diff --git "a/checkpoint-3240/trainer_state.json" "b/checkpoint-3240/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/checkpoint-3240/trainer_state.json"
@@ -0,0 +1,19485 @@
+{
+  "best_metric": 1.0341861248016357,
+  "best_model_checkpoint": "./qlora-out/checkpoint-3240",
+  "epoch": 1.0,
+  "eval_steps": 1620,
+  "global_step": 3240,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 1e-05,
+      "loss": 1.0903,
+      "step": 1
+    },
+    {
+      "epoch": 0.0,
+      "eval_loss": 1.7198632955551147,
+      "eval_runtime": 24.3895,
+      "eval_samples_per_second": 111.892,
+      "eval_steps_per_second": 14.022,
+      "step": 1
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2e-05,
+      "loss": 0.9968,
+      "step": 2
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3e-05,
+      "loss": 0.9124,
+      "step": 3
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4e-05,
+      "loss": 1.4168,
+      "step": 4
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5e-05,
+      "loss": 1.1852,
+      "step": 5
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6e-05,
+      "loss": 0.9936,
+      "step": 6
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 7e-05,
+      "loss": 1.188,
+      "step": 7
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8e-05,
+      "loss": 0.9967,
+      "step": 8
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9e-05,
+      "loss": 1.0747,
+      "step": 9
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0001,
+      "loss": 0.8314,
+      "step": 10
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.999998951046439e-05,
+      "loss": 1.1101,
+      "step": 11
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.999995804186196e-05,
+      "loss": 0.7529,
+      "step": 12
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.999990559420591e-05,
+      "loss": 1.0091,
+      "step": 13
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.999983216751826e-05,
+      "loss": 1.075,
+      "step": 14
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.999973776182981e-05,
+      "loss": 1.0245,
+      "step": 15
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.999962237718015e-05,
+      "loss": 1.2007,
+      "step": 16
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.999948601361773e-05,
+      "loss": 0.9599,
+      "step": 17
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.999932867119973e-05,
+      "loss": 1.0614,
+      "step": 18
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.99991503499922e-05,
+      "loss": 1.0841,
+      "step": 19
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.999895105006994e-05,
+      "loss": 1.0794,
+      "step": 20
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.999873077151658e-05,
+      "loss": 1.1938,
+      "step": 21
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.999848951442455e-05,
+      "loss": 1.1148,
+      "step": 22
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.999822727889507e-05,
+      "loss": 0.9307,
+      "step": 23
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.999794406503817e-05,
+      "loss": 0.8877,
+      "step": 24
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.999763987297265e-05,
+      "loss": 0.8661,
+      "step": 25
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.999731470282622e-05,
+      "loss": 1.1864,
+      "step": 26
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.999696855473524e-05,
+      "loss": 0.9525,
+      "step": 27
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.9996601428845e-05,
+      "loss": 0.9294,
+      "step": 28
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.99962133253095e-05,
+      "loss": 0.9421,
+      "step": 29
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.99958042442916e-05,
+      "loss": 0.9744,
+      "step": 30
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.999537418596294e-05,
+      "loss": 1.1042,
+      "step": 31
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.999492315050396e-05,
+      "loss": 0.9822,
+      "step": 32
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.999445113810392e-05,
+      "loss": 0.9979,
+      "step": 33
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.999395814896085e-05,
+      "loss": 0.9346,
+      "step": 34
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.999344418328162e-05,
+      "loss": 0.7803,
+      "step": 35
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.999290924128185e-05,
+      "loss": 1.1201,
+      "step": 36
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.999235332318602e-05,
+      "loss": 1.1814,
+      "step": 37
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.999177642922737e-05,
+      "loss": 1.1256,
+      "step": 38
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.999117855964796e-05,
+      "loss": 1.2464,
+      "step": 39
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.999055971469864e-05,
+      "loss": 0.9625,
+      "step": 40
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.998991989463907e-05,
+      "loss": 0.9457,
+      "step": 41
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.998925909973769e-05,
+      "loss": 1.1738,
+      "step": 42
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.998857733027178e-05,
+      "loss": 1.1494,
+      "step": 43
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.998787458652739e-05,
+      "loss": 1.1065,
+      "step": 44
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.998715086879938e-05,
+      "loss": 0.9335,
+      "step": 45
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.998640617739139e-05,
+      "loss": 0.9792,
+      "step": 46
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.998564051261591e-05,
+      "loss": 0.9249,
+      "step": 47
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.998485387479418e-05,
+      "loss": 1.009,
+      "step": 48
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.998404626425626e-05,
+      "loss": 0.8926,
+      "step": 49
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.998321768134102e-05,
+      "loss": 0.9349,
+      "step": 50
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.99823681263961e-05,
+      "loss": 1.1918,
+      "step": 51
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.998149759977795e-05,
+      "loss": 1.0911,
+      "step": 52
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.998060610185187e-05,
+      "loss": 1.2647,
+      "step": 53
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.997969363299186e-05,
+      "loss": 1.2362,
+      "step": 54
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.997876019358084e-05,
+      "loss": 1.1019,
+      "step": 55
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.997780578401039e-05,
+      "loss": 0.9889,
+      "step": 56
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.997683040468102e-05,
+      "loss": 1.0521,
+      "step": 57
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.997583405600195e-05,
+      "loss": 1.0627,
+      "step": 58
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.997481673839124e-05,
+      "loss": 1.0359,
+      "step": 59
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.997377845227576e-05,
+      "loss": 1.133,
+      "step": 60
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.997271919809111e-05,
+      "loss": 1.1642,
+      "step": 61
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.997163897628175e-05,
+      "loss": 1.003,
+      "step": 62
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.997053778730095e-05,
+      "loss": 1.2627,
+      "step": 63
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.99694156316107e-05,
+      "loss": 1.1546,
+      "step": 64
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.99682725096819e-05,
+      "loss": 1.118,
+      "step": 65
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.996710842199412e-05,
+      "loss": 1.0603,
+      "step": 66
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.996592336903581e-05,
+      "loss": 0.8922,
+      "step": 67
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.996471735130422e-05,
+      "loss": 1.0922,
+      "step": 68
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.996349036930532e-05,
+      "loss": 1.1192,
+      "step": 69
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.996224242355399e-05,
+      "loss": 1.0947,
+      "step": 70
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.99609735145738e-05,
+      "loss": 1.1523,
+      "step": 71
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.995968364289718e-05,
+      "loss": 1.2092,
+      "step": 72
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.995837280906535e-05,
+      "loss": 1.4025,
+      "step": 73
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.995704101362828e-05,
+      "loss": 1.0057,
+      "step": 74
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.995568825714479e-05,
+      "loss": 1.4157,
+      "step": 75
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.995431454018245e-05,
+      "loss": 1.0842,
+      "step": 76
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.995291986331767e-05,
+      "loss": 1.0096,
+      "step": 77
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.995150422713562e-05,
+      "loss": 1.0146,
+      "step": 78
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.995006763223028e-05,
+      "loss": 0.961,
+      "step": 79
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.99486100792044e-05,
+      "loss": 1.2261,
+      "step": 80
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.994713156866954e-05,
+      "loss": 0.8999,
+      "step": 81
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.994563210124611e-05,
+      "loss": 0.8811,
+      "step": 82
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.994411167756318e-05,
+      "loss": 1.0196,
+      "step": 83
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.994257029825876e-05,
+      "loss": 1.0266,
+      "step": 84
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.994100796397954e-05,
+      "loss": 1.2607,
+      "step": 85
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.993942467538106e-05,
+      "loss": 1.1348,
+      "step": 86
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.993782043312765e-05,
+      "loss": 1.0473,
+      "step": 87
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.993619523789241e-05,
+      "loss": 1.0596,
+      "step": 88
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.993454909035723e-05,
+      "loss": 1.3649,
+      "step": 89
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.993288199121283e-05,
+      "loss": 1.0814,
+      "step": 90
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.993119394115867e-05,
+      "loss": 1.0992,
+      "step": 91
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.992948494090302e-05,
+      "loss": 0.97,
+      "step": 92
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.992775499116298e-05,
+      "loss": 0.8215,
+      "step": 93
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.992600409266438e-05,
+      "loss": 0.9276,
+      "step": 94
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.992423224614185e-05,
+      "loss": 1.268,
+      "step": 95
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.992243945233885e-05,
+      "loss": 1.2018,
+      "step": 96
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.99206257120076e-05,
+      "loss": 0.923,
+      "step": 97
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.991879102590911e-05,
+      "loss": 1.1708,
+      "step": 98
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.991693539481316e-05,
+      "loss": 1.0734,
+      "step": 99
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.991505881949837e-05,
+      "loss": 1.0928,
+      "step": 100
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.991316130075208e-05,
+      "loss": 0.9874,
+      "step": 101
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.991124283937049e-05,
+      "loss": 0.9709,
+      "step": 102
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.990930343615853e-05,
+      "loss": 1.1645,
+      "step": 103
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.990734309192994e-05,
+      "loss": 1.4461,
+      "step": 104
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.990536180750725e-05,
+      "loss": 1.0004,
+      "step": 105
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.990335958372176e-05,
+      "loss": 1.0044,
+      "step": 106
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.990133642141359e-05,
+      "loss": 1.06,
+      "step": 107
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.989929232143159e-05,
+      "loss": 1.2315,
+      "step": 108
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.989722728463344e-05,
+      "loss": 0.9924,
+      "step": 109
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.989514131188559e-05,
+      "loss": 1.0325,
+      "step": 110
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.989303440406328e-05,
+      "loss": 1.2602,
+      "step": 111
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.989090656205052e-05,
+      "loss": 1.2366,
+      "step": 112
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.988875778674013e-05,
+      "loss": 0.9532,
+      "step": 113
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.988658807903368e-05,
+      "loss": 1.0827,
+      "step": 114
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.988439743984154e-05,
+      "loss": 1.2726,
+      "step": 115
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.988218587008286e-05,
+      "loss": 1.1021,
+      "step": 116
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.98799533706856e-05,
+      "loss": 0.7431,
+      "step": 117
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.987769994258644e-05,
+      "loss": 0.9969,
+      "step": 118
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.987542558673088e-05,
+      "loss": 1.0436,
+      "step": 119
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.987313030407323e-05,
+      "loss": 0.9666,
+      "step": 120
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.987081409557653e-05,
+      "loss": 1.3716,
+      "step": 121
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.98684769622126e-05,
+      "loss": 1.6575,
+      "step": 122
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.986611890496207e-05,
+      "loss": 1.214,
+      "step": 123
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.986373992481435e-05,
+      "loss": 1.2334,
+      "step": 124
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.98613400227676e-05,
+      "loss": 0.9869,
+      "step": 125
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.985891919982878e-05,
+      "loss": 0.9415,
+      "step": 126
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.985647745701362e-05,
+      "loss": 0.9803,
+      "step": 127
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.985401479534664e-05,
+      "loss": 1.0033,
+      "step": 128
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.985153121586111e-05,
+      "loss": 1.096,
+      "step": 129
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.984902671959911e-05,
+      "loss": 0.9001,
+      "step": 130
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.984650130761146e-05,
+      "loss": 1.1148,
+      "step": 131
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.98439549809578e-05,
+      "loss": 1.0367,
+      "step": 132
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.984138774070651e-05,
+      "loss": 0.7038,
+      "step": 133
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.983879958793476e-05,
+      "loss": 1.0327,
+      "step": 134
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.983619052372848e-05,
+      "loss": 1.178,
+      "step": 135
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.983356054918238e-05,
+      "loss": 1.2365,
+      "step": 136
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.983090966539998e-05,
+      "loss": 1.1545,
+      "step": 137
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.98282378734935e-05,
+      "loss": 0.9914,
+      "step": 138
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.982554517458404e-05,
+      "loss": 0.9601,
+      "step": 139
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.982283156980132e-05,
+      "loss": 0.9457,
+      "step": 140
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.9820097060284e-05,
+      "loss": 1.2792,
+      "step": 141
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.981734164717935e-05,
+      "loss": 1.2215,
+      "step": 142
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.981456533164355e-05,
+      "loss": 0.9904,
+      "step": 143
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.981176811484148e-05,
+      "loss": 1.17,
+      "step": 144
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.980894999794678e-05,
+      "loss": 1.1273,
+      "step": 145
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.980611098214189e-05,
+      "loss": 0.8269,
+      "step": 146
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.980325106861801e-05,
+      "loss": 1.0958,
+      "step": 147
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.98003702585751e-05,
+      "loss": 1.0681,
+      "step": 148
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.979746855322192e-05,
+      "loss": 1.0474,
+      "step": 149
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.979454595377594e-05,
+      "loss": 0.9635,
+      "step": 150
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.979160246146343e-05,
+      "loss": 1.1489,
+      "step": 151
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.978863807751944e-05,
+      "loss": 1.0464,
+      "step": 152
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.978565280318776e-05,
+      "loss": 1.4461,
+      "step": 153
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.978264663972099e-05,
+      "loss": 1.331,
+      "step": 154
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.97796195883804e-05,
+      "loss": 0.7461,
+      "step": 155
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.977657165043612e-05,
+      "loss": 0.9986,
+      "step": 156
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.977350282716702e-05,
+      "loss": 1.0282,
+      "step": 157
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.977041311986071e-05,
+      "loss": 1.1152,
+      "step": 158
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.976730252981355e-05,
+      "loss": 0.7992,
+      "step": 159
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.97641710583307e-05,
+      "loss": 0.8974,
+      "step": 160
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.976101870672609e-05,
+      "loss": 1.0244,
+      "step": 161
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.975784547632236e-05,
+      "loss": 1.1199,
+      "step": 162
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.975465136845095e-05,
+      "loss": 1.2053,
+      "step": 163
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.975143638445205e-05,
+      "loss": 0.9617,
+      "step": 164
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.974820052567459e-05,
+      "loss": 0.9789,
+      "step": 165
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.974494379347631e-05,
+      "loss": 1.0906,
+      "step": 166
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.974166618922367e-05,
+      "loss": 0.9101,
+      "step": 167
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.973836771429184e-05,
+      "loss": 0.9122,
+      "step": 168
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.973504837006486e-05,
+      "loss": 1.3924,
+      "step": 169
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.973170815793543e-05,
+      "loss": 1.1919,
+      "step": 170
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.972834707930505e-05,
+      "loss": 0.9108,
+      "step": 171
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.972496513558398e-05,
+      "loss": 1.0504,
+      "step": 172
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.97215623281912e-05,
+      "loss": 1.2294,
+      "step": 173
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.971813865855448e-05,
+      "loss": 1.4461,
+      "step": 174
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.971469412811032e-05,
+      "loss": 0.9336,
+      "step": 175
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.971122873830398e-05,
+      "loss": 1.1092,
+      "step": 176
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.970774249058948e-05,
+      "loss": 0.7988,
+      "step": 177
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.970423538642958e-05,
+      "loss": 1.0546,
+      "step": 178
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.97007074272958e-05,
+      "loss": 0.9638,
+      "step": 179
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.96971586146684e-05,
+      "loss": 1.0265,
+      "step": 180
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.96935889500364e-05,
+      "loss": 0.858,
+      "step": 181
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.968999843489754e-05,
+      "loss": 0.8436,
+      "step": 182
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.968638707075839e-05,
+      "loss": 1.1531,
+      "step": 183
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.968275485913417e-05,
+      "loss": 1.1621,
+      "step": 184
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.967910180154889e-05,
+      "loss": 1.3556,
+      "step": 185
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.967542789953531e-05,
+      "loss": 1.1165,
+      "step": 186
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.967173315463494e-05,
+      "loss": 1.0506,
+      "step": 187
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.966801756839803e-05,
+      "loss": 1.0308,
+      "step": 188
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.966428114238353e-05,
+      "loss": 0.8997,
+      "step": 189
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.966052387815923e-05,
+      "loss": 1.0588,
+      "step": 190
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.965674577730158e-05,
+      "loss": 1.1032,
+      "step": 191
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.96529468413958e-05,
+      "loss": 1.1754,
+      "step": 192
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.964912707203586e-05,
+      "loss": 1.0954,
+      "step": 193
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.964528647082446e-05,
+      "loss": 1.0769,
+      "step": 194
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.964142503937305e-05,
+      "loss": 0.8973,
+      "step": 195
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.96375427793018e-05,
+      "loss": 1.2487,
+      "step": 196
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.963363969223966e-05,
+      "loss": 1.001,
+      "step": 197
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.962971577982428e-05,
+      "loss": 1.2647,
+      "step": 198
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.962577104370205e-05,
+      "loss": 0.8697,
+      "step": 199
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.962180548552812e-05,
+      "loss": 0.934,
+      "step": 200
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.961781910696636e-05,
+      "loss": 1.0915,
+      "step": 201
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.961381190968939e-05,
+      "loss": 1.1844,
+      "step": 202
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.960978389537854e-05,
+      "loss": 1.0862,
+      "step": 203
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.96057350657239e-05,
+      "loss": 1.1122,
+      "step": 204
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.960166542242429e-05,
+      "loss": 0.9725,
+      "step": 205
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.959757496718723e-05,
+      "loss": 0.9477,
+      "step": 206
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.959346370172901e-05,
+      "loss": 1.1749,
+      "step": 207
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.958933162777469e-05,
+      "loss": 0.8628,
+      "step": 208
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.958517874705793e-05,
+      "loss": 0.9431,
+      "step": 209
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.958100506132127e-05,
+      "loss": 1.1566,
+      "step": 210
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.957681057231586e-05,
+      "loss": 1.08,
+      "step": 211
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.957259528180165e-05,
+      "loss": 1.224,
+      "step": 212
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.956835919154732e-05,
+      "loss": 0.9874,
+      "step": 213
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.956410230333022e-05,
+      "loss": 1.1741,
+      "step": 214
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.955982461893648e-05,
+      "loss": 1.1225,
+      "step": 215
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.955552614016093e-05,
+      "loss": 1.0827,
+      "step": 216
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.955120686880713e-05,
+      "loss": 1.1954,
+      "step": 217
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.954686680668738e-05,
+      "loss": 1.0271,
+      "step": 218
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.954250595562267e-05,
+      "loss": 1.4446,
+      "step": 219
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.953812431744276e-05,
+      "loss": 0.9276,
+      "step": 220
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.953372189398606e-05,
+      "loss": 1.1267,
+      "step": 221
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.952929868709978e-05,
+      "loss": 1.3764,
+      "step": 222
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.952485469863982e-05,
+      "loss": 0.9892,
+      "step": 223
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.952038993047076e-05,
+      "loss": 1.1135,
+      "step": 224
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.951590438446597e-05,
+      "loss": 1.0696,
+      "step": 225
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.951139806250747e-05,
+      "loss": 1.0701,
+      "step": 226
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.950687096648606e-05,
+      "loss": 1.302,
+      "step": 227
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.95023230983012e-05,
+      "loss": 1.2495,
+      "step": 228
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.949775445986111e-05,
+      "loss": 0.8602,
+      "step": 229
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.949316505308271e-05,
+      "loss": 0.7673,
+      "step": 230
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.948855487989161e-05,
+      "loss": 1.0864,
+      "step": 231
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.948392394222215e-05,
+      "loss": 1.051,
+      "step": 232
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.947927224201741e-05,
+      "loss": 1.1546,
+      "step": 233
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.947459978122912e-05,
+      "loss": 1.1426,
+      "step": 234
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.946990656181781e-05,
+      "loss": 0.8608,
+      "step": 235
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.946519258575263e-05,
+      "loss": 1.1615,
+      "step": 236
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.946045785501149e-05,
+      "loss": 0.9881,
+      "step": 237
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.945570237158097e-05,
+      "loss": 1.002,
+      "step": 238
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.945092613745642e-05,
+      "loss": 1.1672,
+      "step": 239
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.944612915464183e-05,
+      "loss": 1.0557,
+      "step": 240
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.944131142514994e-05,
+      "loss": 1.0746,
+      "step": 241
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.943647295100217e-05,
+      "loss": 0.9525,
+      "step": 242
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.943161373422868e-05,
+      "loss": 0.9577,
+      "step": 243
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.942673377686828e-05,
+      "loss": 1.1239,
+      "step": 244
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.942183308096853e-05,
+      "loss": 1.0171,
+      "step": 245
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.941691164858564e-05,
+      "loss": 1.1317,
+      "step": 246
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.941196948178457e-05,
+      "loss": 0.9689,
+      "step": 247
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.940700658263897e-05,
+      "loss": 0.7445,
+      "step": 248
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.940202295323116e-05,
+      "loss": 1.0133,
+      "step": 249
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.93970185956522e-05,
+      "loss": 1.0325,
+      "step": 250
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.939199351200182e-05,
+      "loss": 0.8932,
+      "step": 251
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.938694770438844e-05,
+      "loss": 0.9418,
+      "step": 252
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.93818811749292e-05,
+      "loss": 1.0395,
+      "step": 253
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.93767939257499e-05,
+      "loss": 0.8413,
+      "step": 254
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.93716859589851e-05,
+      "loss": 1.0509,
+      "step": 255
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.936655727677794e-05,
+      "loss": 1.0705,
+      "step": 256
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.936140788128039e-05,
+      "loss": 1.2903,
+      "step": 257
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.9356237774653e-05,
+      "loss": 1.0083,
+      "step": 258
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.935104695906506e-05,
+      "loss": 1.0945,
+      "step": 259
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.934583543669453e-05,
+      "loss": 1.0874,
+      "step": 260
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.93406032097281e-05,
+      "loss": 0.9813,
+      "step": 261
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.933535028036107e-05,
+      "loss": 1.2657,
+      "step": 262
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.933007665079752e-05,
+      "loss": 0.8038,
+      "step": 263
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.932478232325013e-05,
+      "loss": 0.9384,
+      "step": 264
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.931946729994031e-05,
+      "loss": 1.0333,
+      "step": 265
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.931413158309814e-05,
+      "loss": 1.0549,
+      "step": 266
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.930877517496242e-05,
+      "loss": 0.9159,
+      "step": 267
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.930339807778055e-05,
+      "loss": 1.0205,
+      "step": 268
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.92980002938087e-05,
+      "loss": 1.0295,
+      "step": 269
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.929258182531167e-05,
+      "loss": 1.1828,
+      "step": 270
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.928714267456293e-05,
+      "loss": 0.9118,
+      "step": 271
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.928168284384467e-05,
+      "loss": 1.109,
+      "step": 272
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.927620233544772e-05,
+      "loss": 1.2246,
+      "step": 273
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.927070115167161e-05,
+      "loss": 1.0302,
+      "step": 274
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.926517929482453e-05,
+      "loss": 1.094,
+      "step": 275
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.925963676722335e-05,
+      "loss": 1.1204,
+      "step": 276
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.925407357119358e-05,
+      "loss": 0.9183,
+      "step": 277
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.924848970906949e-05,
+      "loss": 1.0012,
+      "step": 278
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.924288518319394e-05,
+      "loss": 0.8502,
+      "step": 279
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.923725999591847e-05,
+      "loss": 1.082,
+      "step": 280
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.923161414960331e-05,
+      "loss": 1.1287,
+      "step": 281
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.922594764661737e-05,
+      "loss": 0.8271,
+      "step": 282
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.922026048933819e-05,
+      "loss": 0.9643,
+      "step": 283
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.921455268015201e-05,
+      "loss": 1.0429,
+      "step": 284
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.920882422145372e-05,
+      "loss": 1.0911,
+      "step": 285
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.920307511564686e-05,
+      "loss": 1.1694,
+      "step": 286
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.919730536514366e-05,
+      "loss": 1.0064,
+      "step": 287
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.919151497236498e-05,
+      "loss": 1.0022,
+      "step": 288
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.918570393974041e-05,
+      "loss": 1.0601,
+      "step": 289
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.91798722697081e-05,
+      "loss": 1.1219,
+      "step": 290
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.917401996471494e-05,
+      "loss": 0.9813,
+      "step": 291
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.916814702721642e-05,
+      "loss": 1.1111,
+      "step": 292
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.916225345967676e-05,
+      "loss": 1.0724,
+      "step": 293
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.915633926456875e-05,
+      "loss": 1.1459,
+      "step": 294
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.915040444437389e-05,
+      "loss": 1.3103,
+      "step": 295
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.914444900158233e-05,
+      "loss": 1.1705,
+      "step": 296
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.913847293869285e-05,
+      "loss": 1.0025,
+      "step": 297
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.913247625821289e-05,
+      "loss": 1.0549,
+      "step": 298
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.912645896265858e-05,
+      "loss": 0.9296,
+      "step": 299
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.912042105455463e-05,
+      "loss": 1.0679,
+      "step": 300
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.911436253643445e-05,
