750912925e-07,
+      "loss": 0.6387,
+      "step": 21420
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 4.7060638790310137e-07,
+      "loss": 0.4598,
+      "step": 21421
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 4.6803476829707354e-07,
+      "loss": 0.4942,
+      "step": 21422
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 4.6546314869104566e-07,
+      "loss": 0.5833,
+      "step": 21423
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 4.6289152908501773e-07,
+      "loss": 0.4274,
+      "step": 21424
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 4.603199094789899e-07,
+      "loss": 0.4849,
+      "step": 21425
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 4.57748289872962e-07,
+      "loss": 0.4597,
+      "step": 21426
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 4.551766702669341e-07,
+      "loss": 0.4261,
+      "step": 21427
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 4.5260505066090626e-07,
+      "loss": 0.5798,
+      "step": 21428
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 4.500334310548784e-07,
+      "loss": 0.6175,
+      "step": 21429
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 4.4746181144885056e-07,
+      "loss": 0.635,
+      "step": 21430
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 4.4489019184282257e-07,
+      "loss": 0.4876,
+      "step": 21431
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 4.4231857223679474e-07,
+      "loss": 0.5271,
+      "step": 21432
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 4.3974695263076687e-07,
+      "loss": 0.6946,
+      "step": 21433
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 4.3717533302473904e-07,
+      "loss": 0.4546,
+      "step": 21434
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 4.346037134187111e-07,
+      "loss": 0.8643,
+      "step": 21435
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 4.3203209381268323e-07,
+      "loss": 0.4988,
+      "step": 21436
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 4.294604742066554e-07,
+      "loss": 0.4408,
+      "step": 21437
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 4.268888546006275e-07,
+      "loss": 0.5889,
+      "step": 21438
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 4.243172349945996e-07,
+      "loss": 0.4304,
+      "step": 21439
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 4.2174561538857176e-07,
+      "loss": 0.5223,
+      "step": 21440
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 4.191739957825439e-07,
+      "loss": 0.7286,
+      "step": 21441
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 4.1660237617651606e-07,
+      "loss": 0.5561,
+      "step": 21442
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 4.1403075657048807e-07,
+      "loss": 0.4916,
+      "step": 21443
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 4.1145913696446024e-07,
+      "loss": 0.6601,
+      "step": 21444
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 4.0888751735843236e-07,
+      "loss": 0.4858,
+      "step": 21445
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 4.0631589775240443e-07,
+      "loss": 0.5773,
+      "step": 21446
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 4.037442781463766e-07,
+      "loss": 0.4348,
+      "step": 21447
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 4.011726585403487e-07,
+      "loss": 0.6455,
+      "step": 21448
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 3.986010389343209e-07,
+      "loss": 0.5857,
+      "step": 21449
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 3.960294193282929e-07,
+      "loss": 0.6139,
+      "step": 21450
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 3.934577997222651e-07,
+      "loss": 0.4795,
+      "step": 21451
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 3.908861801162372e-07,
+      "loss": 0.6517,
+      "step": 21452
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 3.8831456051020933e-07,
+      "loss": 0.5687,
+      "step": 21453
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 3.857429409041815e-07,
+      "loss": 0.6169,
+      "step": 21454
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 3.8317132129815357e-07,
+      "loss": 0.472,
+      "step": 21455
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 3.8059970169212574e-07,
+      "loss": 0.8172,
+      "step": 21456
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 3.780280820860978e-07,
+      "loss": 0.5637,
+      "step": 21457
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 3.7545646248007e-07,
+      "loss": 0.4782,
+      "step": 21458
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 3.728848428740421e-07,
+      "loss": 0.684,
+      "step": 21459
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 3.703132232680142e-07,
+      "loss": 0.5439,
+      "step": 21460
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 3.6774160366198635e-07,
+      "loss": 0.6665,
+      "step": 21461
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 3.6516998405595847e-07,
+      "loss": 0.4347,
+      "step": 21462
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 3.625983644499306e-07,
+      "loss": 0.442,
+      "step": 21463
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 3.600267448439027e-07,
+      "loss": 0.5313,
+      "step": 21464
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 3.5745512523787483e-07,
+      "loss": 0.34,
+      "step": 21465
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 3.5488350563184695e-07,
+      "loss": 0.5179,
+      "step": 21466
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 3.5231188602581907e-07,