+      "loss": 1.0683,
+      "step": 301
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.910828341084006e-05,
+      "loss": 1.4028,
+      "step": 302
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.910218368032217e-05,
+      "loss": 1.1333,
+      "step": 303
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.909606334744013e-05,
+      "loss": 1.0858,
+      "step": 304
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.908992241476188e-05,
+      "loss": 0.9282,
+      "step": 305
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.908376088486407e-05,
+      "loss": 0.8669,
+      "step": 306
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.907757876033194e-05,
+      "loss": 0.877,
+      "step": 307
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.90713760437594e-05,
+      "loss": 0.8888,
+      "step": 308
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.906515273774902e-05,
+      "loss": 1.1074,
+      "step": 309
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.905890884491195e-05,
+      "loss": 0.9132,
+      "step": 310
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.905264436786804e-05,
+      "loss": 1.1339,
+      "step": 311
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.904635930924573e-05,
+      "loss": 0.996,
+      "step": 312
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.904005367168212e-05,
+      "loss": 1.175,
+      "step": 313
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.903372745782293e-05,
+      "loss": 1.0231,
+      "step": 314
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.902738067032253e-05,
+      "loss": 0.9804,
+      "step": 315
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.90210133118439e-05,
+      "loss": 1.0926,
+      "step": 316
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.90146253850587e-05,
+      "loss": 0.9795,
+      "step": 317
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.900821689264715e-05,
+      "loss": 0.8829,
+      "step": 318
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.900178783729816e-05,
+      "loss": 1.0532,
+      "step": 319
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.899533822170922e-05,
+      "loss": 1.0241,
+      "step": 320
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.898886804858648e-05,
+      "loss": 1.0349,
+      "step": 321
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.898237732064471e-05,
+      "loss": 1.028,
+      "step": 322
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.897586604060729e-05,
+      "loss": 1.2101,
+      "step": 323
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.896933421120622e-05,
+      "loss": 1.1477,
+      "step": 324
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.896278183518216e-05,
+      "loss": 1.0768,
+      "step": 325
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.895620891528437e-05,
+      "loss": 1.271,
+      "step": 326
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.894961545427068e-05,
+      "loss": 1.0788,
+      "step": 327
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.894300145490762e-05,
+      "loss": 1.2054,
+      "step": 328
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.893636691997031e-05,
+      "loss": 1.1744,
+      "step": 329
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.892971185224245e-05,
+      "loss": 1.0626,
+      "step": 330
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.892303625451639e-05,
+      "loss": 0.8727,
+      "step": 331
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.89163401295931e-05,
+      "loss": 1.2652,
+      "step": 332
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.890962348028213e-05,
+      "loss": 0.8391,
+      "step": 333
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.890288630940168e-05,
+      "loss": 0.9487,
+      "step": 334
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.889612861977853e-05,
+      "loss": 0.9691,
+      "step": 335
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.88893504142481e-05,
+      "loss": 1.1253,
+      "step": 336
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.888255169565436e-05,
+      "loss": 1.177,
+      "step": 337
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.887573246684997e-05,
+      "loss": 1.0515,
+      "step": 338
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.886889273069613e-05,
+      "loss": 1.0485,
+      "step": 339
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.886203249006265e-05,
+      "loss": 1.1079,
+      "step": 340
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.8855151747828e-05,
+      "loss": 1.0639,
+      "step": 341
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.884825050687918e-05,
+      "loss": 1.1871,
+      "step": 342
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.884132877011183e-05,
+      "loss": 1.124,
+      "step": 343
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.883438654043018e-05,
+      "loss": 0.9883,
+      "step": 344
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.882742382074707e-05,
+      "loss": 1.1675,
+      "step": 345
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.882044061398392e-05,
+      "loss": 0.9886,
+      "step": 346
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.881343692307075e-05,
+      "loss": 1.023,
+      "step": 347
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.88064127509462e-05,
+      "loss": 0.9649,
+      "step": 348
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.879936810055745e-05,
+      "loss": 0.8554,
+      "step": 349
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.879230297486034e-05,
+      "loss": 1.1711,
+      "step": 350
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.878521737681923e-05,
+      "loss": 1.0751,
+      "step": 351
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.877811130940713e-05,
+      "loss": 1.2364,
+      "step": 352
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.877098477560562e-05,
+      "loss": 1.0071,
+      "step": 353
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.876383777840484e-05,
+      "loss": 1.0568,
+      "step": 354
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.875667032080353e-05,
+      "loss": 0.8281,
+      "step": 355
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.874948240580904e-05,
+      "loss": 1.5203,
+      "step": 356
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.874227403643728e-05,
+      "loss": 1.0448,
+      "step": 357
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.873504521571278e-05,
+      "loss": 1.0515,
+      "step": 358
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.872779594666855e-05,
+      "loss": 0.848,
+      "step": 359
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.872052623234632e-05,
+      "loss": 1.2485,
+      "step": 360
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.871323607579627e-05,
+      "loss": 1.0984,
+      "step": 361
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.870592548007724e-05,
+      "loss": 0.9803,
+      "step": 362
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.869859444825663e-05,
+      "loss": 1.1524,
+      "step": 363
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.869124298341039e-05,
+      "loss": 0.8972,
+      "step": 364
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.868387108862307e-05,
+      "loss": 0.9175,
+      "step": 365
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.867647876698775e-05,
+      "loss": 1.119,
+      "step": 366
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.866906602160616e-05,
+      "loss": 0.9837,
+      "step": 367
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.866163285558851e-05,
+      "loss": 0.9582,
+      "step": 368
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.865417927205363e-05,
+      "loss": 1.0033,
+      "step": 369
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.864670527412891e-05,
+      "loss": 1.0095,
+      "step": 370
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.86392108649503e-05,
+      "loss": 1.1471,
+      "step": 371
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.863169604766231e-05,
+      "loss": 1.0538,
+      "step": 372
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.862416082541802e-05,
+      "loss": 0.9899,
+      "step": 373
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.861660520137908e-05,
+      "loss": 1.0831,
+      "step": 374
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.860902917871567e-05,
+      "loss": 1.0769,
+      "step": 375
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.860143276060655e-05,
+      "loss": 1.0256,
+      "step": 376
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.859381595023904e-05,
+      "loss": 1.0917,
+      "step": 377
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.858617875080903e-05,
+      "loss": 1.0534,
+      "step": 378
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.857852116552093e-05,
+      "loss": 0.8816,
+      "step": 379
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.857084319758772e-05,
+      "loss": 1.1839,
+      "step": 380
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.856314485023094e-05,
+      "loss": 1.0795,
+      "step": 381
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.855542612668066e-05,
+      "loss": 1.1038,
+      "step": 382
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.854768703017552e-05,
+      "loss": 1.2518,
+      "step": 383
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.853992756396272e-05,
+      "loss": 0.7819,
+      "step": 384
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.853214773129796e-05,
+      "loss": 1.1036,
+      "step": 385
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.852434753544552e-05,
+      "loss": 0.9392,
+      "step": 386
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.851652697967824e-05,
+      "loss": 1.0506,
+      "step": 387
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.850868606727745e-05,
+      "loss": 1.0029,
+      "step": 388
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.850082480153307e-05,
+      "loss": 1.232,
+      "step": 389
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.849294318574351e-05,
+      "loss": 0.9588,
+      "step": 390
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.84850412232158e-05,
+      "loss": 1.2153,
+      "step": 391
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.847711891726544e-05,
+      "loss": 0.8345,
+      "step": 392
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.846917627121644e-05,
+      "loss": 1.1035,
+      "step": 393
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.846121328840143e-05,
+      "loss": 1.4204,
+      "step": 394
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.845322997216153e-05,
+      "loss": 1.0985,
+      "step": 395
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.844522632584637e-05,
+      "loss": 0.9537,
+      "step": 396
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.843720235281414e-05,
+      "loss": 1.2347,
+      "step": 397
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.842915805643155e-05,
+      "loss": 0.8923,
+      "step": 398
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.842109344007386e-05,
+      "loss": 1.1004,
+      "step": 399
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.84130085071248e-05,
+      "loss": 1.189,
+      "step": 400
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.840490326097666e-05,
+      "loss": 0.9773,
+      "step": 401
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.839677770503027e-05,
+      "loss": 0.8302,
+      "step": 402
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.838863184269495e-05,
+      "loss": 1.1312,
+      "step": 403
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.838046567738857e-05,
+      "loss": 1.1039,
+      "step": 404
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.837227921253746e-05,
+      "loss": 1.0369,
+      "step": 405
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.836407245157655e-05,
+      "loss": 1.0106,
+      "step": 406
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.835584539794924e-05,
+      "loss": 0.8394,
+      "step": 407
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.834759805510743e-05,
+      "loss": 1.4596,
+      "step": 408
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.833933042651156e-05,
+      "loss": 1.0272,
+      "step": 409
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.833104251563056e-05,
+      "loss": 1.1613,
+      "step": 410
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.832273432594193e-05,
+      "loss": 1.0418,
+      "step": 411
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.831440586093157e-05,
+      "loss": 1.1828,
+      "step": 412
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.830605712409399e-05,
+      "loss": 0.9779,
+      "step": 413
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.829768811893214e-05,
+      "loss": 1.1952,
+      "step": 414
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.828929884895752e-05,
+      "loss": 1.2963,
+      "step": 415
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.828088931769013e-05,
+      "loss": 1.3445,
+      "step": 416
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.82724595286584e-05,
+      "loss": 1.1619,
+      "step": 417
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.826400948539934e-05,
+      "loss": 1.2252,
+      "step": 418
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.825553919145845e-05,
+      "loss": 1.0359,
+      "step": 419
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.824704865038968e-05,
+      "loss": 0.9922,
+      "step": 420
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.823853786575549e-05,
+      "loss": 1.0357,
+      "step": 421
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.823000684112691e-05,
+      "loss": 1.1699,
+      "step": 422
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.822145558008334e-05,
+      "loss": 1.0295,
+      "step": 423
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.821288408621275e-05,
+      "loss": 1.434,
+      "step": 424
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.820429236311158e-05,
+      "loss": 1.2936,
+      "step": 425
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.819568041438476e-05,
+      "loss": 0.934,
+      "step": 426
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.81870482436457e-05,
+      "loss": 0.9372,
+      "step": 427
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.817839585451629e-05,
+      "loss": 1.0567,
+      "step": 428
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.816972325062694e-05,
+      "loss": 1.0504,
+      "step": 429
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.816103043561648e-05,
+      "loss": 0.9764,
+      "step": 430
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.815231741313227e-05,
+      "loss": 0.9363,
+      "step": 431
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.814358418683013e-05,
+      "loss": 1.0778,
+      "step": 432
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.813483076037437e-05,
+      "loss": 0.8744,
+      "step": 433
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.812605713743774e-05,
+      "loss": 1.2378,
+      "step": 434
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.811726332170153e-05,
+      "loss": 1.0535,
+      "step": 435
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.810844931685541e-05,
+      "loss": 1.0929,
+      "step": 436
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.80996151265976e-05,
+      "loss": 1.0065,
+      "step": 437
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.809076075463477e-05,
+      "loss": 1.094,
+      "step": 438
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.808188620468204e-05,
+      "loss": 0.9107,
+      "step": 439
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.8072991480463e-05,
+      "loss": 1.0198,
+      "step": 440
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.806407658570973e-05,
+      "loss": 0.9693,
+      "step": 441
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.805514152416273e-05,
+      "loss": 0.9099,
+      "step": 442
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.8046186299571e-05,
+      "loss": 0.9807,
+      "step": 443
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.8037210915692e-05,
+      "loss": 1.4131,
+      "step": 444
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.802821537629161e-05,
+      "loss": 0.9845,
+      "step": 445
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.801919968514419e-05,
+      "loss": 1.1458,
+      "step": 446
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.801016384603259e-05,
+      "loss": 0.9317,
+      "step": 447
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.800110786274803e-05,
+      "loss": 0.922,
+      "step": 448
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.799203173909027e-05,
+      "loss": 1.0796,
+      "step": 449
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.798293547886748e-05,
+      "loss": 1.0336,
+      "step": 450
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.797381908589627e-05,
+      "loss": 1.0539,
+      "step": 451
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.796468256400171e-05,
+      "loss": 1.166,
+      "step": 452
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.795552591701731e-05,
+      "loss": 1.221,
+      "step": 453
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.794634914878504e-05,
+      "loss": 0.964,
+      "step": 454
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.79371522631553e-05,
+      "loss": 0.8731,
+      "step": 455
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.792793526398694e-05,
+      "loss": 1.1294,
+      "step": 456
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.791869815514723e-05,
+      "loss": 1.0135,
+      "step": 457
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.790944094051187e-05,
+      "loss": 0.8291,
+      "step": 458
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.790016362396505e-05,
+      "loss": 0.9696,
+      "step": 459
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.789086620939936e-05,
+      "loss": 1.3319,
+      "step": 460
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.78815487007158e-05,
+      "loss": 1.0598,
+      "step": 461
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.787221110182383e-05,
+      "loss": 0.9759,
+      "step": 462
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.786285341664134e-05,
+      "loss": 0.7522,
+      "step": 463
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.785347564909463e-05,
+      "loss": 0.9594,
+      "step": 464
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.784407780311845e-05,
+      "loss": 1.4998,
+      "step": 465
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.783465988265595e-05,
+      "loss": 1.0285,
+      "step": 466
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.782522189165873e-05,
+      "loss": 1.0501,
+      "step": 467
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.781576383408677e-05,
+      "loss": 0.8006,
+      "step": 468
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.780628571390852e-05,
+      "loss": 1.2391,
+      "step": 469
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.779678753510081e-05,
+      "loss": 1.1994,
+      "step": 470
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.77872693016489e-05,
+      "loss": 1.0251,
+      "step": 471
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.777773101754649e-05,
+      "loss": 1.0818,
+      "step": 472
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.776817268679561e-05,
+      "loss": 1.3074,
+      "step": 473
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.775859431340682e-05,
+      "loss": 1.1576,
+      "step": 474
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.774899590139896e-05,
+      "loss": 1.0429,
+      "step": 475
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.773937745479942e-05,
+      "loss": 1.0724,
+      "step": 476
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.772973897764389e-05,
+      "loss": 1.0283,
+      "step": 477
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.772008047397647e-05,
+      "loss": 1.1658,
+      "step": 478
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.771040194784973e-05,
+      "loss": 1.2198,
+      "step": 479
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.770070340332456e-05,
+      "loss": 0.8815,
+      "step": 480
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.769098484447033e-05,
+      "loss": 0.9419,
+      "step": 481
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.768124627536474e-05,
+      "loss": 0.9714,
+      "step": 482
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.767148770009393e-05,
+      "loss": 0.8902,
+      "step": 483
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.766170912275239e-05,
+      "loss": 1.1096,
+      "step": 484
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.765191054744305e-05,
+      "loss": 0.7925,
+      "step": 485
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.76420919782772e-05,
+      "loss": 1.0977,
+      "step": 486
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.763225341937455e-05,
+      "loss": 1.0678,
+      "step": 487
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.762239487486315e-05,
+      "loss": 0.9857,
+      "step": 488
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.761251634887949e-05,
+      "loss": 0.9883,
+      "step": 489
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.760261784556839e-05,
+      "loss": 1.1017,
+      "step": 490
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.759269936908308e-05,
+      "loss": 1.1257,
+      "step": 491
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.758276092358518e-05,
+      "loss": 1.1032,
+      "step": 492
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.757280251324468e-05,
+      "loss": 1.3212,
+      "step": 493
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.756282414223994e-05,
+      "loss": 0.9329,
+      "step": 494
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.755282581475769e-05,
+      "loss": 1.2639,
+      "step": 495
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.754280753499305e-05,
+      "loss": 1.0943,
+      "step": 496
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.753276930714949e-05,
+      "loss": 0.9838,
+      "step": 497
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.75227111354389e-05,
+      "loss": 0.955,
+      "step": 498
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.751263302408147e-05,
+      "loss": 0.8907,
+      "step": 499
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.75025349773058e-05,
+      "loss": 0.9984,
+      "step": 500
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.749241699934884e-05,
+      "loss": 1.0013,
+      "step": 501
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.74822790944559e-05,
+      "loss": 0.8387,
+      "step": 502
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.747212126688067e-05,
+      "loss": 1.0543,
+      "step": 503
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.746194352088518e-05,
+      "loss": 0.8775,
+      "step": 504
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.745174586073981e-05,
+      "loss": 1.2384,
+      "step": 505
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.744152829072333e-05,
+      "loss": 1.1225,
+      "step": 506
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.743129081512283e-05,
+      "loss": 1.1823,
+      "step": 507
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.742103343823376e-05,
+      "loss": 0.8957,
+      "step": 508
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.741075616435995e-05,
+      "loss": 0.9828,
+      "step": 509
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.740045899781352e-05,
+      "loss": 1.156,
+      "step": 510
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.7390141942915e-05,
+      "loss": 0.914,
+      "step": 511
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.737980500399322e-05,
+      "loss": 1.0343,
+      "step": 512
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.736944818538537e-05,
+      "loss": 1.1493,
+      "step": 513
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.735907149143694e-05,
+      "loss": 1.2059,
+      "step": 514
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.734867492650186e-05,
+      "loss": 0.8702,
+      "step": 515
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.733825849494231e-05,
+      "loss": 1.0307,
+      "step": 516
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.732782220112883e-05,
+      "loss": 1.1787,
+      "step": 517
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.73173660494403e-05,
+      "loss": 1.1051,
+      "step": 518
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.730689004426392e-05,
+      "loss": 1.054,
+      "step": 519
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.729639418999523e-05,
+      "loss": 1.1327,
+      "step": 520
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.72858784910381e-05,
+      "loss": 0.7935,
+      "step": 521
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.727534295180471e-05,
+      "loss": 1.1044,
+      "step": 522
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.726478757671559e-05,
+      "loss": 0.9102,
+      "step": 523
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.725421237019957e-05,
+      "loss": 1.1146,
+      "step": 524
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.724361733669381e-05,
+      "loss": 1.2427,
+      "step": 525
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.723300248064379e-05,
+      "loss": 1.2277,
+      "step": 526
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.722236780650334e-05,
+      "loss": 0.9673,
+      "step": 527
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.721171331873451e-05,
+      "loss": 0.9208,
+      "step": 528
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.720103902180776e-05,
+      "loss": 1.0357,
+      "step": 529
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.719034492020183e-05,
+      "loss": 0.9537,
+      "step": 530
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.717963101840375e-05,
+      "loss": 1.0881,
+      "step": 531
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.716889732090889e-05,
+      "loss": 1.0546,
+      "step": 532
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.715814383222091e-05,
+      "loss": 1.1128,
+      "step": 533
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.714737055685176e-05,
+      "loss": 1.1669,
+      "step": 534
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.713657749932172e-05,
+      "loss": 1.0863,
+      "step": 535
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.712576466415935e-05,
+      "loss": 1.1035,
+      "step": 536
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.71149320559015e-05,
+      "loss": 1.0527,
+      "step": 537
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.710407967909336e-05,
+      "loss": 1.2313,
+      "step": 538
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.709320753828837e-05,
+      "loss": 1.3694,
+      "step": 539
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.708231563804828e-05,
+      "loss": 0.9448,
+      "step": 540
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.707140398294312e-05,
+      "loss": 1.0452,
+      "step": 541
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.706047257755124e-05,
+      "loss": 1.0658,
+      "step": 542
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.704952142645925e-05,
+      "loss": 1.1795,
+      "step": 543
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.703855053426202e-05,
+      "loss": 0.8224,
+      "step": 544
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.702755990556275e-05,
+      "loss": 1.2008,
+      "step": 545
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.701654954497293e-05,
+      "loss": 0.9103,
+      "step": 546
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.700551945711228e-05,
+      "loss": 1.1161,
+      "step": 547
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.699446964660881e-05,
+      "loss": 1.2181,
+      "step": 548
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.698340011809883e-05,
+      "loss": 1.0115,
+      "step": 549
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.697231087622691e-05,
+      "loss": 1.0503,
+      "step": 550
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.696120192564586e-05,
+      "loss": 1.023,
+      "step": 551
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.695007327101684e-05,
+      "loss": 1.293,
+      "step": 552
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.693892491700919e-05,
+      "loss": 1.0001,
+      "step": 553
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.692775686830055e-05,
+      "loss": 1.3113,
+      "step": 554
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.691656912957685e-05,
+      "loss": 1.1561,
+      "step": 555
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.690536170553226e-05,
+      "loss": 1.079,
+      "step": 556
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.689413460086919e-05,
+      "loss": 0.9921,
+      "step": 557
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.68828878202983e-05,
+      "loss": 1.1777,
+      "step": 558
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.687162136853857e-05,
+      "loss": 0.8292,
+      "step": 559
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.686033525031719e-05,
+      "loss": 0.9962,
+      "step": 560
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.684902947036959e-05,
+      "loss": 1.1259,
+      "step": 561
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.683770403343947e-05,
+      "loss": 1.2877,
+      "step": 562
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.682635894427878e-05,
+      "loss": 1.0675,
+      "step": 563
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.68149942076477e-05,
+      "loss": 0.9408,
+      "step": 564
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.680360982831466e-05,
+      "loss": 1.2813,
+      "step": 565
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.679220581105635e-05,
+      "loss": 1.1893,
+      "step": 566
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.678078216065767e-05,
+      "loss": 1.1375,
+      "step": 567
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.676933888191177e-05,
+      "loss": 1.0743,
+      "step": 568
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.675787597962007e-05,
+      "loss": 1.0379,
+      "step": 569
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.674639345859214e-05,
+      "loss": 1.131,
+      "step": 570
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.673489132364586e-05,
+      "loss": 0.9675,
+      "step": 571
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.67233695796073e-05,
+      "loss": 0.982,
+      "step": 572
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.671182823131078e-05,
+      "loss": 0.8347,
+      "step": 573
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.670026728359884e-05,
+      "loss": 1.1948,
+      "step": 574
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.668868674132224e-05,
+      "loss": 1.0214,
+      "step": 575
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.667708660933994e-05,
+      "loss": 1.061,
+      "step": 576
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.666546689251915e-05,
+      "loss": 0.9996,
+      "step": 577
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.66538275957353e-05,
+      "loss": 1.0239,
+      "step": 578
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.6642168723872e-05,
+      "loss": 1.2928,
+      "step": 579
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.663049028182111e-05,
+      "loss": 1.1614,
+      "step": 580
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.661879227448269e-05,
+      "loss": 0.9677,
+      "step": 581
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.660707470676502e-05,
+      "loss": 1.034,
+      "step": 582
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.659533758358455e-05,
+      "loss": 1.0127,
+      "step": 583
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.658358090986594e-05,
+      "loss": 1.16,
+      "step": 584
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.657180469054213e-05,
+      "loss": 1.4194,
+      "step": 585
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.656000893055414e-05,
+      "loss": 0.9417,
+      "step": 586
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.65481936348513e-05,
+      "loss": 1.1485,
+      "step": 587
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.653635880839106e-05,
+      "loss": 1.2903,
+      "step": 588
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.652450445613912e-05,
+      "loss": 0.7741,
+      "step": 589
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.651263058306932e-05,
+      "loss": 1.2572,
+      "step": 590
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.650073719416372e-05,
+      "loss": 1.0564,
+      "step": 591
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.648882429441257e-05,
+      "loss": 1.0292,
+      "step": 592
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.647689188881431e-05,
+      "loss": 1.1883,
+      "step": 593
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.646493998237556e-05,
+      "loss": 0.9445,
+      "step": 594
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.645296858011109e-05,
+      "loss": 1.0996,
+      "step": 595
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.64409776870439e-05,
+      "loss": 1.0008,
+      "step": 596
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.642896730820514e-05,
+      "loss": 1.1168,
+      "step": 597
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.641693744863413e-05,
+      "loss": 1.0211,
+      "step": 598
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.64048881133784e-05,
+      "loss": 1.0488,
+      "step": 599
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.639281930749362e-05,
+      "loss": 1.2482,
+      "step": 600
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.638073103604364e-05,
+      "loss": 1.3107,
+      "step": 601
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.636862330410044e-05,
+      "loss": 0.9444,
+      "step": 602
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.635649611674424e-05,
+      "loss": 1.237,
+      "step": 603
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.634434947906336e-05,
+      "loss": 1.0475,
+      "step": 604
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.633218339615433e-05,
+      "loss": 1.1722,
+      "step": 605
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.631999787312178e-05,
+      "loss": 0.8417,
+      "step": 606
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.630779291507854e-05,
+      "loss": 1.1301,
+      "step": 607
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.62955685271456e-05,
+      "loss": 1.1955,
+      "step": 608
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.628332471445206e-05,
+      "loss": 1.1969,
+      "step": 609
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.627106148213522e-05,
+      "loss": 1.2363,
+      "step": 610
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.625877883534048e-05,
+      "loss": 0.9747,
+      "step": 611
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.624647677922142e-05,
+      "loss": 1.1625,
+      "step": 612
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.623415531893978e-05,
+      "loss": 1.0624,
+      "step": 613
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.622181445966539e-05,
+      "loss": 0.8885,
+      "step": 614
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.620945420657624e-05,
+      "loss": 0.7688,
+      "step": 615
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.619707456485848e-05,
+      "loss": 0.9219,
+      "step": 616
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.618467553970635e-05,
+      "loss": 1.0255,
+      "step": 617
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.61722571363223e-05,
+      "loss": 0.9286,
+      "step": 618
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.615981935991683e-05,
+      "loss": 1.1151,
+      "step": 619
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.61473622157086e-05,
+      "loss": 0.958,
+      "step": 620
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.613488570892437e-05,
+      "loss": 1.0244,
+      "step": 621
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.61223898447991e-05,
+      "loss": 1.433,
+      "step": 622
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.61098746285758e-05,
+      "loss": 0.8944,
+      "step": 623
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.609734006550562e-05,
+      "loss": 0.9741,
+      "step": 624
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.608478616084784e-05,
+      "loss": 1.1113,
+      "step": 625
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.607221291986983e-05,
+      "loss": 1.2754,
+      "step": 626
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.60596203478471e-05,
+      "loss": 1.0289,
+      "step": 627
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.604700845006326e-05,
+      "loss": 0.998,
+      "step": 628
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.603437723181001e-05,
+      "loss": 0.9494,
+      "step": 629
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.602172669838721e-05,
+      "loss": 1.1897,
+      "step": 630
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.600905685510275e-05,
+      "loss": 0.7718,
+      "step": 631
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.599636770727269e-05,
+      "loss": 0.9619,
+      "step": 632
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.598365926022115e-05,
+      "loss": 1.0731,
+      "step": 633
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.597093151928034e-05,
+      "loss": 1.0211,
+      "step": 634
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.595818448979061e-05,
+      "loss": 0.7484,
+      "step": 635
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.594541817710037e-05,
+      "loss": 0.7567,
+      "step": 636
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.593263258656613e-05,
+      "loss": 1.0972,
+      "step": 637
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.591982772355247e-05,
+      "loss": 1.0645,
+      "step": 638
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.59070035934321e-05,
+      "loss": 1.0851,
+      "step": 639
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.589416020158578e-05,
+      "loss": 1.0303,
+      "step": 640
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.588129755340231e-05,
+      "loss": 0.9137,
+      "step": 641
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.58684156542787e-05,
+      "loss": 1.0701,
+      "step": 642
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.58555145096199e-05,
+      "loss": 1.0387,
+      "step": 643
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.584259412483897e-05,
+      "loss": 0.9965,
+      "step": 644
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.582965450535715e-05,
+      "loss": 1.1828,
+      "step": 645
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.581669565660359e-05,
+      "loss": 1.021,
+      "step": 646
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.58037175840156e-05,
+      "loss": 1.1555,
+      "step": 647
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.579072029303854e-05,
+      "loss": 0.8512,
+      "step": 648
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.577770378912584e-05,
+      "loss": 1.2018,
+      "step": 649
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.576466807773899e-05,
+      "loss": 1.2347,
+      "step": 650
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.575161316434749e-05,
+      "loss": 1.0443,
+      "step": 651
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.5738539054429e-05,
+      "loss": 1.3083,
+      "step": 652
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.572544575346911e-05,
+      "loss": 1.1882,
+      "step": 653
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.571233326696159e-05,
+      "loss": 1.0114,
+      "step": 654
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.569920160040815e-05,
+      "loss": 1.2273,
+      "step": 655
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.56860507593186e-05,
+      "loss": 1.0298,
+      "step": 656
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.567288074921082e-05,
+      "loss": 1.1075,
+      "step": 657
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.565969157561065e-05,
+      "loss": 0.8936,
+      "step": 658
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.564648324405206e-05,
+      "loss": 1.1156,
+      "step": 659
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.563325576007701e-05,
+      "loss": 1.0966,
+      "step": 660
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.562000912923552e-05,
+      "loss": 0.9397,
+      "step": 661
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.56067433570856e-05,
+      "loss": 1.0421,
+      "step": 662
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.559345844919334e-05,
+      "loss": 1.021,
+      "step": 663
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.558015441113285e-05,
+      "loss": 0.8029,
+      "step": 664
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.556683124848625e-05,
+      "loss": 0.9206,
+      "step": 665
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.555348896684366e-05,
+      "loss": 0.919,
+      "step": 666
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.554012757180331e-05,
+      "loss": 1.0957,
+      "step": 667
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.552674706897136e-05,
+      "loss": 0.7554,
+      "step": 668
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.551334746396203e-05,
+      "loss": 1.1522,
+      "step": 669
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.549992876239753e-05,
+      "loss": 0.9873,
+      "step": 670
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.548649096990812e-05,
+      "loss": 1.2701,
+      "step": 671
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.547303409213202e-05,
+      "loss": 0.8475,
+      "step": 672
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.545955813471551e-05,
+      "loss": 1.0528,
+      "step": 673
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.544606310331284e-05,
+      "loss": 0.97,
+      "step": 674
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.543254900358629e-05,
+      "loss": 1.0467,
+      "step": 675
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.541901584120612e-05,
+      "loss": 1.2231,
+      "step": 676
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.540546362185057e-05,
+      "loss": 1.0857,
+      "step": 677
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.539189235120592e-05,
+      "loss": 1.0792,
+      "step": 678
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.537830203496642e-05,
+      "loss": 0.7925,
+      "step": 679
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.536469267883433e-05,
+      "loss": 0.9108,
+      "step": 680
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.535106428851984e-05,
+      "loss": 1.0923,
+      "step": 681
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.533741686974121e-05,
+      "loss": 1.0835,
+      "step": 682
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.532375042822463e-05,
+      "loss": 0.968,
+      "step": 683
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.531006496970429e-05,
+      "loss": 1.0502,
+      "step": 684
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.529636049992234e-05,
+      "loss": 1.0617,
+      "step": 685
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.528263702462894e-05,
+      "loss": 1.2041,
+      "step": 686
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.526889454958219e-05,
+      "loss": 1.0652,
+      "step": 687
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.525513308054819e-05,
+      "loss": 1.1646,
+      "step": 688
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.524135262330098e-05,
+      "loss": 1.3245,
+      "step": 689
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.52275531836226e-05,
+      "loss": 1.0545,
+      "step": 690
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.521373476730303e-05,
+      "loss": 0.8111,
+      "step": 691
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.519989738014022e-05,
+      "loss": 1.1101,
+      "step": 692
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.518604102794009e-05,
+      "loss": 0.9642,
+      "step": 693
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.51721657165165e-05,
+      "loss": 1.2172,
+      "step": 694
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.515827145169127e-05,
+      "loss": 0.9922,
+      "step": 695
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.514435823929417e-05,
+      "loss": 1.02,
+      "step": 696
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.513042608516297e-05,
+      "loss": 0.927,
+      "step": 697
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.511647499514326e-05,
+      "loss": 1.0792,
+      "step": 698
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.510250497508873e-05,
+      "loss": 1.0793,
+      "step": 699
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.508851603086093e-05,
+      "loss": 1.1018,
+      "step": 700
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.507450816832933e-05,
+      "loss": 1.1461,
+      "step": 701
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.50604813933714e-05,
+      "loss": 1.045,
+      "step": 702
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.50464357118725e-05,
+      "loss": 1.1387,
+      "step": 703
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.503237112972594e-05,
+      "loss": 1.2943,
+      "step": 704
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.501828765283295e-05,
+      "loss": 1.298,
+      "step": 705
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.50041852871027e-05,
+      "loss": 1.0739,
+      "step": 706
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.499006403845229e-05,
+      "loss": 1.2861,
+      "step": 707
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.497592391280673e-05,
+      "loss": 1.0856,
+      "step": 708
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.496176491609893e-05,
+      "loss": 1.1481,
+      "step": 709
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.494758705426978e-05,
+      "loss": 0.8815,
+      "step": 710
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.493339033326801e-05,
+      "loss": 0.9163,
+      "step": 711
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.491917475905035e-05,
+      "loss": 0.8947,
+      "step": 712
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.490494033758133e-05,
+      "loss": 0.9286,
+      "step": 713
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.489068707483349e-05,
+      "loss": 1.0503,
+      "step": 714
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.487641497678723e-05,
+      "loss": 1.0833,
+      "step": 715
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.486212404943084e-05,
+      "loss": 1.2958,
+      "step": 716
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.484781429876055e-05,
+      "loss": 1.0803,
+      "step": 717
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.483348573078046e-05,
+      "loss": 1.1405,
+      "step": 718
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.481913835150255e-05,
+      "loss": 1.1074,
+      "step": 719
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.480477216694673e-05,
+      "loss": 1.0363,
+      "step": 720
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.479038718314079e-05,
+      "loss": 0.9753,
+      "step": 721
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.47759834061204e-05,
+      "loss": 1.0439,
+      "step": 722
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.476156084192911e-05,
+      "loss": 1.0519,
+      "step": 723
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.474711949661835e-05,
+      "loss": 1.1586,
+      "step": 724
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.473265937624746e-05,
+      "loss": 1.1556,
+      "step": 725
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.471818048688363e-05,
+      "loss": 0.8811,
+      "step": 726
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.470368283460194e-05,
+      "loss": 0.962,
+      "step": 727
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.468916642548533e-05,
+      "loss": 1.0572,
+      "step": 728
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.467463126562461e-05,
+      "loss": 1.1101,
+      "step": 729
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.466007736111847e-05,
+      "loss": 1.1112,
+      "step": 730
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.464550471807345e-05,
+      "loss": 1.1914,
+      "step": 731
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.463091334260396e-05,
+      "loss": 0.8854,
+      "step": 732
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.461630324083228e-05,
+      "loss": 0.9479,
+      "step": 733
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.460167441888854e-05,
+      "loss": 0.8752,
+      "step": 734
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.458702688291073e-05,
+      "loss": 1.1942,
+      "step": 735
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.457236063904464e-05,
+      "loss": 1.0841,
+      "step": 736
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.455767569344401e-05,
+      "loss": 1.1861,
+      "step": 737
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.454297205227034e-05,
+      "loss": 1.1128,
+      "step": 738
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.452824972169299e-05,
+      "loss": 0.8303,
+      "step": 739
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.45135087078892e-05,
+      "loss": 1.0511,
+      "step": 740
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.449874901704403e-05,
+      "loss": 1.0746,
+      "step": 741
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.448397065535036e-05,
+      "loss": 0.935,
+      "step": 742
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.446917362900891e-05,
+      "loss": 1.0426,
+      "step": 743
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.445435794422825e-05,
+      "loss": 0.9834,
+      "step": 744
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.443952360722477e-05,
+      "loss": 1.149,
+      "step": 745
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.442467062422266e-05,
+      "loss": 0.8691,
+      "step": 746
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.440979900145399e-05,
+      "loss": 1.0306,
+      "step": 747
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.439490874515858e-05,
+      "loss": 0.8991,
+      "step": 748
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.437999986158412e-05,
+      "loss": 1.0767,
+      "step": 749
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.436507235698612e-05,
+      "loss": 0.7964,
+      "step": 750
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.435012623762784e-05,
+      "loss": 0.9606,
+      "step": 751
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.433516150978044e-05,
+      "loss": 1.2193,
+      "step": 752
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.432017817972281e-05,
+      "loss": 1.1044,
+      "step": 753
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.430517625374172e-05,
+      "loss": 1.028,
+      "step": 754
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.429015573813163e-05,
+      "loss": 1.1132,
+      "step": 755
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.427511663919491e-05,
+      "loss": 1.0307,
+      "step": 756
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.42600589632417e-05,
+      "loss": 0.8855,
+      "step": 757
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.424498271658991e-05,
+      "loss": 0.8868,
+      "step": 758
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.422988790556523e-05,
+      "loss": 1.0945,
+      "step": 759
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.421477453650118e-05,
+      "loss": 0.8206,
+      "step": 760
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.419964261573906e-05,
+      "loss": 1.0949,
+      "step": 761
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.418449214962793e-05,
+      "loss": 0.9823,
+      "step": 762
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.416932314452465e-05,
+      "loss": 1.1499,
+      "step": 763
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.415413560679385e-05,
+      "loss": 1.1132,
+      "step": 764
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.413892954280792e-05,
+      "loss": 1.1692,
+      "step": 765
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.412370495894707e-05,
+      "loss": 1.0275,
+      "step": 766
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.410846186159925e-05,
+      "loss": 1.1828,
+      "step": 767
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.409320025716017e-05,
+      "loss": 0.9516,
+      "step": 768
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.407792015203332e-05,
+      "loss": 1.1898,
+      "step": 769
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.406262155262994e-05,
+      "loss": 1.2173,
+      "step": 770
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.404730446536904e-05,
+      "loss": 0.9497,
+      "step": 771
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.403196889667742e-05,
+      "loss": 1.1062,
+      "step": 772
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.401661485298954e-05,
+      "loss": 0.9888,
+      "step": 773
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.400124234074771e-05,
+      "loss": 1.0737,
+      "step": 774
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.398585136640194e-05,
+      "loss": 1.0244,
+      "step": 775
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.397044193641e-05,
+      "loss": 1.2858,
+      "step": 776
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.395501405723741e-05,
+      "loss": 1.129,
+      "step": 777
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.393956773535742e-05,
+      "loss": 0.8439,
+      "step": 778
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.392410297725098e-05,
+      "loss": 1.0561,
+      "step": 779
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.390861978940686e-05,
+      "loss": 1.0668,
+      "step": 780
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.389311817832152e-05,
+      "loss": 1.3748,
+      "step": 781
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.38775981504991e-05,
+      "loss": 1.1372,
+      "step": 782
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.386205971245157e-05,
+      "loss": 1.4245,
+      "step": 783
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.384650287069856e-05,
+      "loss": 1.1035,
+      "step": 784
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.38309276317674e-05,
+      "loss": 1.1305,
+      "step": 785
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.381533400219318e-05,
+      "loss": 1.0785,
+      "step": 786
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.379972198851874e-05,
+      "loss": 1.0058,
+      "step": 787
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.378409159729454e-05,
+      "loss": 1.0262,
+      "step": 788
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.37684428350788e-05,
+      "loss": 0.9302,
+      "step": 789
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.37527757084375e-05,
+      "loss": 1.0855,
+      "step": 790
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.373709022394423e-05,
+      "loss": 1.1637,
+      "step": 791
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.372138638818035e-05,
+      "loss": 1.0654,
+      "step": 792
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.37056642077349e-05,
+      "loss": 0.9723,
+      "step": 793
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.36899236892046e-05,
+      "loss": 1.1771,
+      "step": 794
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.367416483919387e-05,
+      "loss": 1.2174,
+      "step": 795
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.365838766431488e-05,
+      "loss": 1.1088,
+      "step": 796
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.364259217118737e-05,
+      "loss": 1.0142,
+      "step": 797
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.362677836643889e-05,
+      "loss": 1.016,
+      "step": 798
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.36109462567046e-05,
+      "loss": 0.8343,
+      "step": 799
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.359509584862736e-05,
+      "loss": 1.1529,
+      "step": 800
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.35792271488577e-05,
+      "loss": 1.0846,
+      "step": 801
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.356334016405384e-05,
+      "loss": 1.0623,
+      "step": 802
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.354743490088166e-05,
+      "loss": 1.22,
+      "step": 803
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.353151136601471e-05,
+      "loss": 1.0062,
+      "step": 804
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.351556956613423e-05,
+      "loss": 0.8984,
+      "step": 805
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.349960950792907e-05,
+      "loss": 1.0969,
+      "step": 806
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.34836311980958e-05,
+      "loss": 1.1753,
+      "step": 807
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.34676346433386e-05,
+      "loss": 1.0486,
+      "step": 808
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.345161985036937e-05,
+      "loss": 0.9182,
+      "step": 809
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.343558682590756e-05,
+      "loss": 1.2035,
+      "step": 810
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.341953557668037e-05,
+      "loss": 1.2915,
+      "step": 811
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.340346610942258e-05,
+      "loss": 0.9568,
+      "step": 812
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.338737843087669e-05,
+      "loss": 1.1717,
+      "step": 813
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.337127254779272e-05,
+      "loss": 1.2607,
+      "step": 814
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.335514846692845e-05,
+      "loss": 1.0643,
+      "step": 815
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.333900619504923e-05,
+      "loss": 0.9427,
+      "step": 816
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.332284573892807e-05,
+      "loss": 1.023,
+      "step": 817
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.330666710534556e-05,
+      "loss": 1.0624,
+      "step": 818
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.329047030108999e-05,
+      "loss": 1.2784,
+      "step": 819
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.327425533295724e-05,
+      "loss": 1.1581,
+      "step": 820
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.325802220775077e-05,
+      "loss": 1.0811,
+      "step": 821
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.324177093228174e-05,
+      "loss": 1.0437,
+      "step": 822
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.322550151336887e-05,
+      "loss": 1.2757,
+      "step": 823
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.32092139578385e-05,
+      "loss": 1.2486,
+      "step": 824
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.31929082725246e-05,
+      "loss": 1.0548,
+      "step": 825
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.317658446426871e-05,
+      "loss": 0.9905,
+      "step": 826
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 9.316024253992e-05,
+      "loss": 1.0326,
+      "step": 827
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 9.314388250633526e-05,
+      "loss": 0.8985,
+      "step": 828
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 9.312750437037885e-05,
+      "loss": 0.9206,
+      "step": 829
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 9.31111081389227e-05,
+      "loss": 1.2454,
+      "step": 830
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 9.309469381884641e-05,
+      "loss": 1.1604,
+      "step": 831
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 9.30782614170371e-05,
+      "loss": 0.9268,
+      "step": 832
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 9.306181094038949e-05,
+      "loss": 1.1387,
+      "step": 833
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 9.304534239580591e-05,
+      "loss": 0.9817,
+      "step": 834
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 9.302885579019627e-05,
+      "loss": 0.9782,
+      "step": 835
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 9.301235113047802e-05,
+      "loss": 0.8383,
+      "step": 836
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 9.29958284235762e-05,
+      "loss": 1.1395,
+      "step": 837
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 9.297928767642346e-05,
+      "loss": 1.412,
+      "step": 838
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 9.296272889595996e-05,
+      "loss": 0.9339,
+      "step": 839
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 9.294615208913348e-05,
+      "loss": 1.2804,
+      "step": 840
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 9.292955726289935e-05,
+      "loss": 1.1075,
+      "step": 841
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 9.291294442422042e-05,
+      "loss": 1.0823,
+      "step": 842
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 9.289631358006715e-05,
+      "loss": 1.0634,
+      "step": 843
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 9.287966473741751e-05,
+      "loss": 1.0057,
+      "step": 844
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 9.286299790325708e-05,
+      "loss": 1.0421,
+      "step": 845
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 9.284631308457892e-05,
+      "loss": 1.0428,
+      "step": 846
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 9.282961028838369e-05,
+      "loss": 1.0293,
+      "step": 847
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 9.281288952167956e-05,
+      "loss": 0.8872,
+      "step": 848
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 9.279615079148227e-05,
+      "loss": 1.042,
+      "step": 849
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 9.277939410481507e-05,
+      "loss": 1.0147,
+      "step": 850
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 9.276261946870874e-05,
+      "loss": 0.9357,
+      "step": 851
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 9.274582689020164e-05,
+      "loss": 1.0062,
+      "step": 852
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 9.27290163763396e-05,
+      "loss": 0.9703,
+      "step": 853
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 9.271218793417601e-05,
+      "loss": 1.0753,
+      "step": 854
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 9.269534157077177e-05,
+      "loss": 1.0565,
+      "step": 855
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 9.267847729319528e-05,
+      "loss": 1.1252,
+      "step": 856
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 9.26615951085225e-05,
+      "loss": 0.9188,
+      "step": 857
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 9.264469502383688e-05,
+      "loss": 0.9531,
+      "step": 858
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.262777704622938e-05,
+      "loss": 1.1892,
+      "step": 859
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.261084118279847e-05,
+      "loss": 1.0052,
+      "step": 860
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.259388744065011e-05,
+      "loss": 1.0061,
+      "step": 861
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.25769158268978e-05,
+      "loss": 1.0272,
+      "step": 862
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.255992634866248e-05,
+      "loss": 1.0298,
+      "step": 863
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.254291901307266e-05,
+      "loss": 1.0334,
+      "step": 864
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.252589382726426e-05,
+      "loss": 1.0407,
+      "step": 865
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.250885079838078e-05,
+      "loss": 1.1145,
+      "step": 866
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.249178993357311e-05,
+      "loss": 0.9961,
+      "step": 867
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.24747112399997e-05,
+      "loss": 0.9202,
+      "step": 868
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.245761472482645e-05,
+      "loss": 1.0506,
+      "step": 869
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.244050039522672e-05,
+      "loss": 1.306,
+      "step": 870
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.242336825838141e-05,
+      "loss": 1.0618,
+      "step": 871
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.24062183214788e-05,
+      "loss": 1.2713,
+      "step": 872
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.23890505917147e-05,
+      "loss": 1.2182,
+      "step": 873
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.237186507629237e-05,
+      "loss": 1.1474,
+      "step": 874
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.235466178242254e-05,
+      "loss": 1.0809,
+      "step": 875
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.23374407173234e-05,
+      "loss": 0.9283,
+      "step": 876
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.232020188822055e-05,
+      "loss": 1.2767,
+      "step": 877
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.230294530234713e-05,
+      "loss": 1.0073,
+      "step": 878
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.228567096694366e-05,
+      "loss": 0.9024,
+      "step": 879
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.226837888925813e-05,
+      "loss": 0.7884,
+      "step": 880
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.225106907654599e-05,
+      "loss": 1.0218,
+      "step": 881
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.223374153607009e-05,
+      "loss": 1.1999,
+      "step": 882
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.221639627510076e-05,
+      "loss": 1.0681,
+      "step": 883
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.219903330091575e-05,
+      "loss": 0.9637,
+      "step": 884
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.218165262080023e-05,
+      "loss": 1.0658,
+      "step": 885
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.216425424204682e-05,
+      "loss": 0.997,
+      "step": 886
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.214683817195557e-05,
+      "loss": 1.133,
+      "step": 887
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.212940441783391e-05,
+      "loss": 1.0083,
+      "step": 888
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.211195298699674e-05,
+      "loss": 1.0298,
+      "step": 889
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.209448388676635e-05,
+      "loss": 0.9923,
+      "step": 890
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.207699712447246e-05,
+      "loss": 1.0881,
+      "step": 891
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.205949270745217e-05,
+      "loss": 1.2925,
+      "step": 892
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.204197064305001e-05,
+      "loss": 1.2008,
+      "step": 893
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.202443093861795e-05,
+      "loss": 1.0326,
+      "step": 894
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.200687360151528e-05,
+      "loss": 0.9873,
+      "step": 895
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.198929863910874e-05,
+      "loss": 0.9965,
+      "step": 896
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.197170605877247e-05,
+      "loss": 0.8583,
+      "step": 897
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.195409586788799e-05,
+      "loss": 0.8444,
+      "step": 898
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.19364680738442e-05,
+      "loss": 1.2214,
+      "step": 899
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.191882268403743e-05,
+      "loss": 1.0057,
+      "step": 900
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.190115970587129e-05,
+      "loss": 1.1308,
+      "step": 901
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.188347914675689e-05,
+      "loss": 1.1619,
+      "step": 902
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.186578101411265e-05,
+      "loss": 0.9622,
+      "step": 903
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.184806531536437e-05,
+      "loss": 0.7437,
+      "step": 904
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.183033205794525e-05,
+      "loss": 0.9469,
+      "step": 905
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.181258124929581e-05,
+      "loss": 1.1917,
+      "step": 906
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.179481289686398e-05,