+      "loss": 0.3814,
+      "step": 21467
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 3.497402664197912e-07,
+      "loss": 0.5214,
+      "step": 21468
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 3.471686468137633e-07,
+      "loss": 0.5397,
+      "step": 21469
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 3.4459702720773543e-07,
+      "loss": 0.4245,
+      "step": 21470
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 3.420254076017076e-07,
+      "loss": 0.3032,
+      "step": 21471
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 3.3945378799567967e-07,
+      "loss": 0.5633,
+      "step": 21472
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 3.3688216838965184e-07,
+      "loss": 0.5839,
+      "step": 21473
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 3.343105487836239e-07,
+      "loss": 0.6703,
+      "step": 21474
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 3.317389291775961e-07,
+      "loss": 0.6425,
+      "step": 21475
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 3.291673095715682e-07,
+      "loss": 0.4564,
+      "step": 21476
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 3.265956899655403e-07,
+      "loss": 0.8035,
+      "step": 21477
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 3.2402407035951245e-07,
+      "loss": 0.3618,
+      "step": 21478
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 3.2145245075348457e-07,
+      "loss": 0.6645,
+      "step": 21479
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 3.188808311474567e-07,
+      "loss": 0.5786,
+      "step": 21480
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 3.163092115414288e-07,
+      "loss": 0.6951,
+      "step": 21481
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 3.1373759193540093e-07,
+      "loss": 0.4157,
+      "step": 21482
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 3.1116597232937305e-07,
+      "loss": 0.5753,
+      "step": 21483
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 3.0859435272334517e-07,
+      "loss": 0.481,
+      "step": 21484
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 3.060227331173173e-07,
+      "loss": 0.2902,
+      "step": 21485
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 3.034511135112894e-07,
+      "loss": 0.846,
+      "step": 21486
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 3.0087949390526153e-07,
+      "loss": 0.8225,
+      "step": 21487
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 2.9830787429923365e-07,
+      "loss": 0.5049,
+      "step": 21488
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 2.9573625469320577e-07,
+      "loss": 0.3043,
+      "step": 21489
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 2.9316463508717795e-07,
+      "loss": 0.492,
+      "step": 21490
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 2.9059301548115e-07,
+      "loss": 0.3332,
+      "step": 21491
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 2.880213958751222e-07,
+      "loss": 1.0263,
+      "step": 21492
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 2.8544977626909425e-07,
+      "loss": 0.7276,
+      "step": 21493
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 2.8287815666306643e-07,
+      "loss": 0.4764,
+      "step": 21494
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 2.8030653705703855e-07,
+      "loss": 0.4647,
+      "step": 21495
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 2.7773491745101067e-07,
+      "loss": 0.6661,
+      "step": 21496
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 2.751632978449828e-07,
+      "loss": 0.6511,
+      "step": 21497
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 2.725916782389549e-07,
+      "loss": 0.4882,
+      "step": 21498
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 2.7002005863292703e-07,
+      "loss": 0.5497,
+      "step": 21499
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 2.6744843902689915e-07,
+      "loss": 0.51,
+      "step": 21500
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 2.6487681942087127e-07,
+      "loss": 0.5617,
+      "step": 21501
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 2.6230519981484344e-07,
+      "loss": 0.6799,
+      "step": 21502
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 2.597335802088155e-07,
+      "loss": 0.4733,
+      "step": 21503
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 2.571619606027877e-07,
+      "loss": 0.4829,
+      "step": 21504
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 2.5459034099675975e-07,
+      "loss": 0.5804,
+      "step": 21505
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 2.520187213907319e-07,
+      "loss": 0.7854,
+      "step": 21506
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 2.4944710178470405e-07,
+      "loss": 0.6529,
+      "step": 21507
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 2.468754821786761e-07,
+      "loss": 0.3278,
+      "step": 21508
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 2.443038625726483e-07,
+      "loss": 0.7383,
+      "step": 21509
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 2.4173224296662036e-07,
+      "loss": 0.4238,
+      "step": 21510
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 2.3916062336059253e-07,
+      "loss": 0.5313,
+      "step": 21511
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 2.3658900375456462e-07,
+      "loss": 0.9371,
+      "step": 21512
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 2.3401738414853677e-07,
+      "loss": 0.4114,
+      "step": 21513
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 2.3144576454250886e-07,
+      "loss": 0.5596,
+      "step": 21514
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 2.28874144936481e-07,
+      "loss": 0.5191,
+      "step": 21515
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 2.2630252533045313e-07,
+      "loss": 0.8131,
+      "step": 21516
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 2.2373090572442528e-07,
+      "loss": 0.7436,
+      "step": 21517
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 2.2115928611839737e-07,
+      "loss": 0.5759,
+      "step": 21518
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 2.1858766651236952e-07,
+      "loss": 0.325,
+      "step": 21519
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 2.1601604690634161e-07,
+      "loss": 0.6338,
+      "step": 21520
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 2.1344442730031376e-07,
+      "loss": 0.3681,
+      "step": 21521
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 2.1087280769428588e-07,
+      "loss": 0.6867,
+      "step": 21522
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 2.0830118808825803e-07,
+      "loss": 0.6219,
+      "step": 21523
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 2.0572956848223012e-07,
+      "loss": 0.5245,
+      "step": 21524
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 2.0315794887620222e-07,
+      "loss": 0.6507,
+      "step": 21525
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 2.0058632927017436e-07,
+      "loss": 0.6749,
+      "step": 21526
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 1.9801470966414646e-07,
+      "loss": 0.899,
+      "step": 21527
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 1.954430900581186e-07,
+      "loss": 0.3551,
+      "step": 21528
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 1.9287147045209075e-07,
+      "loss": 0.4399,
+      "step": 21529
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 1.9029985084606287e-07,
+      "loss": 0.5531,
+      "step": 21530
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 1.87728231240035e-07,
+      "loss": 0.6662,
+      "step": 21531
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 1.851566116340071e-07,
+      "loss": 0.4288,
+      "step": 21532
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 1.8258499202797923e-07,
+      "loss": 0.463,
+      "step": 21533
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 1.8001337242195135e-07,
+      "loss": 0.522,
+      "step": 21534
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 1.7744175281592347e-07,
+      "loss": 0.7955,
+      "step": 21535
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 1.748701332098956e-07,
+      "loss": 0.7069,
+      "step": 21536
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 1.7229851360386771e-07,
+      "loss": 0.5081,
+      "step": 21537
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 1.6972689399783984e-07,
+      "loss": 0.5686,
+      "step": 21538
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 1.6715527439181196e-07,
+      "loss": 0.5947,
+      "step": 21539
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 1.645836547857841e-07,
+      "loss": 0.4957,
+      "step": 21540
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 1.6201203517975622e-07,
+      "loss": 0.4963,
+      "step": 21541
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 1.5944041557372834e-07,
+      "loss": 0.6578,
+      "step": 21542
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 1.5686879596770046e-07,
+      "loss": 0.4365,
+      "step": 21543
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 1.5429717636167258e-07,
+      "loss": 0.4538,
+      "step": 21544
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 1.517255567556447e-07,
+      "loss": 0.2623,
+      "step": 21545
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 1.4915393714961683e-07,
+      "loss": 0.443,
+      "step": 21546
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 1.4658231754358897e-07,
+      "loss": 0.7967,
+      "step": 21547
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 1.440106979375611e-07,
+      "loss": 0.7543,
+      "step": 21548
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 1.4143907833153321e-07,
+      "loss": 0.4787,
+      "step": 21549
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 1.3886745872550533e-07,
+      "loss": 0.4228,
+      "step": 21550
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 1.3629583911947745e-07,
+      "loss": 0.7563,
+      "step": 21551
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 1.3372421951344958e-07,
+      "loss": 0.4859,
+      "step": 21552
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 1.3115259990742172e-07,
+      "loss": 0.5304,
+      "step": 21553
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 1.2858098030139384e-07,
+      "loss": 0.8538,
+      "step": 21554
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 1.2600936069536594e-07,
+      "loss": 0.7563,
+      "step": 21555
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 1.2343774108933806e-07,
+      "loss": 0.8381,
+      "step": 21556
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 1.2086612148331018e-07,
+      "loss": 0.5994,
+      "step": 21557
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 1.1829450187728231e-07,
+      "loss": 0.947,
+      "step": 21558
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 1.1572288227125443e-07,
+      "loss": 0.5977,
+      "step": 21559
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 1.1315126266522657e-07,
+      "loss": 0.6505,
+      "step": 21560
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 1.1057964305919869e-07,
+      "loss": 0.5051,