+      "loss": 1.2573,
+      "step": 907
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.177702700810501e-05,
+      "loss": 1.0097,
+      "step": 908
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.175922359048155e-05,
+      "loss": 1.1992,
+      "step": 909
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.174140265146356e-05,
+      "loss": 0.9729,
+      "step": 910
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.17235641985284e-05,
+      "loss": 1.1661,
+      "step": 911
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.170570823916074e-05,
+      "loss": 0.9286,
+      "step": 912
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.168783478085262e-05,
+      "loss": 0.9889,
+      "step": 913
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.166994383110338e-05,
+      "loss": 0.8308,
+      "step": 914
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.165203539741976e-05,
+      "loss": 1.108,
+      "step": 915
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.16341094873158e-05,
+      "loss": 1.0764,
+      "step": 916
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.161616610831287e-05,
+      "loss": 0.7979,
+      "step": 917
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.159820526793969e-05,
+      "loss": 1.1523,
+      "step": 918
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.15802269737323e-05,
+      "loss": 1.1686,
+      "step": 919
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.156223123323405e-05,
+      "loss": 0.7966,
+      "step": 920
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.15442180539956e-05,
+      "loss": 0.863,
+      "step": 921
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.152618744357498e-05,
+      "loss": 0.8154,
+      "step": 922
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.150813940953747e-05,
+      "loss": 0.9996,
+      "step": 923
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 9.149007395945569e-05,
+      "loss": 1.1306,
+      "step": 924
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 9.147199110090959e-05,
+      "loss": 1.1132,
+      "step": 925
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 9.145389084148639e-05,
+      "loss": 0.9914,
+      "step": 926
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 9.143577318878063e-05,
+      "loss": 1.1176,
+      "step": 927
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 9.141763815039412e-05,
+      "loss": 1.0175,
+      "step": 928
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 9.139948573393602e-05,
+      "loss": 1.2434,
+      "step": 929
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 9.13813159470227e-05,
+      "loss": 0.9241,
+      "step": 930
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 9.136312879727791e-05,
+      "loss": 0.9755,
+      "step": 931
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 9.134492429233261e-05,
+      "loss": 1.078,
+      "step": 932
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 9.132670243982508e-05,
+      "loss": 1.0478,
+      "step": 933
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 9.130846324740088e-05,
+      "loss": 1.0523,
+      "step": 934
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 9.129020672271283e-05,
+      "loss": 1.1237,
+      "step": 935
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 9.127193287342102e-05,
+      "loss": 1.0925,
+      "step": 936
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 9.125364170719284e-05,
+      "loss": 0.9607,
+      "step": 937
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 9.12353332317029e-05,
+      "loss": 1.1398,
+      "step": 938
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 9.12170074546331e-05,
+      "loss": 0.9437,
+      "step": 939
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 9.119866438367263e-05,
+      "loss": 1.0044,
+      "step": 940
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 9.118030402651785e-05,
+      "loss": 1.1422,
+      "step": 941
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 9.116192639087245e-05,
+      "loss": 1.1251,
+      "step": 942
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 9.114353148444735e-05,
+      "loss": 1.0989,
+      "step": 943
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 9.112511931496071e-05,
+      "loss": 1.0736,
+      "step": 944
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 9.110668989013791e-05,
+      "loss": 1.0213,
+      "step": 945
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 9.108824321771162e-05,
+      "loss": 1.0006,
+      "step": 946
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 9.106977930542171e-05,
+      "loss": 0.8421,
+      "step": 947
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 9.10512981610153e-05,
+      "loss": 1.0094,
+      "step": 948
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 9.103279979224675e-05,
+      "loss": 1.076,
+      "step": 949
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 9.101428420687759e-05,
+      "loss": 0.8985,
+      "step": 950
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 9.099575141267666e-05,
+      "loss": 0.899,
+      "step": 951
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 9.097720141741994e-05,
+      "loss": 0.9563,
+      "step": 952
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 9.09586342288907e-05,
+      "loss": 1.1305,
+      "step": 953
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 9.094004985487935e-05,
+      "loss": 1.0931,
+      "step": 954
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 9.092144830318358e-05,
+      "loss": 1.3143,
+      "step": 955
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 9.090282958160823e-05,
+      "loss": 1.1012,
+      "step": 956
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 9.088419369796539e-05,
+      "loss": 0.9937,
+      "step": 957
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 9.08655406600743e-05,
+      "loss": 0.9995,
+      "step": 958
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 9.084687047576148e-05,
+      "loss": 0.9602,
+      "step": 959
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 9.082818315286055e-05,
+      "loss": 1.0777,
+      "step": 960
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 9.080947869921238e-05,
+      "loss": 1.0665,
+      "step": 961
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 9.079075712266501e-05,
+      "loss": 0.9808,
+      "step": 962
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 9.077201843107365e-05,
+      "loss": 1.1814,
+      "step": 963
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 9.075326263230073e-05,
+      "loss": 1.1952,
+      "step": 964
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 9.073448973421582e-05,
+      "loss": 0.928,
+      "step": 965
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 9.071569974469569e-05,
+      "loss": 1.0121,
+      "step": 966
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 9.069689267162424e-05,
+      "loss": 1.0551,
+      "step": 967
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 9.067806852289263e-05,
+      "loss": 0.8802,
+      "step": 968
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 9.065922730639906e-05,
+      "loss": 0.992,
+      "step": 969
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 9.0640369030049e-05,
+      "loss": 1.016,
+      "step": 970
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 9.062149370175501e-05,
+      "loss": 1.0736,
+      "step": 971
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 9.060260132943682e-05,
+      "loss": 1.3418,
+      "step": 972
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 9.058369192102133e-05,
+      "loss": 1.0168,
+      "step": 973
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 9.056476548444258e-05,
+      "loss": 1.1812,
+      "step": 974
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 9.054582202764175e-05,
+      "loss": 0.8268,
+      "step": 975
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 9.052686155856716e-05,
+      "loss": 0.9526,
+      "step": 976
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 9.050788408517426e-05,
+      "loss": 0.9626,
+      "step": 977
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 9.048888961542566e-05,
+      "loss": 0.9268,
+      "step": 978
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 9.046987815729108e-05,
+      "loss": 1.1682,
+      "step": 979
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 9.045084971874738e-05,
+      "loss": 0.8847,
+      "step": 980
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 9.043180430777853e-05,
+      "loss": 1.1201,
+      "step": 981
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 9.041274193237564e-05,
+      "loss": 0.9434,
+      "step": 982
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 9.039366260053692e-05,
+      "loss": 1.3278,
+      "step": 983
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 9.037456632026773e-05,
+      "loss": 0.9788,
+      "step": 984
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 9.035545309958046e-05,
+      "loss": 0.9135,
+      "step": 985
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 9.033632294649473e-05,
+      "loss": 1.1537,
+      "step": 986
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 9.031717586903714e-05,
+      "loss": 0.9037,
+      "step": 987
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 9.029801187524147e-05,
+      "loss": 1.3395,
+      "step": 988
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 9.027883097314857e-05,
+      "loss": 1.2508,
+      "step": 989
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 9.025963317080641e-05,
+      "loss": 0.8922,
+      "step": 990
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 9.024041847627002e-05,
+      "loss": 1.1211,
+      "step": 991
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 9.022118689760152e-05,
+      "loss": 1.1281,
+      "step": 992
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 9.020193844287012e-05,
+      "loss": 0.9806,
+      "step": 993
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 9.018267312015214e-05,
+      "loss": 0.8213,
+      "step": 994
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 9.016339093753093e-05,
+      "loss": 0.8467,
+      "step": 995
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 9.014409190309695e-05,
+      "loss": 1.125,
+      "step": 996
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 9.012477602494769e-05,
+      "loss": 1.0954,
+      "step": 997
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 9.010544331118776e-05,
+      "loss": 1.1711,
+      "step": 998
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 9.00860937699288e-05,
+      "loss": 1.1069,
+      "step": 999
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 9.006672740928952e-05,
+      "loss": 1.3942,
+      "step": 1000
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 9.004734423739566e-05,
+      "loss": 1.1349,
+      "step": 1001
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 9.002794426238008e-05,
+      "loss": 1.437,
+      "step": 1002
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 9.000852749238261e-05,
+      "loss": 1.056,
+      "step": 1003
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.998909393555021e-05,
+      "loss": 1.1027,
+      "step": 1004
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.99696436000368e-05,
+      "loss": 1.1443,
+      "step": 1005
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.99501764940034e-05,
+      "loss": 1.0346,
+      "step": 1006
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.993069262561803e-05,
+      "loss": 0.9881,
+      "step": 1007
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.99111920030558e-05,
+      "loss": 1.1121,
+      "step": 1008
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.989167463449875e-05,
+      "loss": 1.0815,
+      "step": 1009
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.987214052813604e-05,
+      "loss": 1.3458,
+      "step": 1010
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.985258969216383e-05,
+      "loss": 0.9095,
+      "step": 1011
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.983302213478523e-05,
+      "loss": 1.1317,
+      "step": 1012
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.981343786421051e-05,
+      "loss": 0.9822,
+      "step": 1013
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.979383688865679e-05,
+      "loss": 1.2522,
+      "step": 1014
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.977421921634832e-05,
+      "loss": 1.0278,
+      "step": 1015
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.975458485551629e-05,
+      "loss": 0.8931,
+      "step": 1016
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.973493381439893e-05,
+      "loss": 0.9377,
+      "step": 1017
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.971526610124141e-05,
+      "loss": 1.0222,
+      "step": 1018
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.9695581724296e-05,
+      "loss": 1.125,
+      "step": 1019
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.967588069182185e-05,
+      "loss": 1.2003,
+      "step": 1020
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.965616301208516e-05,
+      "loss": 1.2909,
+      "step": 1021
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.963642869335913e-05,
+      "loss": 0.9231,
+      "step": 1022
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.961667774392387e-05,
+      "loss": 1.0822,
+      "step": 1023
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.959691017206653e-05,
+      "loss": 1.1514,
+      "step": 1024
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.957712598608123e-05,
+      "loss": 0.9995,
+      "step": 1025
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.955732519426902e-05,
+      "loss": 0.9597,
+      "step": 1026
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.953750780493797e-05,
+      "loss": 1.1939,
+      "step": 1027
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.951767382640307e-05,
+      "loss": 1.1509,
+      "step": 1028
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.949782326698631e-05,
+      "loss": 0.9629,
+      "step": 1029
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.947795613501658e-05,
+      "loss": 0.9108,
+      "step": 1030
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.945807243882979e-05,
+      "loss": 0.7633,
+      "step": 1031
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.943817218676877e-05,
+      "loss": 1.1782,
+      "step": 1032
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.941825538718327e-05,
+      "loss": 1.2349,
+      "step": 1033
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.939832204843004e-05,
+      "loss": 1.0182,
+      "step": 1034
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.937837217887273e-05,
+      "loss": 0.9708,
+      "step": 1035
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.935840578688191e-05,
+      "loss": 1.0374,
+      "step": 1036
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.933842288083514e-05,
+      "loss": 1.0975,
+      "step": 1037
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.931842346911687e-05,
+      "loss": 1.0734,
+      "step": 1038
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.929840756011846e-05,
+      "loss": 1.012,
+      "step": 1039
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.927837516223824e-05,
+      "loss": 0.8954,
+      "step": 1040
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.925832628388142e-05,
+      "loss": 0.9467,
+      "step": 1041
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.923826093346013e-05,
+      "loss": 1.044,
+      "step": 1042
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.921817911939343e-05,
+      "loss": 0.871,
+      "step": 1043
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.919808085010726e-05,
+      "loss": 1.2205,
+      "step": 1044
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.917796613403451e-05,
+      "loss": 1.0996,
+      "step": 1045
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.915783497961492e-05,
+      "loss": 0.8604,
+      "step": 1046
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.913768739529513e-05,
+      "loss": 1.0283,
+      "step": 1047
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.911752338952875e-05,
+      "loss": 1.1051,
+      "step": 1048
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.909734297077616e-05,
+      "loss": 1.0572,
+      "step": 1049
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.907714614750473e-05,
+      "loss": 1.119,
+      "step": 1050
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.905693292818864e-05,
+      "loss": 1.1195,
+      "step": 1051
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.903670332130901e-05,
+      "loss": 0.7755,
+      "step": 1052
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.90164573353538e-05,
+      "loss": 1.172,
+      "step": 1053
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.899619497881784e-05,
+      "loss": 0.9573,
+      "step": 1054
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.897591626020284e-05,
+      "loss": 0.7479,
+      "step": 1055
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.895562118801738e-05,
+      "loss": 1.2475,
+      "step": 1056
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.893530977077689e-05,
+      "loss": 1.1387,
+      "step": 1057
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.891498201700368e-05,
+      "loss": 1.1615,
+      "step": 1058
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.889463793522687e-05,
+      "loss": 1.1727,
+      "step": 1059
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.887427753398248e-05,
+      "loss": 1.0991,
+      "step": 1060
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.885390082181333e-05,
+      "loss": 1.2653,
+      "step": 1061
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.883350780726916e-05,
+      "loss": 0.9243,
+      "step": 1062
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.881309849890644e-05,
+      "loss": 0.942,
+      "step": 1063
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.879267290528859e-05,
+      "loss": 1.0732,
+      "step": 1064
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.877223103498575e-05,
+      "loss": 1.0764,
+      "step": 1065
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.875177289657502e-05,
+      "loss": 1.0674,
+      "step": 1066
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.873129849864021e-05,
+      "loss": 1.0715,
+      "step": 1067
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.871080784977199e-05,
+      "loss": 0.9097,
+      "step": 1068
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.86903009585679e-05,
+      "loss": 1.1325,
+      "step": 1069
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.86697778336322e-05,
+      "loss": 0.8945,
+      "step": 1070
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.864923848357605e-05,
+      "loss": 1.163,
+      "step": 1071
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.862868291701735e-05,
+      "loss": 1.0493,
+      "step": 1072
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.860811114258085e-05,
+      "loss": 1.1481,
+      "step": 1073
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.85875231688981e-05,
+      "loss": 1.2442,
+      "step": 1074
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.856691900460739e-05,
+      "loss": 1.0598,
+      "step": 1075
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.854629865835387e-05,
+      "loss": 1.0437,
+      "step": 1076
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.852566213878947e-05,
+      "loss": 1.2063,
+      "step": 1077
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.850500945457286e-05,
+      "loss": 1.164,
+      "step": 1078
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.848434061436954e-05,
+      "loss": 0.9412,
+      "step": 1079
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.846365562685177e-05,
+      "loss": 0.9875,
+      "step": 1080
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.844295450069858e-05,
+      "loss": 1.1109,
+      "step": 1081
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.842223724459577e-05,
+      "loss": 1.0125,
+      "step": 1082
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.840150386723595e-05,
+      "loss": 1.0115,
+      "step": 1083
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.838075437731843e-05,
+      "loss": 1.0898,
+      "step": 1084
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.835998878354931e-05,
+      "loss": 1.2442,
+      "step": 1085
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 8.833920709464145e-05,
+      "loss": 0.7628,
+      "step": 1086
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 8.831840931931448e-05,
+      "loss": 0.9859,
+      "step": 1087
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 8.829759546629475e-05,
+      "loss": 0.8464,
+      "step": 1088
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 8.827676554431534e-05,
+      "loss": 1.0601,
+      "step": 1089
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 8.825591956211614e-05,
+      "loss": 1.3557,
+      "step": 1090
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 8.823505752844372e-05,
+      "loss": 1.1644,
+      "step": 1091
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 8.82141794520514e-05,
+      "loss": 0.9357,
+      "step": 1092
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 8.819328534169922e-05,
+      "loss": 0.962,
+      "step": 1093
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 8.817237520615398e-05,
+      "loss": 0.9549,
+      "step": 1094
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 8.815144905418917e-05,
+      "loss": 0.969,
+      "step": 1095
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 8.813050689458502e-05,
+      "loss": 0.9641,
+      "step": 1096
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 8.810954873612848e-05,
+      "loss": 0.9999,
+      "step": 1097
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 8.80885745876132e-05,
+      "loss": 1.1939,
+      "step": 1098
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 8.806758445783953e-05,
+      "loss": 1.2516,
+      "step": 1099
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 8.804657835561456e-05,
+      "loss": 1.1438,
+      "step": 1100
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 8.802555628975204e-05,
+      "loss": 1.0197,
+      "step": 1101
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 8.800451826907245e-05,
+      "loss": 0.9702,
+      "step": 1102
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 8.798346430240296e-05,
+      "loss": 1.2586,
+      "step": 1103
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 8.79623943985774e-05,
+      "loss": 1.2549,
+      "step": 1104
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 8.794130856643634e-05,
+      "loss": 0.8286,
+      "step": 1105
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 8.792020681482698e-05,
+      "loss": 0.9827,
+      "step": 1106
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 8.789908915260322e-05,
+      "loss": 0.971,
+      "step": 1107
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 8.787795558862566e-05,
+      "loss": 1.1003,
+      "step": 1108
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 8.785680613176153e-05,
+      "loss": 1.0052,
+      "step": 1109
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 8.783564079088477e-05,
+      "loss": 0.9428,
+      "step": 1110
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 8.781445957487595e-05,
+      "loss": 1.0577,
+      "step": 1111
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 8.77932624926223e-05,
+      "loss": 0.9651,
+      "step": 1112
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 8.777204955301776e-05,
+      "loss": 1.088,
+      "step": 1113
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 8.775082076496286e-05,
+      "loss": 1.132,
+      "step": 1114
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 8.772957613736483e-05,
+      "loss": 1.1621,
+      "step": 1115
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 8.770831567913746e-05,
+      "loss": 1.0171,
+      "step": 1116
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 8.768703939920132e-05,
+      "loss": 1.149,
+      "step": 1117
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 8.76657473064835e-05,
+      "loss": 1.1375,
+      "step": 1118
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 8.764443940991776e-05,
+      "loss": 1.0254,
+      "step": 1119
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 8.762311571844451e-05,
+      "loss": 0.9781,
+      "step": 1120
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 8.760177624101079e-05,
+      "loss": 1.1408,
+      "step": 1121
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 8.758042098657023e-05,
+      "loss": 1.3126,
+      "step": 1122
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 8.755904996408309e-05,
+      "loss": 1.1723,
+      "step": 1123
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 8.753766318251628e-05,
+      "loss": 1.037,
+      "step": 1124
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 8.751626065084328e-05,
+      "loss": 1.0226,
+      "step": 1125
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 8.749484237804419e-05,
+      "loss": 0.8685,
+      "step": 1126
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 8.747340837310574e-05,
+      "loss": 0.9253,
+      "step": 1127
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 8.745195864502122e-05,
+      "loss": 0.9289,
+      "step": 1128
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 8.743049320279053e-05,
+      "loss": 1.1661,
+      "step": 1129
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 8.740901205542021e-05,
+      "loss": 1.0005,
+      "step": 1130
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 8.73875152119233e-05,
+      "loss": 0.8808,
+      "step": 1131
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 8.736600268131953e-05,
+      "loss": 1.1384,
+      "step": 1132
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 8.73444744726351e-05,
+      "loss": 0.8967,
+      "step": 1133
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 8.73229305949029e-05,
+      "loss": 1.1564,
+      "step": 1134
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 8.73013710571623e-05,
+      "loss": 1.1833,
+      "step": 1135
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 8.727979586845931e-05,
+      "loss": 0.983,
+      "step": 1136
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 8.725820503784648e-05,
+      "loss": 1.0386,
+      "step": 1137
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 8.723659857438289e-05,
+      "loss": 1.2687,
+      "step": 1138
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 8.721497648713423e-05,
+      "loss": 1.1489,
+      "step": 1139
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 8.719333878517273e-05,
+      "loss": 1.2983,
+      "step": 1140
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 8.717168547757716e-05,
+      "loss": 1.2877,
+      "step": 1141
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 8.715001657343284e-05,
+      "loss": 1.026,
+      "step": 1142
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 8.712833208183166e-05,
+      "loss": 0.9495,
+      "step": 1143
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 8.710663201187203e-05,
+      "loss": 1.0973,
+      "step": 1144
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 8.708491637265887e-05,
+      "loss": 1.2346,
+      "step": 1145
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 8.706318517330367e-05,
+      "loss": 0.8069,
+      "step": 1146
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 8.704143842292445e-05,
+      "loss": 1.0632,
+      "step": 1147
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 8.701967613064575e-05,
+      "loss": 1.1616,
+      "step": 1148
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 8.699789830559857e-05,
+      "loss": 0.9137,
+      "step": 1149
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 8.697610495692054e-05,
+      "loss": 0.8736,
+      "step": 1150
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 8.695429609375571e-05,
+      "loss": 1.159,
+      "step": 1151
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 8.693247172525471e-05,
+      "loss": 1.1751,
+      "step": 1152
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 8.69106318605746e-05,
+      "loss": 1.0093,
+      "step": 1153
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 8.6888776508879e-05,
+      "loss": 1.3955,
+      "step": 1154
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 8.686690567933802e-05,
+      "loss": 1.1304,
+      "step": 1155
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 8.684501938112821e-05,
+      "loss": 1.1403,
+      "step": 1156
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 8.682311762343271e-05,
+      "loss": 1.1913,
+      "step": 1157
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 8.680120041544106e-05,
+      "loss": 1.0707,
+      "step": 1158
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 8.677926776634933e-05,
+      "loss": 1.0614,
+      "step": 1159
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 8.675731968536002e-05,
+      "loss": 1.3528,
+      "step": 1160
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 8.673535618168219e-05,
+      "loss": 0.9793,
+      "step": 1161
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 8.671337726453124e-05,
+      "loss": 1.0707,
+      "step": 1162
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 8.669138294312919e-05,
+      "loss": 1.0962,
+      "step": 1163
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 8.666937322670442e-05,
+      "loss": 0.9027,
+      "step": 1164
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 8.66473481244918e-05,
+      "loss": 0.9815,
+      "step": 1165
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 8.662530764573265e-05,
+      "loss": 0.7915,
+      "step": 1166
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 8.660325179967475e-05,
+      "loss": 0.8815,
+      "step": 1167
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 8.658118059557231e-05,
+      "loss": 0.8807,
+      "step": 1168
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 8.655909404268603e-05,
+      "loss": 0.887,
+      "step": 1169
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 8.653699215028296e-05,
+      "loss": 0.92,
+      "step": 1170
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 8.651487492763672e-05,
+      "loss": 1.1966,
+      "step": 1171
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 8.649274238402723e-05,
+      "loss": 1.0268,
+      "step": 1172
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 8.647059452874091e-05,
+      "loss": 0.9146,
+      "step": 1173
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 8.644843137107059e-05,
+      "loss": 1.2465,
+      "step": 1174
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 8.64262529203155e-05,
+      "loss": 0.9218,
+      "step": 1175
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 8.640405918578134e-05,
+      "loss": 1.2877,
+      "step": 1176
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 8.638185017678017e-05,
+      "loss": 0.7926,
+      "step": 1177
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 8.635962590263047e-05,
+      "loss": 1.3074,
+      "step": 1178
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 8.633738637265716e-05,
+      "loss": 1.1851,
+      "step": 1179
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 8.631513159619151e-05,
+      "loss": 0.9712,
+      "step": 1180
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 8.62928615825712e-05,
+      "loss": 1.0755,
+      "step": 1181
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 8.627057634114035e-05,
+      "loss": 1.1217,
+      "step": 1182
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 8.624827588124941e-05,
+      "loss": 1.1524,
+      "step": 1183
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 8.622596021225524e-05,
+      "loss": 0.8748,
+      "step": 1184
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 8.620362934352109e-05,
+      "loss": 0.8761,
+      "step": 1185
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 8.618128328441655e-05,
+      "loss": 0.9481,
+      "step": 1186
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 8.615892204431766e-05,
+      "loss": 1.0855,
+      "step": 1187
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 8.613654563260674e-05,
+      "loss": 0.8907,
+      "step": 1188
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 8.611415405867251e-05,
+      "loss": 0.9524,
+      "step": 1189
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 8.609174733191011e-05,
+      "loss": 1.0817,
+      "step": 1190
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 8.606932546172093e-05,
+      "loss": 1.0578,
+      "step": 1191
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 8.604688845751282e-05,
+      "loss": 0.9919,
+      "step": 1192
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 8.602443632869988e-05,
+      "loss": 1.2283,
+      "step": 1193
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 8.600196908470265e-05,
+      "loss": 1.2266,
+      "step": 1194
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 8.597948673494794e-05,
+      "loss": 1.1176,
+      "step": 1195
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 8.595698928886894e-05,
+      "loss": 1.2027,
+      "step": 1196
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 8.593447675590516e-05,
+      "loss": 0.9252,
+      "step": 1197
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 8.591194914550243e-05,
+      "loss": 1.0251,
+      "step": 1198
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 8.588940646711291e-05,
+      "loss": 1.0495,
+      "step": 1199
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 8.586684873019513e-05,
+      "loss": 1.0459,
+      "step": 1200
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 8.584427594421386e-05,
+      "loss": 1.0862,
+      "step": 1201
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 8.582168811864022e-05,
+      "loss": 1.0942,
+      "step": 1202
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 8.579908526295166e-05,
+      "loss": 1.1189,
+      "step": 1203
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 8.577646738663192e-05,
+      "loss": 0.8528,
+      "step": 1204
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 8.575383449917102e-05,
+      "loss": 1.3154,
+      "step": 1205
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 8.573118661006534e-05,
+      "loss": 1.0356,
+      "step": 1206
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 8.570852372881746e-05,
+      "loss": 0.9972,
+      "step": 1207
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 8.568584586493634e-05,
+      "loss": 1.0858,
+      "step": 1208
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 8.566315302793717e-05,
+      "loss": 1.1971,
+      "step": 1209
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 8.564044522734147e-05,
+      "loss": 1.0698,
+      "step": 1210
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 8.561772247267698e-05,
+      "loss": 0.99,
+      "step": 1211
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 8.559498477347776e-05,
+      "loss": 0.9567,
+      "step": 1212
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 8.557223213928414e-05,
+      "loss": 1.1722,
+      "step": 1213
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 8.554946457964268e-05,
+      "loss": 0.9196,
+      "step": 1214
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 8.552668210410624e-05,
+      "loss": 0.8087,
+      "step": 1215
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 8.550388472223391e-05,
+      "loss": 1.1797,
+      "step": 1216
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 8.548107244359105e-05,
+      "loss": 1.0599,
+      "step": 1217
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 8.545824527774928e-05,
+      "loss": 1.014,
+      "step": 1218
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 8.543540323428645e-05,
+      "loss": 1.1362,
+      "step": 1219
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 8.541254632278665e-05,
+      "loss": 0.8568,
+      "step": 1220
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 8.538967455284024e-05,
+      "loss": 1.0295,
+      "step": 1221
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 8.536678793404376e-05,
+      "loss": 1.0809,
+      "step": 1222
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 8.534388647600003e-05,
+      "loss": 1.297,
+      "step": 1223
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 8.532097018831805e-05,
+      "loss": 1.0955,
+      "step": 1224
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 8.529803908061309e-05,
+      "loss": 1.1114,
+      "step": 1225
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 8.527509316250663e-05,
+      "loss": 1.1213,
+      "step": 1226
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 8.525213244362631e-05,
+      "loss": 1.0566,
+      "step": 1227
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 8.522915693360606e-05,
+      "loss": 1.3259,
+      "step": 1228
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 8.520616664208596e-05,
+      "loss": 1.0596,
+      "step": 1229
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 8.518316157871231e-05,
+      "loss": 0.7937,
+      "step": 1230
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 8.516014175313762e-05,
+      "loss": 1.0566,
+      "step": 1231
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 8.513710717502056e-05,
+      "loss": 0.9662,
+      "step": 1232
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 8.511405785402603e-05,
+      "loss": 1.009,
+      "step": 1233
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 8.509099379982508e-05,
+      "loss": 1.0112,
+      "step": 1234
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 8.506791502209496e-05,
+      "loss": 1.3701,
+      "step": 1235
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 8.504482153051912e-05,
+      "loss": 1.0245,
+      "step": 1236
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 8.502171333478712e-05,
+      "loss": 1.1457,
+      "step": 1237
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 8.499859044459478e-05,
+      "loss": 1.162,
+      "step": 1238
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 8.497545286964399e-05,
+      "loss": 1.1847,
+      "step": 1239
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 8.495230061964288e-05,
+      "loss": 0.9128,
+      "step": 1240
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 8.492913370430568e-05,
+      "loss": 1.1877,
+      "step": 1241
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 8.49059521333528e-05,
+      "loss": 0.9698,
+      "step": 1242
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 8.488275591651081e-05,
+      "loss": 1.2002,
+      "step": 1243
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 8.485954506351241e-05,
+      "loss": 1.1345,
+      "step": 1244
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 8.483631958409644e-05,
+      "loss": 1.0434,
+      "step": 1245
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 8.481307948800787e-05,
+      "loss": 1.1092,
+      "step": 1246
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 8.478982478499782e-05,
+      "loss": 0.9825,
+      "step": 1247
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 8.476655548482353e-05,
+      "loss": 1.1214,
+      "step": 1248
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 8.474327159724837e-05,
+      "loss": 0.9335,
+      "step": 1249
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 8.471997313204181e-05,
+      "loss": 1.0966,
+      "step": 1250
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 8.469666009897948e-05,
+      "loss": 1.066,
+      "step": 1251
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 8.467333250784308e-05,
+      "loss": 0.945,
+      "step": 1252
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 8.464999036842045e-05,
+      "loss": 0.9315,
+      "step": 1253
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 8.462663369050549e-05,
+      "loss": 1.162,
+      "step": 1254
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 8.460326248389825e-05,
+      "loss": 1.0165,
+      "step": 1255
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 8.457987675840484e-05,
+      "loss": 1.1504,
+      "step": 1256
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 8.455647652383749e-05,
+      "loss": 0.9454,
+      "step": 1257
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 8.45330617900145e-05,
+      "loss": 0.9894,
+      "step": 1258
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 8.450963256676026e-05,
+      "loss": 0.8718,
+      "step": 1259
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 8.448618886390522e-05,
+      "loss": 0.9232,
+      "step": 1260
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 8.446273069128593e-05,
+      "loss": 1.1037,
+      "step": 1261
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 8.443925805874502e-05,
+      "loss": 1.0112,
+      "step": 1262
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 8.441577097613113e-05,
+      "loss": 1.1343,
+      "step": 1263
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 8.439226945329907e-05,
+      "loss": 1.1371,
+      "step": 1264
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 8.436875350010957e-05,
+      "loss": 1.1122,
+      "step": 1265
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 8.434522312642954e-05,
+      "loss": 1.0191,
+      "step": 1266
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 8.432167834213188e-05,
+      "loss": 1.1135,
+      "step": 1267
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 8.42981191570955e-05,
+      "loss": 0.8721,
+      "step": 1268
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 8.427454558120544e-05,
+      "loss": 1.0739,
+      "step": 1269
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 8.425095762435273e-05,
+      "loss": 0.9153,
+      "step": 1270
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 8.422735529643444e-05,
+      "loss": 1.0597,
+      "step": 1271
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 8.420373860735366e-05,
+      "loss": 1.2016,
+      "step": 1272
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 8.418010756701951e-05,
+      "loss": 1.1067,
+      "step": 1273
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 8.415646218534714e-05,
+      "loss": 1.0125,
+      "step": 1274
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 8.413280247225768e-05,
+      "loss": 1.0904,
+      "step": 1275
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 8.410912843767837e-05,
+      "loss": 1.0203,
+      "step": 1276
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 8.408544009154238e-05,
+      "loss": 1.0999,
+      "step": 1277
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 8.406173744378886e-05,
+      "loss": 0.7544,
+      "step": 1278
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 8.403802050436303e-05,
+      "loss": 1.3206,
+      "step": 1279
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 8.401428928321607e-05,
+      "loss": 1.0393,
+      "step": 1280
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 8.399054379030517e-05,
+      "loss": 1.0583,
+      "step": 1281
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 8.396678403559348e-05,
+      "loss": 1.1368,
+      "step": 1282
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 8.394301002905017e-05,
+      "loss": 1.3145,
+      "step": 1283
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 8.391922178065036e-05,
+      "loss": 1.1708,
+      "step": 1284
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 8.389541930037516e-05,
+      "loss": 1.1406,
+      "step": 1285
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 8.387160259821166e-05,
+      "loss": 1.1929,
+      "step": 1286
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 8.384777168415287e-05,
+      "loss": 1.1872,
+      "step": 1287
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 8.382392656819785e-05,
+      "loss": 1.2589,
+      "step": 1288
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 8.380006726035151e-05,
+      "loss": 0.8798,
+      "step": 1289
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 8.377619377062482e-05,
+      "loss": 0.9254,
+      "step": 1290
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 8.375230610903464e-05,
+      "loss": 1.0246,
+      "step": 1291
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 8.372840428560378e-05,
+      "loss": 1.181,
+      "step": 1292
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 8.3704488310361e-05,
+      "loss": 1.1213,
+      "step": 1293
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 8.368055819334102e-05,
+      "loss": 1.0877,
+      "step": 1294
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 8.365661394458446e-05,
+      "loss": 0.9647,
+      "step": 1295
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 8.363265557413786e-05,
+      "loss": 0.9793,
+      "step": 1296
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 8.360868309205374e-05,
+      "loss": 1.0019,
+      "step": 1297
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 8.358469650839049e-05,
+      "loss": 0.9371,
+      "step": 1298
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 8.356069583321244e-05,
+      "loss": 1.0595,
+      "step": 1299
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 8.353668107658984e-05,
+      "loss": 1.0967,
+      "step": 1300
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 8.351265224859881e-05,
+      "loss": 1.0826,
+      "step": 1301
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 8.348860935932142e-05,
+      "loss": 1.1784,
+      "step": 1302
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 8.346455241884561e-05,
+      "loss": 0.9622,
+      "step": 1303
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 8.344048143726523e-05,
+      "loss": 0.7652,
+      "step": 1304
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 8.341639642468002e-05,
+      "loss": 0.9198,
+      "step": 1305
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 8.339229739119559e-05,
+      "loss": 1.0617,
+      "step": 1306
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 8.336818434692345e-05,
+      "loss": 1.0669,
+      "step": 1307
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 8.334405730198101e-05,
+      "loss": 0.9805,
+      "step": 1308
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 8.331991626649149e-05,
+      "loss": 1.0601,
+      "step": 1309
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 8.329576125058406e-05,
+      "loss": 1.2402,
+      "step": 1310
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 8.327159226439369e-05,
+      "loss": 0.9223,
+      "step": 1311
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 8.324740931806124e-05,
+      "loss": 1.0774,
+      "step": 1312
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 8.322321242173343e-05,
+      "loss": 1.0107,
+      "step": 1313
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 8.319900158556285e-05,
+      "loss": 1.0229,
+      "step": 1314
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 8.317477681970787e-05,
+      "loss": 1.1003,
+      "step": 1315
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 8.315053813433279e-05,
+      "loss": 1.0458,
+      "step": 1316
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 8.312628553960768e-05,
+      "loss": 1.023,
+      "step": 1317
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 8.310201904570852e-05,
+      "loss": 1.1252,
+      "step": 1318
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 8.307773866281705e-05,
+      "loss": 1.0791,
+      "step": 1319
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 8.305344440112089e-05,
+      "loss": 1.2862,
+      "step": 1320
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 8.302913627081341e-05,
+      "loss": 1.0649,
+      "step": 1321
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 8.300481428209391e-05,
+      "loss": 1.186,
+      "step": 1322
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 8.298047844516742e-05,
+      "loss": 1.0853,
+      "step": 1323
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 8.295612877024482e-05,
+      "loss": 1.0236,
+      "step": 1324
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 8.293176526754274e-05,
+      "loss": 1.2217,
+      "step": 1325
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 8.290738794728367e-05,
+      "loss": 0.9498,
+      "step": 1326
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 8.288299681969593e-05,
+      "loss": 0.9099,
+      "step": 1327
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 8.285859189501352e-05,
+      "loss": 1.1654,
+      "step": 1328
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 8.283417318347632e-05,
+      "loss": 0.9811,
+      "step": 1329
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 8.280974069532998e-05,
+      "loss": 0.7256,
+      "step": 1330
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 8.278529444082589e-05,
+      "loss": 0.9584,
+      "step": 1331
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 8.276083443022126e-05,
+      "loss": 1.172,
+      "step": 1332
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 8.273636067377905e-05,
+      "loss": 1.2019,
+      "step": 1333
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 8.271187318176802e-05,
+      "loss": 1.0216,
+      "step": 1334
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 8.268737196446264e-05,
+      "loss": 0.8307,
+      "step": 1335
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 8.266285703214315e-05,
+      "loss": 1.2034,
+      "step": 1336
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 8.263832839509559e-05,
+      "loss": 0.9552,
+      "step": 1337
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 8.261378606361172e-05,
+      "loss": 1.226,
+      "step": 1338
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 8.258923004798902e-05,
+      "loss": 0.9883,
+      "step": 1339
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 8.256466035853076e-05,
+      "loss": 1.209,
+      "step": 1340
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 8.254007700554593e-05,
+      "loss": 1.0463,
+      "step": 1341
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 8.251547999934925e-05,
+      "loss": 1.1088,
+      "step": 1342
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 8.249086935026114e-05,
+      "loss": 0.9876,
+      "step": 1343
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 8.246624506860779e-05,
+      "loss": 1.1388,
+      "step": 1344
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 8.24416071647211e-05,
+      "loss": 0.7756,
+      "step": 1345
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 8.241695564893864e-05,
+      "loss": 1.148,
+      "step": 1346
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 8.239229053160377e-05,
+      "loss": 0.9824,
+      "step": 1347
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 8.23676118230655e-05,
+      "loss": 1.121,
+      "step": 1348
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 8.234291953367854e-05,
+      "loss": 1.1081,
+      "step": 1349
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 8.231821367380335e-05,
+      "loss": 1.0748,
+      "step": 1350
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 8.229349425380602e-05,
+      "loss": 1.2221,
+      "step": 1351
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 8.226876128405837e-05,
+      "loss": 1.0439,
+      "step": 1352
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 8.22440147749379e-05,
+      "loss": 0.8366,
+      "step": 1353
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 8.221925473682779e-05,
+      "loss": 1.045,
+      "step": 1354
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 8.219448118011688e-05,
+      "loss": 0.9536,
+      "step": 1355
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 8.21696941151997e-05,
+      "loss": 1.0414,
+      "step": 1356
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 8.214489355247643e-05,
+      "loss": 1.0034,
+      "step": 1357
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 8.212007950235294e-05,
+      "loss": 1.2846,
+      "step": 1358
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 8.209525197524074e-05,
+      "loss": 1.1796,
+      "step": 1359
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 8.2070410981557e-05,
+      "loss": 1.2099,
+      "step": 1360
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 8.204555653172455e-05,
+      "loss": 1.0266,
+      "step": 1361
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 8.202068863617183e-05,
+      "loss": 0.8796,
+      "step": 1362
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 8.199580730533297e-05,
+      "loss": 1.1352,
+      "step": 1363
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 8.197091254964768e-05,
+      "loss": 0.972,
+      "step": 1364
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 8.194600437956139e-05,
+      "loss": 1.2997,
+      "step": 1365
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 8.192108280552507e-05,
+      "loss": 0.8087,
+      "step": 1366
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 8.189614783799535e-05,
+      "loss": 1.1032,
+      "step": 1367
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 8.18711994874345e-05,
+      "loss": 1.0998,
+      "step": 1368
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 8.184623776431036e-05,
+      "loss": 0.9931,
+      "step": 1369
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 8.182126267909641e-05,
+      "loss": 0.9477,
+      "step": 1370
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 8.179627424227176e-05,
+      "loss": 1.1975,
+      "step": 1371
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 8.177127246432106e-05,
+      "loss": 1.2068,
+      "step": 1372
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 8.17462573557346e-05,
+      "loss": 0.7652,
+      "step": 1373
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 8.172122892700826e-05,
+      "loss": 1.1108,
+      "step": 1374
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 8.16961871886435e-05,
+      "loss": 1.0075,
+      "step": 1375
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 8.167113215114738e-05,
+      "loss": 1.1984,
+      "step": 1376
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 8.164606382503251e-05,
+      "loss": 0.907,
+      "step": 1377
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 8.162098222081711e-05,
+      "loss": 0.9438,
+      "step": 1378
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 8.159588734902494e-05,
+      "loss": 0.8098,
+      "step": 1379
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 8.157077922018537e-05,
+      "loss": 0.9772,
+      "step": 1380
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 8.154565784483326e-05,
+      "loss": 1.0408,
+      "step": 1381
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 8.15205232335091e-05,
+      "loss": 1.179,
+      "step": 1382
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 8.149537539675892e-05,
+      "loss": 1.022,
+      "step": 1383
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 8.147021434513426e-05,
+      "loss": 0.8428,
+      "step": 1384
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 8.144504008919222e-05,
+      "loss": 1.4412,
+      "step": 1385
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 8.14198526394955e-05,
+      "loss": 0.7293,
+      "step": 1386
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 8.139465200661222e-05,
+      "loss": 0.9062,
+      "step": 1387
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 8.136943820111615e-05,
+      "loss": 1.2521,
+      "step": 1388
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 8.134421123358653e-05,
+      "loss": 1.023,
+      "step": 1389
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 8.13189711146081e-05,
+      "loss": 0.9434,
+      "step": 1390
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 8.129371785477117e-05,
+      "loss": 1.0678,
+      "step": 1391
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 8.12684514646715e-05,
+      "loss": 1.2053,
+      "step": 1392
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 8.124317195491044e-05,
+      "loss": 0.5497,
+      "step": 1393
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 8.121787933609479e-05,
+      "loss": 1.1445,
+      "step": 1394
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 8.119257361883687e-05,
+      "loss": 1.1288,
+      "step": 1395
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 8.116725481375446e-05,
+      "loss": 1.0261,
+      "step": 1396
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 8.114192293147089e-05,
+      "loss": 1.028,
+      "step": 1397
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 8.111657798261494e-05,
+      "loss": 0.9008,
+      "step": 1398
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 8.109121997782087e-05,
+      "loss": 1.0636,
+      "step": 1399
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 8.106584892772844e-05,
+      "loss": 0.927,
+      "step": 1400
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 8.104046484298287e-05,
+      "loss": 1.1516,
+      "step": 1401
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 8.101506773423483e-05,
+      "loss": 1.2668,
+      "step": 1402
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 8.09896576121405e-05,
+      "loss": 0.79,
+      "step": 1403
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 8.096423448736149e-05,
+      "loss": 1.2587,
+      "step": 1404
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 8.093879837056487e-05,
+      "loss": 1.0948,
+      "step": 1405
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 8.091334927242315e-05,
+      "loss": 1.0886,
+      "step": 1406
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 8.088788720361431e-05,
+      "loss": 1.1069,
+      "step": 1407
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 8.086241217482177e-05,
+      "loss": 0.8249,
+      "step": 1408
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 8.083692419673437e-05,
+      "loss": 1.1786,
+      "step": 1409
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 8.081142328004637e-05,
+      "loss": 0.9741,
+      "step": 1410
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 8.078590943545752e-05,
+      "loss": 1.1439,
+      "step": 1411
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 8.076038267367292e-05,
+      "loss": 0.9882,
+      "step": 1412
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 8.073484300540316e-05,
+      "loss": 0.7896,
+      "step": 1413
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 8.070929044136419e-05,
+      "loss": 0.9602,
+      "step": 1414
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 8.068372499227738e-05,
+      "loss": 1.0937,
+      "step": 1415
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 8.065814666886954e-05,
+      "loss": 0.9566,
+      "step": 1416
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 8.063255548187286e-05,
+      "loss": 1.2158,
+      "step": 1417
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 8.06069514420249e-05,
+      "loss": 0.9732,
+      "step": 1418
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 8.058133456006866e-05,
+      "loss": 1.1306,
+      "step": 1419
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 8.055570484675251e-05,
+      "loss": 0.8709,
+      "step": 1420
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 8.05300623128302e-05,
+      "loss": 1.1023,
+      "step": 1421
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 8.050440696906085e-05,
+      "loss": 0.8987,
+      "step": 1422
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 8.047873882620898e-05,
+      "loss": 0.9138,
+      "step": 1423
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 8.045305789504444e-05,
+      "loss": 0.8525,
+      "step": 1424
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 8.042736418634251e-05,
+      "loss": 1.2346,
+      "step": 1425
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 8.040165771088377e-05,
+      "loss": 1.1735,
+      "step": 1426
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 8.037593847945419e-05,
+      "loss": 1.1262,
+      "step": 1427
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 8.035020650284506e-05,
+      "loss": 1.4404,
+      "step": 1428
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 8.032446179185305e-05,
+      "loss": 0.9879,
+      "step": 1429
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 8.029870435728018e-05,
+      "loss": 1.3297,
+      "step": 1430
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 8.027293420993377e-05,
+      "loss": 1.0218,
+      "step": 1431
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 8.02471513606265e-05,
+      "loss": 1.0686,
+      "step": 1432
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 8.022135582017639e-05,
+      "loss": 1.0148,
+      "step": 1433
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 8.019554759940675e-05,
+      "loss": 1.1309,
+      "step": 1434
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 8.016972670914624e-05,
+      "loss": 1.1264,
+      "step": 1435
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 8.014389316022881e-05,
+      "loss": 1.3504,
+      "step": 1436
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 8.011804696349377e-05,
+      "loss": 1.2314,
+      "step": 1437
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 8.009218812978567e-05,
+      "loss": 1.1179,
+      "step": 1438
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 8.006631666995441e-05,
+      "loss": 1.0494,
+      "step": 1439
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 8.004043259485519e-05,
+      "loss": 1.1227,
+      "step": 1440
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 8.001453591534846e-05,
+      "loss": 1.073,
+      "step": 1441
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 7.99886266423e-05,
+      "loss": 1.1664,
+      "step": 1442
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 7.996270478658084e-05,
+      "loss": 0.9982,
+      "step": 1443
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 7.993677035906735e-05,
+      "loss": 0.867,
+      "step": 1444
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 7.99108233706411e-05,
+      "loss": 1.0774,
+      "step": 1445
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 7.988486383218898e-05,
+      "loss": 0.8992,
+      "step": 1446
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 7.985889175460311e-05,
+      "loss": 1.1845,
+      "step": 1447
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 7.983290714878092e-05,
+      "loss": 1.2575,
+      "step": 1448
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 7.980691002562505e-05,
+      "loss": 0.9954,
+      "step": 1449
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 7.978090039604341e-05,
+      "loss": 1.1248,
+      "step": 1450
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 7.975487827094917e-05,
+      "loss": 1.035,
+      "step": 1451
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 7.972884366126072e-05,
+      "loss": 1.0451,
+      "step": 1452
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 7.97027965779017e-05,
+      "loss": 0.9388,
+      "step": 1453
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 7.967673703180096e-05,
+      "loss": 0.8035,
+      "step": 1454
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 7.965066503389264e-05,
+      "loss": 0.992,
+      "step": 1455
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 7.962458059511606e-05,
+      "loss": 0.9104,
+      "step": 1456
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 7.959848372641577e-05,
+      "loss": 1.0711,
+      "step": 1457
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 7.957237443874148e-05,
+      "loss": 1.0798,
+      "step": 1458
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 7.954625274304821e-05,
+      "loss": 1.1155,
+      "step": 1459
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 7.952011865029614e-05,
+      "loss": 1.1897,
+      "step": 1460
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 7.949397217145062e-05,
+      "loss": 1.1669,
+      "step": 1461
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 7.946781331748225e-05,
+      "loss": 0.9898,
+      "step": 1462
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 7.944164209936681e-05,
+      "loss": 1.073,
+      "step": 1463
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 7.941545852808522e-05,
+      "loss": 0.9468,
+      "step": 1464
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 7.938926261462366e-05,
+      "loss": 1.1316,
+      "step": 1465
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 7.936305436997343e-05,
+      "loss": 0.8649,
+      "step": 1466
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 7.933683380513099e-05,
+      "loss": 1.2812,
+      "step": 1467
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 7.931060093109807e-05,
+      "loss": 0.7277,
+      "step": 1468
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 7.928435575888144e-05,
+      "loss": 1.2493,
+      "step": 1469
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 7.925809829949312e-05,
+      "loss": 1.0225,
+      "step": 1470
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 7.923182856395024e-05,
+      "loss": 1.2568,
+      "step": 1471
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 7.920554656327508e-05,
+      "loss": 1.142,
+      "step": 1472
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 7.91792523084951e-05,
+      "loss": 1.1214,
+      "step": 1473
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 7.915294581064286e-05,
+      "loss": 1.2361,
+      "step": 1474
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 7.912662708075611e-05,
+      "loss": 1.0533,
+      "step": 1475
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 7.910029612987766e-05,
+      "loss": 1.1726,
+      "step": 1476
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 7.907395296905554e-05,
+      "loss": 0.9383,
+      "step": 1477
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 7.90475976093428e-05,
+      "loss": 1.0719,
+      "step": 1478
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 7.902123006179767e-05,
+      "loss": 0.9271,
+      "step": 1479
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 7.89948503374835e-05,
+      "loss": 1.1783,
+      "step": 1480
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 7.896845844746872e-05,
+      "loss": 1.094,
+      "step": 1481
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 7.894205440282689e-05,
+      "loss": 1.0178,
+      "step": 1482
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 7.891563821463664e-05,
+      "loss": 0.9915,
+      "step": 1483
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 7.888920989398174e-05,
+      "loss": 0.916,
+      "step": 1484
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 7.886276945195099e-05,
+      "loss": 0.8964,
+      "step": 1485
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 7.88363168996383e-05,
+      "loss": 1.1572,
+      "step": 1486
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 7.880985224814271e-05,
+      "loss": 1.0203,
+      "step": 1487
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 7.878337550856829e-05,
+      "loss": 1.0997,
+      "step": 1488
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 7.875688669202414e-05,
+      "loss": 0.8128,
+      "step": 1489
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 7.873038580962452e-05,
+      "loss": 1.0114,
+      "step": 1490
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 7.870387287248872e-05,
+      "loss": 1.1613,
+      "step": 1491
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 7.867734789174104e-05,
+      "loss": 1.0272,
+      "step": 1492
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 7.865081087851088e-05,
+      "loss": 1.2204,
+      "step": 1493
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 7.862426184393268e-05,
+      "loss": 0.8345,
+      "step": 1494
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 7.859770079914592e-05,
+      "loss": 1.124,
+      "step": 1495
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 7.857112775529514e-05,
+      "loss": 0.8901,
+      "step": 1496
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 7.854454272352987e-05,
+      "loss": 1.1176,
+      "step": 1497
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 7.85179457150047e-05,
+      "loss": 0.8434,
+      "step": 1498
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 7.849133674087924e-05,
+      "loss": 1.2866,
+      "step": 1499
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 7.846471581231814e-05,
+      "loss": 1.1743,
+      "step": 1500