+      "step": 21561
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 1.0800802345317081e-07,
+      "loss": 0.9783,
+      "step": 21562
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 1.0543640384714294e-07,
+      "loss": 0.6227,
+      "step": 21563
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 1.0286478424111506e-07,
+      "loss": 0.7502,
+      "step": 21564
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 1.0029316463508718e-07,
+      "loss": 0.47,
+      "step": 21565
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 9.77215450290593e-08,
+      "loss": 0.5994,
+      "step": 21566
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 9.514992542303144e-08,
+      "loss": 0.4982,
+      "step": 21567
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 9.257830581700356e-08,
+      "loss": 0.567,
+      "step": 21568
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 9.000668621097568e-08,
+      "loss": 0.6278,
+      "step": 21569
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 8.74350666049478e-08,
+      "loss": 0.902,
+      "step": 21570
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 8.486344699891992e-08,
+      "loss": 0.795,
+      "step": 21571
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 8.229182739289205e-08,
+      "loss": 0.423,
+      "step": 21572
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 7.972020778686417e-08,
+      "loss": 0.4641,
+      "step": 21573
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 7.714858818083629e-08,
+      "loss": 0.5763,
+      "step": 21574
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 7.457696857480841e-08,
+      "loss": 0.9515,
+      "step": 21575
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 7.200534896878055e-08,
+      "loss": 0.841,
+      "step": 21576
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 6.943372936275267e-08,
+      "loss": 0.438,
+      "step": 21577
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 6.686210975672479e-08,
+      "loss": 0.5353,
+      "step": 21578
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 6.429049015069692e-08,
+      "loss": 0.4532,
+      "step": 21579
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 6.171887054466903e-08,
+      "loss": 0.5563,
+      "step": 21580
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 5.9147250938641156e-08,
+      "loss": 0.6907,
+      "step": 21581
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 5.657563133261328e-08,
+      "loss": 0.5472,
+      "step": 21582
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 5.4004011726585403e-08,
+      "loss": 0.4991,
+      "step": 21583
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 5.143239212055753e-08,
+      "loss": 0.6009,
+      "step": 21584
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 4.886077251452965e-08,
+      "loss": 0.8325,
+      "step": 21585
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 4.628915290850178e-08,
+      "loss": 0.4582,
+      "step": 21586
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 4.37175333024739e-08,
+      "loss": 0.401,
+      "step": 21587
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 4.1145913696446026e-08,
+      "loss": 0.3575,
+      "step": 21588
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 3.8574294090418146e-08,
+      "loss": 0.4319,
+      "step": 21589
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 3.6002674484390273e-08,
+      "loss": 0.6294,
+      "step": 21590
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 3.3431054878362394e-08,
+      "loss": 0.5124,
+      "step": 21591
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 3.0859435272334514e-08,
+      "loss": 0.6047,
+      "step": 21592
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 2.828781566630664e-08,
+      "loss": 0.7842,
+      "step": 21593
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 2.5716196060278765e-08,
+      "loss": 0.754,
+      "step": 21594
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 2.314457645425089e-08,
+      "loss": 0.5521,
+      "step": 21595
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 2.0572956848223013e-08,
+      "loss": 0.543,
+      "step": 21596
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 1.8001337242195137e-08,
+      "loss": 0.7934,
+      "step": 21597
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 1.5429717636167257e-08,
+      "loss": 0.5515,
+      "step": 21598
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 1.2858098030139383e-08,
+      "loss": 0.9233,
+      "step": 21599
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 1.0286478424111506e-08,
+      "loss": 0.546,
+      "step": 21600
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 7.714858818083629e-09,
+      "loss": 0.7311,
+      "step": 21601
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 5.143239212055753e-09,
+      "loss": 0.5681,
+      "step": 21602
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 2.5716196060278766e-09,
+      "loss": 0.6552,
+      "step": 21603
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 0.0,
+      "loss": 0.4274,
+      "step": 21604
+    },
+    {
+      "epoch": 4.0,
+      "eval_accuracy": 0.7011264818704112,
+      "eval_loss": 0.9610870480537415,
+      "eval_runtime": 40.4854,
+      "eval_samples_per_second": 1423.056,
+      "eval_steps_per_second": 44.485,
+      "step": 21604
+    }
+  ],
+  "logging_steps": 1,
+  "max_steps": 21604,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 4,
+  "save_steps": 500,
+  "total_flos": 1.405051511841864e+16,
+  "train_batch_size": 16,
+  "trial_name": null,
+  "trial_params": null
+}