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 7.843808294049103e-05,
+      "loss": 0.9542,
+      "step": 1501
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 7.841143813657257e-05,
+      "loss": 1.0275,
+      "step": 1502
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 7.838478141174244e-05,
+      "loss": 0.9885,
+      "step": 1503
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 7.835811277718527e-05,
+      "loss": 1.239,
+      "step": 1504
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 7.833143224409075e-05,
+      "loss": 1.017,
+      "step": 1505
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 7.830473982365354e-05,
+      "loss": 0.9559,
+      "step": 1506
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 7.827803552707329e-05,
+      "loss": 1.0444,
+      "step": 1507
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 7.82513193655546e-05,
+      "loss": 1.0596,
+      "step": 1508
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 7.822459135030709e-05,
+      "loss": 1.0206,
+      "step": 1509
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 7.819785149254532e-05,
+      "loss": 0.9799,
+      "step": 1510
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 7.817109980348889e-05,
+      "loss": 1.2142,
+      "step": 1511
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 7.814433629436225e-05,
+      "loss": 1.0349,
+      "step": 1512
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 7.81175609763949e-05,
+      "loss": 1.0714,
+      "step": 1513
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 7.809077386082128e-05,
+      "loss": 1.076,
+      "step": 1514
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 7.806397495888073e-05,
+      "loss": 0.8173,
+      "step": 1515
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 7.803716428181763e-05,
+      "loss": 1.0849,
+      "step": 1516
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 7.801034184088116e-05,
+      "loss": 1.0974,
+      "step": 1517
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 7.798350764732559e-05,
+      "loss": 0.9579,
+      "step": 1518
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 7.795666171241e-05,
+      "loss": 1.19,
+      "step": 1519
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 7.792980404739848e-05,
+      "loss": 1.0387,
+      "step": 1520
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 7.790293466356e-05,
+      "loss": 1.0822,
+      "step": 1521
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 7.787605357216843e-05,
+      "loss": 0.9865,
+      "step": 1522
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 7.78491607845026e-05,
+      "loss": 1.097,
+      "step": 1523
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 7.782225631184623e-05,
+      "loss": 1.071,
+      "step": 1524
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 7.779534016548791e-05,
+      "loss": 0.9471,
+      "step": 1525
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 7.776841235672119e-05,
+      "loss": 1.1299,
+      "step": 1526
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 7.774147289684445e-05,
+      "loss": 1.1195,
+      "step": 1527
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 7.771452179716099e-05,
+      "loss": 1.0155,
+      "step": 1528
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 7.7687559068979e-05,
+      "loss": 1.3389,
+      "step": 1529
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 7.766058472361154e-05,
+      "loss": 0.8215,
+      "step": 1530
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 7.763359877237655e-05,
+      "loss": 0.9589,
+      "step": 1531
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 7.760660122659682e-05,
+      "loss": 1.1295,
+      "step": 1532
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 7.75795920976e-05,
+      "loss": 1.0641,
+      "step": 1533
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 7.755257139671867e-05,
+      "loss": 1.1835,
+      "step": 1534
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 7.752553913529018e-05,
+      "loss": 1.1189,
+      "step": 1535
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 7.749849532465677e-05,
+      "loss": 1.2932,
+      "step": 1536
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 7.747143997616553e-05,
+      "loss": 1.1366,
+      "step": 1537
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 7.744437310116837e-05,
+      "loss": 1.1565,
+      "step": 1538
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 7.741729471102203e-05,
+      "loss": 1.0858,
+      "step": 1539
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 7.739020481708815e-05,
+      "loss": 1.0955,
+      "step": 1540
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 7.736310343073312e-05,
+      "loss": 0.9864,
+      "step": 1541
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 7.733599056332816e-05,
+      "loss": 1.0552,
+      "step": 1542
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 7.730886622624934e-05,
+      "loss": 0.8431,
+      "step": 1543
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 7.728173043087755e-05,
+      "loss": 1.1433,
+      "step": 1544
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 7.725458318859842e-05,
+      "loss": 0.9823,
+      "step": 1545
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 7.722742451080246e-05,
+      "loss": 0.9701,
+      "step": 1546
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 7.720025440888494e-05,
+      "loss": 0.9027,
+      "step": 1547
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 7.717307289424594e-05,
+      "loss": 1.4322,
+      "step": 1548
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 7.71458799782903e-05,
+      "loss": 1.2122,
+      "step": 1549
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 7.711867567242768e-05,
+      "loss": 0.9025,
+      "step": 1550
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 7.709145998807248e-05,
+      "loss": 1.0366,
+      "step": 1551
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 7.706423293664391e-05,
+      "loss": 1.1068,
+      "step": 1552
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 7.703699452956595e-05,
+      "loss": 1.3544,
+      "step": 1553
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 7.70097447782673e-05,
+      "loss": 1.0059,
+      "step": 1554
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 7.698248369418146e-05,
+      "loss": 0.9904,
+      "step": 1555
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 7.695521128874668e-05,
+      "loss": 1.0298,
+      "step": 1556
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 7.692792757340596e-05,
+      "loss": 1.0017,
+      "step": 1557
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 7.690063255960702e-05,
+      "loss": 1.3161,
+      "step": 1558
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 7.687332625880237e-05,
+      "loss": 0.9116,
+      "step": 1559
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 7.68460086824492e-05,
+      "loss": 0.9195,
+      "step": 1560
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 7.681867984200945e-05,
+      "loss": 1.6121,
+      "step": 1561
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 7.679133974894983e-05,
+      "loss": 0.9346,
+      "step": 1562
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 7.676398841474172e-05,
+      "loss": 1.2976,
+      "step": 1563
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 7.673662585086123e-05,
+      "loss": 1.1178,
+      "step": 1564
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 7.670925206878915e-05,
+      "loss": 0.9904,
+      "step": 1565
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 7.668186708001106e-05,
+      "loss": 1.0742,
+      "step": 1566
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 7.665447089601718e-05,
+      "loss": 1.2015,
+      "step": 1567
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 7.662706352830245e-05,
+      "loss": 1.0674,
+      "step": 1568
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 7.659964498836644e-05,
+      "loss": 1.103,
+      "step": 1569
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 7.657221528771352e-05,
+      "loss": 1.3906,
+      "step": 1570
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 7.654477443785265e-05,
+      "loss": 0.6711,
+      "step": 1571
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 7.651732245029753e-05,
+      "loss": 0.9468,
+      "step": 1572
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 7.648985933656646e-05,
+      "loss": 0.8497,
+      "step": 1573
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 7.646238510818249e-05,
+      "loss": 1.2278,
+      "step": 1574
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 7.643489977667326e-05,
+      "loss": 1.1534,
+      "step": 1575
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 7.640740335357115e-05,
+      "loss": 0.7397,
+      "step": 1576
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 7.637989585041313e-05,
+      "loss": 1.1681,
+      "step": 1577
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 7.635237727874083e-05,
+      "loss": 0.8704,
+      "step": 1578
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 7.632484765010052e-05,
+      "loss": 1.1411,
+      "step": 1579
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 7.629730697604314e-05,
+      "loss": 0.8879,
+      "step": 1580
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 7.626975526812426e-05,
+      "loss": 1.0557,
+      "step": 1581
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 7.624219253790403e-05,
+      "loss": 1.0763,
+      "step": 1582
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 7.621461879694726e-05,
+      "loss": 1.1099,
+      "step": 1583
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 7.618703405682341e-05,
+      "loss": 0.963,
+      "step": 1584
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 7.61594383291065e-05,
+      "loss": 1.1306,
+      "step": 1585
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 7.61318316253752e-05,
+      "loss": 1.1609,
+      "step": 1586
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 7.610421395721279e-05,
+      "loss": 1.0919,
+      "step": 1587
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 7.607658533620707e-05,
+      "loss": 0.9219,
+      "step": 1588
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 7.604894577395053e-05,
+      "loss": 0.9198,
+      "step": 1589
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 7.602129528204023e-05,
+      "loss": 0.9718,
+      "step": 1590
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 7.59936338720778e-05,
+      "loss": 1.02,
+      "step": 1591
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 7.596596155566942e-05,
+      "loss": 0.8313,
+      "step": 1592
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 7.593827834442592e-05,
+      "loss": 1.0676,
+      "step": 1593
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 7.591058424996263e-05,
+      "loss": 0.9941,
+      "step": 1594
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 7.588287928389952e-05,
+      "loss": 1.0203,
+      "step": 1595
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 7.585516345786103e-05,
+      "loss": 0.8965,
+      "step": 1596
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 7.582743678347624e-05,
+      "loss": 0.9538,
+      "step": 1597
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 7.579969927237871e-05,
+      "loss": 1.133,
+      "step": 1598
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 7.577195093620664e-05,
+      "loss": 0.8539,
+      "step": 1599
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 7.574419178660268e-05,
+      "loss": 1.1509,
+      "step": 1600
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 7.571642183521405e-05,
+      "loss": 1.0014,
+      "step": 1601
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 7.568864109369251e-05,
+      "loss": 0.9642,
+      "step": 1602
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 7.566084957369437e-05,
+      "loss": 1.282,
+      "step": 1603
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 7.56330472868804e-05,
+      "loss": 0.8232,
+      "step": 1604
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 7.560523424491595e-05,
+      "loss": 0.9664,
+      "step": 1605
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 7.557741045947083e-05,
+      "loss": 0.8691,
+      "step": 1606
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 7.554957594221939e-05,
+      "loss": 0.9148,
+      "step": 1607
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 7.552173070484048e-05,
+      "loss": 1.095,
+      "step": 1608
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 7.549387475901746e-05,
+      "loss": 0.9653,
+      "step": 1609
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 7.546600811643816e-05,
+      "loss": 1.1104,
+      "step": 1610
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 7.543813078879489e-05,
+      "loss": 0.8372,
+      "step": 1611
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 7.541024278778446e-05,
+      "loss": 0.8583,
+      "step": 1612
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 7.538234412510819e-05,
+      "loss": 1.0445,
+      "step": 1613
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 7.535443481247179e-05,
+      "loss": 0.9388,
+      "step": 1614
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 7.532651486158554e-05,
+      "loss": 1.0135,
+      "step": 1615
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 7.52985842841641e-05,
+      "loss": 1.1372,
+      "step": 1616
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 7.527064309192662e-05,
+      "loss": 1.0589,
+      "step": 1617
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 7.524269129659673e-05,
+      "loss": 1.1811,
+      "step": 1618
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 7.521472890990246e-05,
+      "loss": 1.1555,
+      "step": 1619
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 7.518675594357633e-05,
+      "loss": 0.8959,
+      "step": 1620
+    },
+    {
+      "epoch": 0.5,
+      "eval_loss": 1.1007293462753296,
+      "eval_runtime": 24.8051,
+      "eval_samples_per_second": 110.018,
+      "eval_steps_per_second": 13.788,
+      "step": 1620
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 7.515877240935525e-05,
+      "loss": 1.0774,
+      "step": 1621
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 7.513077831898062e-05,
+      "loss": 1.0432,
+      "step": 1622
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 7.510277368419822e-05,
+      "loss": 1.1261,
+      "step": 1623
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 7.507475851675827e-05,
+      "loss": 0.835,
+      "step": 1624
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 7.504673282841544e-05,
+      "loss": 1.2936,
+      "step": 1625
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 7.501869663092876e-05,
+      "loss": 1.0536,
+      "step": 1626
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 7.499064993606171e-05,
+      "loss": 1.0292,
+      "step": 1627
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 7.496259275558217e-05,
+      "loss": 0.9853,
+      "step": 1628
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 7.493452510126241e-05,
+      "loss": 1.1064,
+      "step": 1629
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 7.490644698487909e-05,
+      "loss": 0.9146,
+      "step": 1630
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 7.487835841821325e-05,
+      "loss": 1.069,
+      "step": 1631
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 7.485025941305036e-05,
+      "loss": 1.0743,
+      "step": 1632
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 7.482214998118022e-05,
+      "loss": 1.0122,
+      "step": 1633
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 7.479403013439706e-05,
+      "loss": 1.0293,
+      "step": 1634
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 7.476589988449939e-05,
+      "loss": 1.0461,
+      "step": 1635
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 7.473775924329017e-05,
+      "loss": 1.3707,
+      "step": 1636
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 7.470960822257669e-05,
+      "loss": 1.1007,
+      "step": 1637
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 7.468144683417062e-05,
+      "loss": 1.0038,
+      "step": 1638
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 7.465327508988791e-05,
+      "loss": 1.0347,
+      "step": 1639
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 7.462509300154892e-05,
+      "loss": 1.1155,
+      "step": 1640
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 7.459690058097832e-05,
+      "loss": 1.2323,
+      "step": 1641
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 7.456869784000516e-05,
+      "loss": 1.122,
+      "step": 1642
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 7.454048479046277e-05,
+      "loss": 1.0769,
+      "step": 1643
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 7.45122614441888e-05,
+      "loss": 1.0656,
+      "step": 1644
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 7.448402781302525e-05,
+      "loss": 1.2636,
+      "step": 1645
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 7.445578390881845e-05,
+      "loss": 1.2189,
+      "step": 1646
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 7.4427529743419e-05,
+      "loss": 1.0342,
+      "step": 1647
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 7.439926532868183e-05,
+      "loss": 1.2699,
+      "step": 1648
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 7.437099067646616e-05,
+      "loss": 1.0911,
+      "step": 1649
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 7.434270579863548e-05,
+      "loss": 0.9119,
+      "step": 1650
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 7.431441070705764e-05,
+      "loss": 1.2381,
+      "step": 1651
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 7.428610541360474e-05,
+      "loss": 1.3485,
+      "step": 1652
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 7.425778993015313e-05,
+      "loss": 0.9204,
+      "step": 1653
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 7.422946426858345e-05,
+      "loss": 0.9154,
+      "step": 1654
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 7.420112844078065e-05,
+      "loss": 0.8644,
+      "step": 1655
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 7.41727824586339e-05,
+      "loss": 0.8549,
+      "step": 1656
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 7.414442633403668e-05,
+      "loss": 0.9371,
+      "step": 1657
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 7.411606007888666e-05,
+      "loss": 1.167,
+      "step": 1658
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 7.408768370508576e-05,
+      "loss": 1.0084,
+      "step": 1659
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 7.405929722454026e-05,
+      "loss": 0.9907,
+      "step": 1660
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 7.403090064916054e-05,
+      "loss": 0.6976,
+      "step": 1661
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 7.40024939908613e-05,
+      "loss": 0.941,
+      "step": 1662
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 7.397407726156145e-05,
+      "loss": 0.8085,
+      "step": 1663
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 7.39456504731841e-05,
+      "loss": 1.0396,
+      "step": 1664
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 7.391721363765664e-05,
+      "loss": 1.1232,
+      "step": 1665
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 7.388876676691059e-05,
+      "loss": 1.0389,
+      "step": 1666
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 7.386030987288176e-05,
+      "loss": 0.9865,
+      "step": 1667
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 7.383184296751013e-05,
+      "loss": 1.1715,
+      "step": 1668
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 7.380336606273988e-05,
+      "loss": 1.0214,
+      "step": 1669
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 7.377487917051939e-05,
+      "loss": 1.1848,
+      "step": 1670
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 7.374638230280123e-05,
+      "loss": 1.07,
+      "step": 1671
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 7.371787547154216e-05,
+      "loss": 0.9229,
+      "step": 1672
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 7.368935868870311e-05,
+      "loss": 0.9466,
+      "step": 1673
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 7.36608319662492e-05,
+      "loss": 0.8591,
+      "step": 1674
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 7.363229531614972e-05,
+      "loss": 0.9754,
+      "step": 1675
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 7.36037487503781e-05,
+      "loss": 1.0268,
+      "step": 1676
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 7.357519228091196e-05,
+      "loss": 1.078,
+      "step": 1677
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 7.354662591973305e-05,
+      "loss": 1.3546,
+      "step": 1678
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 7.351804967882732e-05,
+      "loss": 0.9415,
+      "step": 1679
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 7.348946357018479e-05,
+      "loss": 1.1027,
+      "step": 1680
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 7.346086760579968e-05,
+      "loss": 1.1282,
+      "step": 1681
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 7.343226179767033e-05,
+      "loss": 1.1285,
+      "step": 1682
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 7.340364615779917e-05,
+      "loss": 1.142,
+      "step": 1683
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 7.337502069819286e-05,
+      "loss": 0.9609,
+      "step": 1684
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 7.334638543086203e-05,
+      "loss": 1.1116,
+      "step": 1685
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 7.331774036782158e-05,
+      "loss": 1.0108,
+      "step": 1686
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 7.328908552109037e-05,
+      "loss": 1.0038,
+      "step": 1687
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 7.326042090269151e-05,
+      "loss": 1.0142,
+      "step": 1688
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 7.323174652465212e-05,
+      "loss": 1.2512,
+      "step": 1689
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 7.320306239900343e-05,
+      "loss": 0.9588,
+      "step": 1690
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 7.317436853778077e-05,
+      "loss": 1.1786,
+      "step": 1691
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 7.314566495302353e-05,
+      "loss": 1.0566,
+      "step": 1692
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 7.311695165677523e-05,
+      "loss": 1.0892,
+      "step": 1693
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 7.308822866108342e-05,
+      "loss": 0.974,
+      "step": 1694
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 7.305949597799977e-05,
+      "loss": 0.9195,
+      "step": 1695
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 7.303075361957992e-05,
+      "loss": 0.9228,
+      "step": 1696
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 7.300200159788367e-05,
+      "loss": 1.0912,
+      "step": 1697
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 7.297323992497482e-05,
+      "loss": 1.6419,
+      "step": 1698
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 7.294446861292124e-05,
+      "loss": 0.8875,
+      "step": 1699
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 7.291568767379484e-05,
+      "loss": 0.9758,
+      "step": 1700
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 7.288689711967157e-05,
+      "loss": 1.0481,
+      "step": 1701
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 7.285809696263139e-05,
+      "loss": 1.1493,
+      "step": 1702
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 7.282928721475834e-05,
+      "loss": 1.082,
+      "step": 1703
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 7.280046788814044e-05,
+      "loss": 1.1309,
+      "step": 1704
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 7.277163899486975e-05,
+      "loss": 1.0792,
+      "step": 1705
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 7.274280054704233e-05,
+      "loss": 0.6976,
+      "step": 1706
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 7.271395255675826e-05,
+      "loss": 0.9329,
+      "step": 1707
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 7.268509503612162e-05,
+      "loss": 1.1158,
+      "step": 1708
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 7.265622799724049e-05,
+      "loss": 0.9887,
+      "step": 1709
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 7.262735145222696e-05,
+      "loss": 0.9575,
+      "step": 1710
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 7.259846541319706e-05,
+      "loss": 0.9694,
+      "step": 1711
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 7.256956989227084e-05,
+      "loss": 1.0261,
+      "step": 1712
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 7.254066490157237e-05,
+      "loss": 1.0895,
+      "step": 1713
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 7.251175045322959e-05,
+      "loss": 1.147,
+      "step": 1714
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 7.248282655937451e-05,
+      "loss": 1.0118,
+      "step": 1715
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 7.2453893232143e-05,
+      "loss": 0.8309,
+      "step": 1716
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 7.2424950483675e-05,
+      "loss": 1.1046,
+      "step": 1717
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 7.239599832611435e-05,
+      "loss": 1.2299,
+      "step": 1718
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 7.236703677160881e-05,
+      "loss": 1.2031,
+      "step": 1719
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 7.233806583231011e-05,
+      "loss": 1.0082,
+      "step": 1720
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 7.230908552037395e-05,
+      "loss": 1.0869,
+      "step": 1721
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 7.228009584795989e-05,
+      "loss": 0.8797,
+      "step": 1722
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 7.22510968272315e-05,
+      "loss": 1.188,
+      "step": 1723
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 7.22220884703562e-05,
+      "loss": 0.9649,
+      "step": 1724
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 7.219307078950535e-05,
+      "loss": 1.029,
+      "step": 1725
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 7.216404379685427e-05,
+      "loss": 1.0552,
+      "step": 1726
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 7.21350075045821e-05,
+      "loss": 1.0983,
+      "step": 1727
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 7.210596192487198e-05,
+      "loss": 1.0116,
+      "step": 1728
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 7.207690706991083e-05,
+      "loss": 1.1543,
+      "step": 1729
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 7.204784295188959e-05,
+      "loss": 1.1483,
+      "step": 1730
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 7.2018769583003e-05,
+      "loss": 0.7969,
+      "step": 1731
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 7.19896869754497e-05,
+      "loss": 1.0945,
+      "step": 1732
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 7.196059514143221e-05,
+      "loss": 1.125,
+      "step": 1733
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 7.193149409315693e-05,
+      "loss": 1.1732,
+      "step": 1734
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 7.190238384283412e-05,
+      "loss": 0.9398,
+      "step": 1735
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 7.18732644026779e-05,
+      "loss": 0.8941,
+      "step": 1736
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 7.184413578490625e-05,
+      "loss": 1.3027,
+      "step": 1737
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 7.181499800174098e-05,
+      "loss": 1.256,
+      "step": 1738
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 7.178585106540778e-05,
+      "loss": 1.2151,
+      "step": 1739
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 7.175669498813617e-05,
+      "loss": 1.1227,
+      "step": 1740
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 7.172752978215948e-05,
+      "loss": 1.0806,
+      "step": 1741
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 7.169835545971489e-05,
+      "loss": 1.1249,
+      "step": 1742
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 7.16691720330434e-05,
+      "loss": 0.9705,
+      "step": 1743
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 7.163997951438986e-05,
+      "loss": 0.9135,
+      "step": 1744
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 7.161077791600288e-05,
+      "loss": 1.0933,
+      "step": 1745
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 7.158156725013493e-05,
+      "loss": 0.9969,
+      "step": 1746
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 7.155234752904223e-05,
+      "loss": 0.9868,
+      "step": 1747
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 7.152311876498487e-05,
+      "loss": 1.0465,
+      "step": 1748
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 7.149388097022667e-05,
+      "loss": 1.0505,
+      "step": 1749
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 7.146463415703531e-05,
+      "loss": 0.9331,
+      "step": 1750
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 7.143537833768214e-05,
+      "loss": 1.3671,
+      "step": 1751
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 7.14061135244424e-05,
+      "loss": 1.0157,
+      "step": 1752
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 7.137683972959505e-05,
+      "loss": 0.921,
+      "step": 1753
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 7.134755696542286e-05,
+      "loss": 1.0802,
+      "step": 1754
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 7.13182652442123e-05,
+      "loss": 1.246,
+      "step": 1755
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 7.128896457825364e-05,
+      "loss": 1.1682,
+      "step": 1756
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 7.125965497984089e-05,
+      "loss": 1.026,
+      "step": 1757
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 7.123033646127183e-05,
+      "loss": 1.3554,
+      "step": 1758
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 7.120100903484797e-05,
+      "loss": 0.9812,
+      "step": 1759
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 7.117167271287453e-05,
+      "loss": 1.1717,
+      "step": 1760
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 7.114232750766048e-05,
+      "loss": 0.9885,
+      "step": 1761
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 7.111297343151853e-05,
+      "loss": 0.9443,
+      "step": 1762
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 7.108361049676516e-05,
+      "loss": 0.8757,
+      "step": 1763
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 7.105423871572044e-05,
+      "loss": 1.0498,
+      "step": 1764
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 7.102485810070824e-05,
+      "loss": 1.1067,
+      "step": 1765
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 7.099546866405612e-05,
+      "loss": 1.1988,
+      "step": 1766
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 7.096607041809537e-05,
+      "loss": 1.092,
+      "step": 1767
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 7.093666337516094e-05,
+      "loss": 1.0919,
+      "step": 1768
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 7.090724754759147e-05,
+      "loss": 0.9829,
+      "step": 1769
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 7.087782294772927e-05,
+      "loss": 1.0816,
+      "step": 1770
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 7.084838958792039e-05,
+      "loss": 0.9313,
+      "step": 1771
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 7.081894748051451e-05,
+      "loss": 1.0496,
+      "step": 1772
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 7.078949663786499e-05,
+      "loss": 1.0321,
+      "step": 1773
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 7.076003707232886e-05,
+      "loss": 1.1033,
+      "step": 1774
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 7.073056879626681e-05,
+      "loss": 0.9792,
+      "step": 1775
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 7.070109182204316e-05,
+      "loss": 1.0085,
+      "step": 1776
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 7.067160616202593e-05,
+      "loss": 1.1701,
+      "step": 1777
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 7.064211182858673e-05,
+      "loss": 1.003,
+      "step": 1778
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 7.061260883410086e-05,
+      "loss": 0.9982,
+      "step": 1779
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 7.05830971909472e-05,
+      "loss": 0.8691,
+      "step": 1780
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 7.05535769115083e-05,
+      "loss": 0.8333,
+      "step": 1781
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 7.052404800817035e-05,
+      "loss": 1.1291,
+      "step": 1782
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 7.049451049332306e-05,
+      "loss": 1.118,
+      "step": 1783
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 7.046496437935989e-05,
+      "loss": 0.975,
+      "step": 1784
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 7.043540967867782e-05,
+      "loss": 1.0711,
+      "step": 1785
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 7.040584640367744e-05,
+      "loss": 1.0789,
+      "step": 1786
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 7.037627456676294e-05,
+      "loss": 0.9934,
+      "step": 1787
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 7.034669418034216e-05,
+      "loss": 0.7577,
+      "step": 1788
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 7.031710525682645e-05,
+      "loss": 1.0155,
+      "step": 1789
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 7.028750780863078e-05,
+      "loss": 0.882,
+      "step": 1790
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 7.025790184817367e-05,
+      "loss": 1.0951,
+      "step": 1791
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 7.022828738787724e-05,
+      "loss": 0.9762,
+      "step": 1792
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 7.01986644401672e-05,
+      "loss": 1.011,
+      "step": 1793
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 7.016903301747275e-05,
+      "loss": 0.7383,
+      "step": 1794
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 7.01393931322267e-05,
+      "loss": 1.2669,
+      "step": 1795
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 7.010974479686538e-05,
+      "loss": 1.0308,
+      "step": 1796
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 7.008008802382871e-05,
+      "loss": 1.0665,
+      "step": 1797
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 7.00504228255601e-05,
+      "loss": 0.935,
+      "step": 1798
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 7.002074921450653e-05,
+      "loss": 1.0925,
+      "step": 1799
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 6.999106720311845e-05,
+      "loss": 0.9888,
+      "step": 1800
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 6.996137680384994e-05,
+      "loss": 1.0338,
+      "step": 1801
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 6.993167802915853e-05,
+      "loss": 1.1179,
+      "step": 1802
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 6.990197089150525e-05,
+      "loss": 1.0464,
+      "step": 1803
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 6.987225540335467e-05,
+      "loss": 0.7863,
+      "step": 1804
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 6.984253157717487e-05,
+      "loss": 0.8877,
+      "step": 1805
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 6.98127994254374e-05,
+      "loss": 1.2868,
+      "step": 1806
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 6.978305896061735e-05,
+      "loss": 0.9605,
+      "step": 1807
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 6.975331019519322e-05,
+      "loss": 1.3444,
+      "step": 1808
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 6.972355314164705e-05,
+      "loss": 1.1225,
+      "step": 1809
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 6.969378781246436e-05,
+      "loss": 1.1341,
+      "step": 1810
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 6.966401422013415e-05,
+      "loss": 1.1363,
+      "step": 1811
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 6.963423237714883e-05,
+      "loss": 1.1386,
+      "step": 1812
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 6.960444229600431e-05,
+      "loss": 1.0134,
+      "step": 1813
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 6.957464398919997e-05,
+      "loss": 1.0808,
+      "step": 1814
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 6.954483746923865e-05,
+      "loss": 1.1968,
+      "step": 1815
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 6.951502274862655e-05,
+      "loss": 1.1169,
+      "step": 1816
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 6.948519983987344e-05,
+      "loss": 1.0216,
+      "step": 1817
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 6.94553687554924e-05,
+      "loss": 0.7387,
+      "step": 1818
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 6.942552950800004e-05,
+      "loss": 1.125,
+      "step": 1819
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 6.939568210991633e-05,
+      "loss": 1.205,
+      "step": 1820
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 6.93658265737647e-05,
+      "loss": 1.061,
+      "step": 1821
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 6.933596291207196e-05,
+      "loss": 1.0488,
+      "step": 1822
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 6.930609113736837e-05,
+      "loss": 1.0046,
+      "step": 1823
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 6.927621126218755e-05,
+      "loss": 1.2861,
+      "step": 1824
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 6.924632329906656e-05,
+      "loss": 0.966,
+      "step": 1825
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 6.921642726054583e-05,
+      "loss": 1.2003,
+      "step": 1826
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 6.918652315916915e-05,
+      "loss": 1.0995,
+      "step": 1827
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 6.915661100748378e-05,
+      "loss": 1.0292,
+      "step": 1828
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 6.912669081804028e-05,
+      "loss": 1.1036,
+      "step": 1829
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 6.90967626033926e-05,
+      "loss": 0.9856,
+      "step": 1830
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 6.906682637609805e-05,
+      "loss": 0.9197,
+      "step": 1831
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 6.903688214871734e-05,
+      "loss": 0.9357,
+      "step": 1832
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 6.900692993381449e-05,
+      "loss": 1.0031,
+      "step": 1833
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 6.897696974395691e-05,
+      "loss": 0.9832,
+      "step": 1834
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 6.894700159171534e-05,
+      "loss": 1.1124,
+      "step": 1835
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 6.891702548966385e-05,
+      "loss": 1.0723,
+      "step": 1836
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 6.888704145037985e-05,
+      "loss": 0.9013,
+      "step": 1837
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 6.885704948644411e-05,
+      "loss": 0.8807,
+      "step": 1838
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 6.882704961044068e-05,
+      "loss": 0.9514,
+      "step": 1839
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 6.879704183495695e-05,
+      "loss": 1.0419,
+      "step": 1840
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 6.876702617258365e-05,
+      "loss": 1.0767,
+      "step": 1841
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 6.873700263591476e-05,
+      "loss": 1.0016,
+      "step": 1842
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 6.870697123754762e-05,
+      "loss": 1.1651,
+      "step": 1843
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 6.867693199008285e-05,
+      "loss": 0.9581,
+      "step": 1844
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 6.864688490612435e-05,
+      "loss": 1.155,
+      "step": 1845
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 6.86168299982793e-05,
+      "loss": 1.1493,
+      "step": 1846
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 6.858676727915822e-05,
+      "loss": 1.5982,
+      "step": 1847
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 6.855669676137483e-05,
+      "loss": 0.9385,
+      "step": 1848
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 6.85266184575462e-05,
+      "loss": 1.098,
+      "step": 1849
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 6.84965323802926e-05,
+      "loss": 1.2341,
+      "step": 1850
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 6.84664385422376e-05,
+      "loss": 1.0443,
+      "step": 1851
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 6.843633695600802e-05,
+      "loss": 1.1741,
+      "step": 1852
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 6.840622763423391e-05,
+      "loss": 1.0315,
+      "step": 1853
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 6.837611058954858e-05,
+      "loss": 1.0289,
+      "step": 1854
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 6.834598583458862e-05,
+      "loss": 1.0414,
+      "step": 1855
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 6.831585338199376e-05,
+      "loss": 1.0964,
+      "step": 1856
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 6.828571324440707e-05,
+      "loss": 0.9021,
+      "step": 1857
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 6.825556543447476e-05,
+      "loss": 1.1038,
+      "step": 1858
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 6.822540996484631e-05,
+      "loss": 1.0664,
+      "step": 1859
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 6.819524684817438e-05,
+      "loss": 0.9376,
+      "step": 1860
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 6.816507609711487e-05,
+      "loss": 1.014,
+      "step": 1861
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 6.813489772432684e-05,
+      "loss": 1.0861,
+      "step": 1862
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 6.810471174247259e-05,
+      "loss": 0.9739,
+      "step": 1863
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 6.807451816421761e-05,
+      "loss": 0.9956,
+      "step": 1864
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 6.804431700223056e-05,
+      "loss": 0.9404,
+      "step": 1865
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 6.801410826918327e-05,
+      "loss": 1.3265,
+      "step": 1866
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 6.798389197775076e-05,
+      "loss": 1.1082,
+      "step": 1867
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 6.795366814061126e-05,
+      "loss": 1.0341,
+      "step": 1868
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 6.79234367704461e-05,
+      "loss": 0.9736,
+      "step": 1869
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 6.789319787993979e-05,
+      "loss": 0.7658,
+      "step": 1870
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 6.786295148178005e-05,
+      "loss": 1.0874,
+      "step": 1871
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 6.783269758865768e-05,
+      "loss": 1.0889,
+      "step": 1872
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 6.780243621326665e-05,
+      "loss": 1.1022,
+      "step": 1873
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 6.777216736830409e-05,
+      "loss": 0.8778,
+      "step": 1874
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 6.77418910664702e-05,
+      "loss": 0.8237,
+      "step": 1875
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 6.77116073204684e-05,
+      "loss": 0.9944,
+      "step": 1876
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 6.768131614300518e-05,
+      "loss": 1.014,
+      "step": 1877
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 6.765101754679015e-05,
+      "loss": 1.0666,
+      "step": 1878
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 6.762071154453603e-05,
+      "loss": 1.0099,
+      "step": 1879
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 6.759039814895862e-05,
+      "loss": 1.2179,
+      "step": 1880
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 6.756007737277694e-05,
+      "loss": 1.0911,
+      "step": 1881
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 6.752974922871299e-05,
+      "loss": 1.1646,
+      "step": 1882
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 6.749941372949188e-05,
+      "loss": 0.9451,
+      "step": 1883
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 6.746907088784182e-05,
+      "loss": 0.8038,
+      "step": 1884
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 6.743872071649411e-05,
+      "loss": 0.8414,
+      "step": 1885
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 6.740836322818313e-05,
+      "loss": 1.0514,
+      "step": 1886
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 6.737799843564632e-05,
+      "loss": 0.9345,
+      "step": 1887
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 6.734762635162418e-05,
+      "loss": 0.957,
+      "step": 1888
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 6.731724698886022e-05,
+      "loss": 1.035,
+      "step": 1889
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 6.728686036010115e-05,
+      "loss": 0.9117,
+      "step": 1890
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 6.725646647809656e-05,
+      "loss": 0.8381,
+      "step": 1891
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 6.72260653555992e-05,
+      "loss": 1.1935,
+      "step": 1892
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 6.719565700536478e-05,
+      "loss": 1.0916,
+      "step": 1893
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 6.716524144015212e-05,
+      "loss": 1.0774,
+      "step": 1894
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 6.7134818672723e-05,
+      "loss": 1.0375,
+      "step": 1895
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 6.710438871584224e-05,
+      "loss": 0.8994,
+      "step": 1896
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 6.707395158227772e-05,
+      "loss": 1.0408,
+      "step": 1897
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 6.704350728480025e-05,
+      "loss": 1.1238,
+      "step": 1898
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 6.701305583618373e-05,
+      "loss": 0.8306,
+      "step": 1899
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 6.698259724920502e-05,
+      "loss": 1.1842,
+      "step": 1900
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 6.695213153664397e-05,
+      "loss": 0.8855,
+      "step": 1901
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 6.69216587112834e-05,
+      "loss": 0.9718,
+      "step": 1902
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 6.689117878590917e-05,
+      "loss": 0.9205,
+      "step": 1903
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 6.68606917733101e-05,
+      "loss": 1.3419,
+      "step": 1904
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 6.683019768627795e-05,
+      "loss": 1.1337,
+      "step": 1905
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 6.679969653760747e-05,
+      "loss": 0.7672,
+      "step": 1906
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 6.67691883400964e-05,
+      "loss": 0.9755,
+      "step": 1907
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 6.673867310654538e-05,
+      "loss": 1.0797,
+      "step": 1908
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 6.670815084975807e-05,
+      "loss": 1.0841,
+      "step": 1909
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 6.667762158254104e-05,
+      "loss": 1.1982,
+      "step": 1910
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 6.664708531770378e-05,
+      "loss": 1.0894,
+      "step": 1911
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 6.661654206805873e-05,
+      "loss": 1.141,
+      "step": 1912
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 6.658599184642132e-05,
+      "loss": 1.1332,
+      "step": 1913
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 6.655543466560981e-05,
+      "loss": 0.9568,
+      "step": 1914
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 6.652487053844544e-05,
+      "loss": 1.0526,
+      "step": 1915
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 6.649429947775236e-05,
+      "loss": 1.12,
+      "step": 1916
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 6.64637214963576e-05,
+      "loss": 0.8265,
+      "step": 1917
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 6.643313660709114e-05,
+      "loss": 1.0044,
+      "step": 1918
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 6.640254482278581e-05,
+      "loss": 1.089,
+      "step": 1919
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 6.637194615627733e-05,
+      "loss": 0.9623,
+      "step": 1920
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 6.634134062040441e-05,
+      "loss": 1.0361,
+      "step": 1921
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 6.631072822800846e-05,
+      "loss": 0.9829,
+      "step": 1922
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 6.628010899193398e-05,
+      "loss": 1.0511,
+      "step": 1923
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 6.624948292502814e-05,
+      "loss": 1.0011,
+      "step": 1924
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 6.621885004014112e-05,
+      "loss": 1.0013,
+      "step": 1925
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 6.618821035012592e-05,
+      "loss": 1.0973,
+      "step": 1926
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 6.615756386783834e-05,
+      "loss": 1.0911,
+      "step": 1927
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.61269106061371e-05,
+      "loss": 1.1074,
+      "step": 1928
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.609625057788372e-05,
+      "loss": 1.3054,
+      "step": 1929
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.606558379594262e-05,
+      "loss": 1.0119,
+      "step": 1930
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.603491027318097e-05,
+      "loss": 0.9643,
+      "step": 1931
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.600423002246885e-05,
+      "loss": 1.0327,
+      "step": 1932
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.597354305667908e-05,
+      "loss": 1.0064,
+      "step": 1933
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.594284938868737e-05,
+      "loss": 0.9205,
+      "step": 1934
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.59121490313722e-05,
+      "loss": 0.9257,
+      "step": 1935
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.588144199761487e-05,
+      "loss": 1.049,
+      "step": 1936
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.585072830029949e-05,
+      "loss": 0.9125,
+      "step": 1937
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.582000795231295e-05,
+      "loss": 0.8406,
+      "step": 1938
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.578928096654494e-05,
+      "loss": 0.8304,
+      "step": 1939
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.575854735588794e-05,
+      "loss": 1.0656,
+      "step": 1940
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.572780713323719e-05,
+      "loss": 1.2346,
+      "step": 1941
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.56970603114907e-05,
+      "loss": 1.1596,
+      "step": 1942
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.56663069035493e-05,
+      "loss": 1.0183,
+      "step": 1943
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.563554692231655e-05,
+      "loss": 0.8952,
+      "step": 1944
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.560478038069873e-05,
+      "loss": 1.1148,
+      "step": 1945
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.557400729160493e-05,
+      "loss": 0.9085,
+      "step": 1946
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.554322766794697e-05,
+      "loss": 1.042,
+      "step": 1947
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.55124415226394e-05,
+      "loss": 1.1216,
+      "step": 1948
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.548164886859955e-05,
+      "loss": 0.9356,
+      "step": 1949
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.545084971874738e-05,
+      "loss": 0.977,
+      "step": 1950
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.542004408600569e-05,
+      "loss": 1.0234,
+      "step": 1951
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.538923198329993e-05,
+      "loss": 0.8402,
+      "step": 1952
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.53584134235583e-05,
+      "loss": 0.8707,
+      "step": 1953
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.53275884197117e-05,
+      "loss": 0.9848,
+      "step": 1954
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.52967569846937e-05,
+      "loss": 1.0752,
+      "step": 1955
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.526591913144061e-05,
+      "loss": 0.9174,
+      "step": 1956
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.523507487289145e-05,
+      "loss": 1.0929,
+      "step": 1957
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.520422422198787e-05,
+      "loss": 0.9353,
+      "step": 1958
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.517336719167421e-05,
+      "loss": 1.4227,
+      "step": 1959
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.514250379489753e-05,
+      "loss": 1.0392,
+      "step": 1960
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.511163404460756e-05,
+      "loss": 1.145,
+      "step": 1961
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.508075795375665e-05,
+      "loss": 0.8454,
+      "step": 1962
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.504987553529984e-05,
+      "loss": 0.8107,
+      "step": 1963
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.50189868021948e-05,
+      "loss": 1.0409,
+      "step": 1964
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.498809176740189e-05,
+      "loss": 1.3317,
+      "step": 1965
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.495719044388409e-05,
+      "loss": 1.1917,
+      "step": 1966
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.492628284460702e-05,
+      "loss": 0.931,
+      "step": 1967
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.489536898253893e-05,
+      "loss": 0.8686,
+      "step": 1968
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.48644488706507e-05,
+      "loss": 1.3127,
+      "step": 1969
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.483352252191584e-05,
+      "loss": 1.0578,
+      "step": 1970
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.480258994931048e-05,
+      "loss": 1.0356,
+      "step": 1971
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.477165116581333e-05,
+      "loss": 0.6953,
+      "step": 1972
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.474070618440576e-05,
+      "loss": 1.1527,
+      "step": 1973
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.470975501807168e-05,
+      "loss": 0.9629,
+      "step": 1974
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.467879767979764e-05,
+      "loss": 0.9815,
+      "step": 1975
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.464783418257277e-05,
+      "loss": 1.0326,
+      "step": 1976
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.461686453938876e-05,
+      "loss": 1.0595,
+      "step": 1977
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.45858887632399e-05,
+      "loss": 1.018,
+      "step": 1978
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.455490686712307e-05,
+      "loss": 1.0029,
+      "step": 1979
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.452391886403767e-05,
+      "loss": 1.0541,
+      "step": 1980
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.449292476698572e-05,
+      "loss": 1.1802,
+      "step": 1981
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.446192458897173e-05,
+      "loss": 1.014,
+      "step": 1982
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.443091834300283e-05,
+      "loss": 1.0219,
+      "step": 1983
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.439990604208868e-05,
+      "loss": 0.9729,
+      "step": 1984
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.436888769924142e-05,
+      "loss": 1.0904,
+      "step": 1985
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.433786332747578e-05,
+      "loss": 1.0736,
+      "step": 1986
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.430683293980905e-05,
+      "loss": 0.9895,
+      "step": 1987
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.427579654926095e-05,
+      "loss": 0.9904,
+      "step": 1988
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.424475416885382e-05,
+      "loss": 1.0981,
+      "step": 1989
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.421370581161243e-05,
+      "loss": 1.2544,
+      "step": 1990
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.418265149056412e-05,
+      "loss": 1.1331,
+      "step": 1991
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.415159121873868e-05,
+      "loss": 0.7367,
+      "step": 1992
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.412052500916845e-05,
+      "loss": 1.1033,
+      "step": 1993
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.408945287488823e-05,
+      "loss": 1.3143,
+      "step": 1994
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.405837482893528e-05,
+      "loss": 1.192,
+      "step": 1995
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.402729088434942e-05,
+      "loss": 0.9367,
+      "step": 1996
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.399620105417285e-05,
+      "loss": 1.1553,
+      "step": 1997
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.396510535145033e-05,
+      "loss": 0.9773,
+      "step": 1998
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.3934003789229e-05,
+      "loss": 1.0711,
+      "step": 1999
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.390289638055851e-05,
+      "loss": 1.2339,
+      "step": 2000
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.387178313849096e-05,
+      "loss": 0.9115,
+      "step": 2001
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.384066407608087e-05,
+      "loss": 1.0086,
+      "step": 2002
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.380953920638524e-05,
+      "loss": 0.8988,
+      "step": 2003
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.377840854246348e-05,
+      "loss": 0.8311,
+      "step": 2004
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.374727209737744e-05,
+      "loss": 0.9145,
+      "step": 2005
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.371612988419137e-05,
+      "loss": 1.0694,
+      "step": 2006
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.3684981915972e-05,
+      "loss": 1.1212,
+      "step": 2007
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.365382820578845e-05,
+      "loss": 1.0768,
+      "step": 2008
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.362266876671217e-05,
+      "loss": 1.2388,
+      "step": 2009
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.359150361181715e-05,
+      "loss": 1.2467,
+      "step": 2010
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.356033275417967e-05,
+      "loss": 1.1344,
+      "step": 2011
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.352915620687847e-05,
+      "loss": 1.0802,
+      "step": 2012
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.349797398299463e-05,
+      "loss": 1.1091,
+      "step": 2013
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.346678609561165e-05,
+      "loss": 0.9194,
+      "step": 2014
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.343559255781537e-05,
+      "loss": 1.1533,
+      "step": 2015
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.340439338269402e-05,
+      "loss": 1.045,
+      "step": 2016
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.33731885833382e-05,
+      "loss": 1.0561,
+      "step": 2017
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.334197817284088e-05,
+      "loss": 1.1247,
+      "step": 2018
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.331076216429735e-05,
+      "loss": 1.1044,
+      "step": 2019
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.327954057080526e-05,
+      "loss": 0.9763,
+      "step": 2020
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.324831340546463e-05,
+      "loss": 1.0923,
+      "step": 2021
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.321708068137779e-05,
+      "loss": 1.0268,
+      "step": 2022
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.318584241164942e-05,
+      "loss": 0.9005,
+      "step": 2023
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.315459860938649e-05,
+      "loss": 1.2152,
+      "step": 2024
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.312334928769835e-05,
+      "loss": 1.0222,
+      "step": 2025
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.30920944596966e-05,
+      "loss": 0.9222,
+      "step": 2026
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.306083413849524e-05,
+      "loss": 0.9446,
+      "step": 2027
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.302956833721047e-05,
+      "loss": 0.9467,
+      "step": 2028
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.299829706896087e-05,
+      "loss": 0.9007,
+      "step": 2029
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.296702034686725e-05,
+      "loss": 1.1401,
+      "step": 2030
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.293573818405279e-05,
+      "loss": 0.8826,
+      "step": 2031
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.290445059364286e-05,
+      "loss": 1.2152,
+      "step": 2032
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.287315758876518e-05,
+      "loss": 0.9786,
+      "step": 2033
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.284185918254968e-05,
+      "loss": 0.9813,
+      "step": 2034
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.281055538812861e-05,
+      "loss": 1.2591,
+      "step": 2035
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.277924621863649e-05,
+      "loss": 0.8275,
+      "step": 2036
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.274793168721002e-05,
+      "loss": 1.1275,
+      "step": 2037
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.271661180698824e-05,
+      "loss": 0.9995,
+      "step": 2038
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.268528659111233e-05,
+      "loss": 1.0665,
+      "step": 2039
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.265395605272581e-05,
+      "loss": 1.0394,
+      "step": 2040
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.26226202049744e-05,
+      "loss": 1.1027,
+      "step": 2041
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.259127906100602e-05,
+      "loss": 1.1526,
+      "step": 2042
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.255993263397083e-05,
+      "loss": 1.0435,
+      "step": 2043
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.252858093702121e-05,
+      "loss": 1.1474,
+      "step": 2044
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.249722398331176e-05,
+      "loss": 1.1542,
+      "step": 2045
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.246586178599928e-05,
+      "loss": 0.9045,
+      "step": 2046
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.243449435824276e-05,
+      "loss": 1.3024,
+      "step": 2047
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.240312171320336e-05,
+      "loss": 0.9711,
+      "step": 2048
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.237174386404449e-05,
+      "loss": 0.7768,
+      "step": 2049
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.23403608239317e-05,
+      "loss": 1.064,
+      "step": 2050
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.230897260603277e-05,
+      "loss": 1.14,
+      "step": 2051
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.227757922351755e-05,
+      "loss": 1.1622,
+      "step": 2052
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.224618068955816e-05,
+      "loss": 1.0429,
+      "step": 2053
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.221477701732884e-05,
+      "loss": 1.0023,
+      "step": 2054
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.218336822000598e-05,
+      "loss": 0.9548,
+      "step": 2055
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.215195431076813e-05,
+      "loss": 0.9158,
+      "step": 2056
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.212053530279597e-05,
+      "loss": 0.9072,
+      "step": 2057
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.208911120927234e-05,
+      "loss": 0.8596,
+      "step": 2058
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.205768204338221e-05,
+      "loss": 0.9733,
+      "step": 2059
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.202624781831268e-05,
+      "loss": 0.7668,
+      "step": 2060
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.199480854725296e-05,
+      "loss": 1.2297,
+      "step": 2061
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.196336424339439e-05,
+      "loss": 1.2033,
+      "step": 2062
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.193191491993039e-05,
+      "loss": 1.0316,
+      "step": 2063
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.190046059005655e-05,
+      "loss": 1.0828,
+      "step": 2064
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.186900126697051e-05,
+      "loss": 1.0459,
+      "step": 2065
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.183753696387199e-05,
+      "loss": 1.1223,
+      "step": 2066
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.180606769396286e-05,
+      "loss": 1.1681,
+      "step": 2067
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.177459347044703e-05,
+      "loss": 1.1054,
+      "step": 2068
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.17431143065305e-05,
+      "loss": 0.8328,
+      "step": 2069
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.171163021542133e-05,
+      "loss": 0.9898,
+      "step": 2070
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.168014121032969e-05,
+      "loss": 1.0329,
+      "step": 2071
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.164864730446776e-05,
+      "loss": 1.2302,
+      "step": 2072
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.161714851104979e-05,
+      "loss": 1.0931,
+      "step": 2073
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.158564484329213e-05,
+      "loss": 1.1175,
+      "step": 2074
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.155413631441307e-05,
+      "loss": 1.0914,
+      "step": 2075
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.152262293763305e-05,
+      "loss": 1.0668,
+      "step": 2076
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.149110472617448e-05,
+      "loss": 0.8983,
+      "step": 2077
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.145958169326182e-05,
+      "loss": 1.1101,
+      "step": 2078
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.142805385212156e-05,
+      "loss": 1.1401,
+      "step": 2079
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.139652121598218e-05,
+      "loss": 1.0104,
+      "step": 2080
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.136498379807418e-05,
+      "loss": 1.0857,
+      "step": 2081
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.133344161163012e-05,
+      "loss": 1.1454,
+      "step": 2082
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.130189466988447e-05,
+      "loss": 1.2005,
+      "step": 2083
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.127034298607375e-05,
+      "loss": 0.8893,
+      "step": 2084
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.123878657343648e-05,
+      "loss": 1.2587,
+      "step": 2085
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.120722544521311e-05,
+      "loss": 0.9204,
+      "step": 2086
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.117565961464614e-05,
+      "loss": 0.9433,
+      "step": 2087
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.114408909497998e-05,
+      "loss": 1.02,
+      "step": 2088
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.111251389946105e-05,
+      "loss": 1.1814,
+      "step": 2089
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.108093404133773e-05,
+      "loss": 1.0929,
+      "step": 2090
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.104934953386029e-05,
+      "loss": 1.3206,
+      "step": 2091
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.101776039028104e-05,
+      "loss": 0.8155,
+      "step": 2092
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.098616662385419e-05,
+      "loss": 0.9347,
+      "step": 2093
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.095456824783592e-05,
+      "loss": 0.8972,
+      "step": 2094
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.092296527548427e-05,
+      "loss": 0.8613,
+      "step": 2095
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.0891357720059314e-05,
+      "loss": 1.172,
+      "step": 2096
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.085974559482296e-05,
+      "loss": 0.8049,
+      "step": 2097
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.0828128913039085e-05,
+      "loss": 0.7073,
+      "step": 2098
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.079650768797346e-05,
+      "loss": 0.9336,
+      "step": 2099
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.076488193289375e-05,
+      "loss": 0.9501,
+      "step": 2100
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.0733251661069557e-05,
+      "loss": 0.9844,
+      "step": 2101
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.070161688577233e-05,
+      "loss": 0.9716,
+      "step": 2102
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.066997762027546e-05,
+      "loss": 0.9908,
+      "step": 2103
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.063833387785418e-05,
+      "loss": 1.0144,
+      "step": 2104
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.06066856717856e-05,
+      "loss": 0.9148,
+      "step": 2105
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.057503301534875e-05,
+      "loss": 1.1145,
+      "step": 2106
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.0543375921824485e-05,
+      "loss": 0.9586,
+      "step": 2107
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.0511714404495546e-05,
+      "loss": 0.8205,
+      "step": 2108
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.048004847664649e-05,
+      "loss": 1.0161,
+      "step": 2109
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.044837815156377e-05,
+      "loss": 1.0831,
+      "step": 2110
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.041670344253565e-05,
+      "loss": 0.9721,
+      "step": 2111
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.038502436285227e-05,
+      "loss": 1.0811,
+      "step": 2112
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.035334092580558e-05,
+      "loss": 0.9918,
+      "step": 2113
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.032165314468935e-05,
+      "loss": 1.0303,
+      "step": 2114
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.028996103279918e-05,
+      "loss": 0.8197,
+      "step": 2115
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.0258264603432526e-05,
+      "loss": 0.8403,
+      "step": 2116
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.0226563869888596e-05,
+      "loss": 0.9733,
+      "step": 2117
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.019485884546843e-05,
+      "loss": 1.4516,
+      "step": 2118
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.016314954347484e-05,
+      "loss": 0.9806,
+      "step": 2119
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.013143597721251e-05,
+      "loss": 1.0129,
+      "step": 2120
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.009971815998785e-05,
+      "loss": 0.941,
+      "step": 2121
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.006799610510905e-05,
+      "loss": 1.0758,
+      "step": 2122
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 6.0036269825886094e-05,
+      "loss": 0.969,
+      "step": 2123
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 6.000453933563075e-05,
+      "loss": 1.1753,
+      "step": 2124
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.9972804647656544e-05,
+      "loss": 1.1388,
+      "step": 2125
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.994106577527877e-05,
+      "loss": 0.9837,
+      "step": 2126
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.9909322731814456e-05,
+      "loss": 1.218,
+      "step": 2127
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.987757553058236e-05,
+      "loss": 1.1808,
+      "step": 2128
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.9845824184903086e-05,
+      "loss": 0.9543,
+      "step": 2129
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.981406870809888e-05,
+      "loss": 1.0587,
+      "step": 2130
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.978230911349374e-05,
+      "loss": 1.0521,
+      "step": 2131
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.9750545414413403e-05,
+      "loss": 0.9899,
+      "step": 2132
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.9718777624185327e-05,
+      "loss": 0.986,
+      "step": 2133
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.96870057561387e-05,
+      "loss": 1.1204,
+      "step": 2134
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.9655229823604406e-05,
+      "loss": 0.9447,
+      "step": 2135
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.962344983991503e-05,
+      "loss": 1.0577,
+      "step": 2136
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.959166581840484e-05,
+      "loss": 0.9506,
+      "step": 2137
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.9559877772409855e-05,
+      "loss": 1.013,
+      "step": 2138
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.9528085715267746e-05,
+      "loss": 0.8285,
+      "step": 2139
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.949628966031785e-05,
+      "loss": 1.066,
+      "step": 2140
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.946448962090121e-05,
+      "loss": 1.0296,
+      "step": 2141
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.943268561036053e-05,
+      "loss": 1.1153,
+      "step": 2142
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.94008776420402e-05,
+      "loss": 1.2563,
+      "step": 2143
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.9369065729286245e-05,
+      "loss": 0.9768,
+      "step": 2144
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.9337249885446324e-05,
+      "loss": 1.1626,
+      "step": 2145
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.930543012386981e-05,
+      "loss": 1.0542,
+      "step": 2146
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.927360645790765e-05,
+      "loss": 1.0801,
+      "step": 2147
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.92417789009125e-05,
+      "loss": 0.953,
+      "step": 2148
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.9209947466238605e-05,
+      "loss": 1.235,
+      "step": 2149
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.9178112167241805e-05,
+      "loss": 0.7847,
+      "step": 2150
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.9146273017279654e-05,
+      "loss": 1.0142,
+      "step": 2151
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.911443002971122e-05,
+      "loss": 0.9631,
+      "step": 2152
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.9082583217897284e-05,
+      "loss": 1.2026,
+      "step": 2153
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.9050732595200106e-05,
+      "loss": 0.8694,
+      "step": 2154
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.901887817498367e-05,
+      "loss": 0.9415,
+      "step": 2155
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.898701997061349e-05,
+      "loss": 0.8743,
+      "step": 2156
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.895515799545666e-05,
+      "loss": 0.8093,
+      "step": 2157
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.892329226288189e-05,
+      "loss": 1.281,
+      "step": 2158
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.8891422786259435e-05,
+      "loss": 0.976,
+      "step": 2159
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.885954957896115e-05,
+      "loss": 0.9168,
+      "step": 2160
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.882767265436042e-05,
+      "loss": 1.1541,
+      "step": 2161
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.879579202583223e-05,
+      "loss": 0.9527,
+      "step": 2162
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.8763907706753084e-05,
+      "loss": 0.8176,
+      "step": 2163
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.873201971050107e-05,
+      "loss": 0.9299,
+      "step": 2164
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.8700128050455795e-05,
+      "loss": 0.9456,
+      "step": 2165
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.8668232739998384e-05,
+      "loss": 0.9079,
+      "step": 2166
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.863633379251155e-05,
+      "loss": 1.0385,
+      "step": 2167
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.860443122137946e-05,
+      "loss": 0.9485,
+      "step": 2168
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.857252503998788e-05,
+      "loss": 1.0627,
+      "step": 2169
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.854061526172402e-05,
+      "loss": 0.9716,
+      "step": 2170
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.850870189997665e-05,
+      "loss": 0.9839,
+      "step": 2171
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.847678496813601e-05,
+      "loss": 0.9705,
+      "step": 2172
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.844486447959388e-05,
+      "loss": 1.1802,
+      "step": 2173
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.841294044774346e-05,
+      "loss": 0.9474,
+      "step": 2174
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.8381012885979504e-05,
+      "loss": 1.05,
+      "step": 2175
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.8349081807698236e-05,
+      "loss": 1.274,
+      "step": 2176
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.831714722629732e-05,
+      "loss": 1.068,
+      "step": 2177
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.828520915517592e-05,
+      "loss": 0.6944,
+      "step": 2178
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.8253267607734684e-05,
+      "loss": 1.2252,
+      "step": 2179
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.822132259737565e-05,
+      "loss": 1.1181,
+      "step": 2180
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.818937413750236e-05,
+      "loss": 1.419,
+      "step": 2181
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.8157422241519824e-05,
+      "loss": 0.973,
+      "step": 2182
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.812546692283444e-05,
+      "loss": 0.8865,
+      "step": 2183
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.809350819485407e-05,
+      "loss": 1.0995,
+      "step": 2184
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.8061546070987994e-05,
+      "loss": 1.0659,
+      "step": 2185
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.802958056464694e-05,
+      "loss": 1.1365,
+      "step": 2186
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.799761168924305e-05,
+      "loss": 0.9673,
+      "step": 2187
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.7965639458189835e-05,
+      "loss": 0.9529,
+      "step": 2188
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.7933663884902275e-05,
+      "loss": 1.0578,
+      "step": 2189
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.7901684982796714e-05,
+      "loss": 1.2926,
+      "step": 2190
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.786970276529093e-05,
+      "loss": 1.1744,
+      "step": 2191
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.7837717245804045e-05,
+      "loss": 0.7069,
+      "step": 2192
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.78057284377566e-05,
+      "loss": 1.1554,
+      "step": 2193
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.777373635457048e-05,
+      "loss": 1.1782,
+      "step": 2194
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.774174100966899e-05,
+      "loss": 1.0752,
+      "step": 2195
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.7709742416476785e-05,
+      "loss": 0.9599,
+      "step": 2196
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.7677740588419885e-05,
+      "loss": 0.734,
+      "step": 2197
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.764573553892564e-05,
+      "loss": 0.9601,
+      "step": 2198
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.761372728142278e-05,
+      "loss": 0.8812,
+      "step": 2199
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.7581715829341396e-05,
+      "loss": 1.2377,
+      "step": 2200
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.754970119611289e-05,
+      "loss": 1.2632,
+      "step": 2201
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.7517683395169994e-05,
+      "loss": 0.8723,
+      "step": 2202
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.7485662439946785e-05,
+      "loss": 1.0657,
+      "step": 2203
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.7453638343878665e-05,
+      "loss": 0.9452,
+      "step": 2204
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.742161112040236e-05,
+      "loss": 0.929,
+      "step": 2205
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.7389580782955896e-05,
+      "loss": 1.2247,
+      "step": 2206
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.735754734497859e-05,
+      "loss": 1.2153,
+      "step": 2207
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.732551081991109e-05,
+      "loss": 1.1432,
+      "step": 2208
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.7293471221195326e-05,
+      "loss": 0.891,
+      "step": 2209
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.726142856227452e-05,
+      "loss": 1.0079,
+      "step": 2210
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.7229382856593187e-05,
+      "loss": 1.2301,
+      "step": 2211
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.719733411759707e-05,
+      "loss": 1.0544,
+      "step": 2212
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.716528235873326e-05,
+      "loss": 1.0224,
+      "step": 2213
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.713322759345008e-05,
+      "loss": 1.3068,
+      "step": 2214
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.710116983519711e-05,
+      "loss": 1.108,
+      "step": 2215
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.706910909742518e-05,
+      "loss": 1.1158,
+      "step": 2216
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.7037045393586374e-05,
+      "loss": 1.0657,
+      "step": 2217
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.7004978737134055e-05,
+      "loss": 1.1184,
+      "step": 2218
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.697290914152277e-05,
+      "loss": 1.2764,
+      "step": 2219
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.6940836620208346e-05,
+      "loss": 1.1916,
+      "step": 2220
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.690876118664778e-05,
+      "loss": 1.0452,
+      "step": 2221
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.687668285429938e-05,
+      "loss": 0.9442,
+      "step": 2222
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.684460163662259e-05,
+      "loss": 1.1398,
+      "step": 2223
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.6812517547078094e-05,
+      "loss": 0.9965,
+      "step": 2224
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.678043059912777e-05,
+      "loss": 0.8657,
+      "step": 2225
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.674834080623472e-05,
+      "loss": 0.9936,
+      "step": 2226
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.671624818186323e-05,
+      "loss": 0.673,
+      "step": 2227
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.668415273947876e-05,
+      "loss": 1.3878,
+      "step": 2228
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.6652054492547965e-05,
+      "loss": 1.0554,
+      "step": 2229
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.6619953454538664e-05,
+      "loss": 1.2773,
+      "step": 2230
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.6587849638919854e-05,
+      "loss": 1.0511,
+      "step": 2231
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.655574305916172e-05,
+      "loss": 1.1802,
+      "step": 2232
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.652363372873559e-05,
+      "loss": 0.9121,
+      "step": 2233
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.64915216611139e-05,
+      "loss": 1.0013,
+      "step": 2234
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.645940686977033e-05,
+      "loss": 0.9847,
+      "step": 2235
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.6427289368179605e-05,
+      "loss": 1.0125,
+      "step": 2236
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.6395169169817665e-05,
+      "loss": 1.2868,
+      "step": 2237
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.6363046288161525e-05,
+      "loss": 1.0541,
+      "step": 2238
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.633092073668936e-05,
+      "loss": 1.1426,
+      "step": 2239
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.629879252888046e-05,
+      "loss": 1.0363,
+      "step": 2240
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.6266661678215216e-05,
+      "loss": 0.9409,
+      "step": 2241
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.6234528198175144e-05,
+      "loss": 1.0181,
+      "step": 2242
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.620239210224284e-05,
+      "loss": 1.1209,
+      "step": 2243
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.617025340390203e-05,
+      "loss": 0.9037,
+      "step": 2244
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.613811211663751e-05,
+      "loss": 0.996,
+      "step": 2245
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.610596825393516e-05,
+      "loss": 1.0058,
+      "step": 2246
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.607382182928196e-05,
+      "loss": 0.9988,
+      "step": 2247
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.604167285616593e-05,
+      "loss": 0.9486,
+      "step": 2248
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.600952134807621e-05,
+      "loss": 0.8843,
+      "step": 2249
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.597736731850295e-05,
+      "loss": 1.1879,
+      "step": 2250
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.594521078093741e-05,
+      "loss": 0.9918,
+      "step": 2251
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 5.591305174887185e-05,
+      "loss": 1.0278,
+      "step": 2252
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 5.5880890235799636e-05,
+      "loss": 1.0731,
+      "step": 2253
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 5.5848726255215104e-05,
+      "loss": 0.8735,
+      "step": 2254
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 5.581655982061367e-05,
+      "loss": 0.8486,
+      "step": 2255
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 5.578439094549178e-05,
+      "loss": 0.9558,
+      "step": 2256
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 5.575221964334692e-05,
+      "loss": 1.2244,
+      "step": 2257
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 5.572004592767754e-05,
+      "loss": 1.0885,
+      "step": 2258
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 5.568786981198315e-05,
+      "loss": 0.9424,
+      "step": 2259
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 5.565569130976422e-05,
+      "loss": 0.9583,
+      "step": 2260
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 5.562351043452227e-05,
+      "loss": 0.961,
+      "step": 2261
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 5.559132719975984e-05,
+      "loss": 0.9632,
+      "step": 2262
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 5.555914161898035e-05,
+      "loss": 1.0893,
+      "step": 2263
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 5.55269537056883e-05,
+      "loss": 0.8318,
+      "step": 2264
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 5.5494763473389145e-05,
+      "loss": 1.1214,
+      "step": 2265
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 5.546257093558932e-05,
+      "loss": 1.0031,
+      "step": 2266
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 5.543037610579619e-05,
+      "loss": 1.0347,
+      "step": 2267
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 5.539817899751812e-05,
+      "loss": 1.1812,
+      "step": 2268
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 5.536597962426441e-05,
+      "loss": 1.1635,
+      "step": 2269
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 5.533377799954532e-05,
+      "loss": 0.9454,
+      "step": 2270
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 5.530157413687208e-05,
+      "loss": 0.9831,
+      "step": 2271
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 5.526936804975681e-05,
+      "loss": 1.0144,
+      "step": 2272
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 5.5237159751712584e-05,
+      "loss": 1.1738,
+      "step": 2273
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 5.520494925625339e-05,
+      "loss": 1.3576,
+      "step": 2274
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 5.517273657689419e-05,
+      "loss": 1.2042,
+      "step": 2275
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 5.5140521727150805e-05,
+      "loss": 1.0726,
+      "step": 2276
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 5.510830472054001e-05,
+      "loss": 1.0942,
+      "step": 2277
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 5.5076085570579416e-05,
+      "loss": 1.1613,
+      "step": 2278
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 5.5043864290787615e-05,
+      "loss": 1.398,
+      "step": 2279
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 5.5011640894684056e-05,
+      "loss": 1.2372,
+      "step": 2280
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 5.497941539578908e-05,
+      "loss": 1.3297,
+      "step": 2281
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 5.494718780762388e-05,
+      "loss": 0.8617,
+      "step": 2282
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 5.491495814371058e-05,
+      "loss": 0.9996,
+      "step": 2283
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 5.488272641757215e-05,
+      "loss": 0.9369,
+      "step": 2284
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 5.4850492642732406e-05,
+      "loss": 1.1553,
+      "step": 2285
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 5.481825683271607e-05,
+      "loss": 1.0319,
+      "step": 2286
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 5.478601900104864e-05,
+      "loss": 1.0369,
+      "step": 2287
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 5.475377916125655e-05,
+      "loss": 0.8311,
+      "step": 2288
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 5.472153732686702e-05,
+      "loss": 0.9436,
+      "step": 2289
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 5.4689293511408155e-05,
+      "loss": 0.9344,
+      "step": 2290
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 5.465704772840882e-05,
+      "loss": 1.1732,
+      "step": 2291
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 5.4624799991398766e-05,
+      "loss": 1.1709,
+      "step": 2292
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 5.459255031390855e-05,
+      "loss": 1.2112,
+      "step": 2293
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 5.456029870946954e-05,
+      "loss": 0.8844,
+      "step": 2294
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 5.4528045191613896e-05,
+      "loss": 0.8973,
+      "step": 2295
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 5.4495789773874594e-05,
+      "loss": 0.9388,
+      "step": 2296
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 5.446353246978542e-05,
+      "loss": 0.8854,
+      "step": 2297
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 5.443127329288092e-05,
+      "loss": 0.9479,
+      "step": 2298
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 5.439901225669649e-05,
+      "loss": 1.4247,
+      "step": 2299
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 5.43667493747682e-05,
+      "loss": 1.1235,
+      "step": 2300
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 5.433448466063301e-05,
+      "loss": 1.0211,
+      "step": 2301
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 5.430221812782855e-05,
+      "loss": 0.7795,
+      "step": 2302
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 5.4269949789893305e-05,
+      "loss": 1.0832,
+      "step": 2303
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 5.4237679660366435e-05,
+      "loss": 1.0783,
+      "step": 2304
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 5.420540775278789e-05,
+      "loss": 1.3059,
+      "step": 2305
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 5.4173134080698385e-05,
+      "loss": 0.7759,
+      "step": 2306
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 5.4140858657639315e-05,
+      "loss": 0.8972,
+      "step": 2307
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 5.410858149715289e-05,
+      "loss": 1.1036,
+      "step": 2308
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 5.407630261278197e-05,
+      "loss": 0.9387,
+      "step": 2309
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 5.4044022018070214e-05,
+      "loss": 0.8905,
+      "step": 2310
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 5.401173972656193e-05,
+      "loss": 1.042,
+      "step": 2311
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 5.3979455751802175e-05,
+      "loss": 1.0472,
+      "step": 2312
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 5.394717010733672e-05,
+      "loss": 0.8218,
+      "step": 2313
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 5.391488280671199e-05,
+      "loss": 1.4798,
+      "step": 2314
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 5.388259386347517e-05,
+      "loss": 0.9367,
+      "step": 2315
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 5.385030329117408e-05,
+      "loss": 1.0795,
+      "step": 2316
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.3818011103357246e-05,
+      "loss": 0.9519,
+      "step": 2317
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.378571731357388e-05,
+      "loss": 1.2071,
+      "step": 2318
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.375342193537385e-05,
+      "loss": 0.9891,
+      "step": 2319
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.37211249823077e-05,
+      "loss": 0.849,
+      "step": 2320
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.3688826467926626e-05,
+      "loss": 0.9098,
+      "step": 2321
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.365652640578249e-05,
+      "loss": 1.0196,
+      "step": 2322
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.3624224809427794e-05,
+      "loss": 0.87,
+      "step": 2323
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.3591921692415704e-05,
+      "loss": 0.977,
+      "step": 2324
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.3559617068299974e-05,
+      "loss": 1.0277,
+      "step": 2325
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.352731095063506e-05,
+      "loss": 1.031,
+      "step": 2326
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.349500335297597e-05,
+      "loss": 1.0833,
+      "step": 2327
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.346269428887842e-05,
+      "loss": 0.8167,
+      "step": 2328
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.343038377189866e-05,
+      "loss": 1.008,
+      "step": 2329
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.3398071815593585e-05,
+      "loss": 0.9416,
+      "step": 2330
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.336575843352071e-05,
+      "loss": 0.995,
+      "step": 2331
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.3333443639238104e-05,
+      "loss": 1.002,
+      "step": 2332
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.330112744630447e-05,
+      "loss": 1.1715,
+      "step": 2333
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.3268809868279094e-05,
+      "loss": 0.7859,
+      "step": 2334
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.3236490918721794e-05,
+      "loss": 1.0159,
+      "step": 2335
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.320417061119303e-05,
+      "loss": 1.2871,
+      "step": 2336
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.31718489592538e-05,
+      "loss": 1.058,
+      "step": 2337
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.313952597646568e-05,
+      "loss": 1.1358,
+      "step": 2338
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.310720167639076e-05,
+      "loss": 1.0783,
+      "step": 2339
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.307487607259175e-05,
+      "loss": 0.9496,
+      "step": 2340
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.304254917863184e-05,
+      "loss": 0.9201,
+      "step": 2341
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.3010221008074826e-05,
+      "loss": 0.7768,
+      "step": 2342
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.2977891574484994e-05,
+      "loss": 0.7972,
+      "step": 2343
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.294556089142716e-05,
+      "loss": 0.9322,
+      "step": 2344
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.2913228972466685e-05,
+      "loss": 1.1433,
+      "step": 2345
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.2880895831169474e-05,
+      "loss": 0.9315,
+      "step": 2346
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.2848561481101874e-05,
+      "loss": 0.8807,
+      "step": 2347
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.28162259358308e-05,
+      "loss": 0.9244,
+      "step": 2348
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.278388920892362e-05,
+      "loss": 1.3168,
+      "step": 2349
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.2751551313948246e-05,
+      "loss": 1.2252,
+      "step": 2350
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.271921226447305e-05,
+      "loss": 1.2448,
+      "step": 2351
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.2686872074066914e-05,
+      "loss": 1.239,
+      "step": 2352
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.265453075629915e-05,
+      "loss": 1.052,
+      "step": 2353
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.2622188324739584e-05,
+      "loss": 1.2043,
+      "step": 2354
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.258984479295852e-05,
+      "loss": 1.006,
+      "step": 2355
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.25575001745267e-05,
+      "loss": 0.782,
+      "step": 2356
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.252515448301529e-05,
+      "loss": 1.0688,
+      "step": 2357
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.2492807731995965e-05,
+      "loss": 0.9452,
+      "step": 2358
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.246045993504083e-05,
+      "loss": 1.0262,
+      "step": 2359
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.242811110572242e-05,
+      "loss": 1.1952,
+      "step": 2360
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.2395761257613684e-05,
+      "loss": 0.8887,
+      "step": 2361
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.236341040428803e-05,
+      "loss": 1.0041,
+      "step": 2362
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.2331058559319256e-05,
+      "loss": 1.0233,
+      "step": 2363
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.229870573628163e-05,
+      "loss": 0.8465,
+      "step": 2364
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.2266351948749784e-05,
+      "loss": 0.9831,
+      "step": 2365
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.2233997210298745e-05,
+      "loss": 1.0394,
+      "step": 2366
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.220164153450395e-05,
+      "loss": 0.8489,
+      "step": 2367
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.21692849349413e-05,
+      "loss": 0.9297,
+      "step": 2368
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.2136927425186985e-05,
+      "loss": 0.9187,
+      "step": 2369
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.210456901881761e-05,
+      "loss": 0.9708,
+      "step": 2370
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.207220972941016e-05,
+      "loss": 0.9978,
+      "step": 2371
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.2039849570541996e-05,
+      "loss": 0.8685,
+      "step": 2372
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.200748855579086e-05,
+      "loss": 1.0287,
+      "step": 2373
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.197512669873482e-05,
+      "loss": 1.0544,
+      "step": 2374
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.194276401295231e-05,
+      "loss": 1.0128,
+      "step": 2375
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.191040051202208e-05,
+      "loss": 0.8994,
+      "step": 2376
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.1878036209523295e-05,
+      "loss": 1.3186,
+      "step": 2377
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.184567111903541e-05,
+      "loss": 0.8332,
+      "step": 2378
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.181330525413821e-05,
+      "loss": 1.0021,
+      "step": 2379
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.1780938628411794e-05,
+      "loss": 1.0491,
+      "step": 2380
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.174857125543662e-05,
+      "loss": 1.2825,
+      "step": 2381
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.171620314879342e-05,
+      "loss": 1.0651,
+      "step": 2382
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.168383432206327e-05,
+      "loss": 0.9851,
+      "step": 2383
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.1651464788827506e-05,
+      "loss": 1.2986,
+      "step": 2384
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.1619094562667804e-05,
+      "loss": 0.8112,
+      "step": 2385
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.158672365716608e-05,
+      "loss": 0.9047,
+      "step": 2386
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.155435208590461e-05,
+      "loss": 1.0125,
+      "step": 2387
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.152197986246586e-05,
+      "loss": 0.9906,
+      "step": 2388
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.148960700043264e-05,
+      "loss": 0.9123,
+      "step": 2389
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.1457233513387994e-05,
+      "loss": 0.9956,
+      "step": 2390
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.142485941491522e-05,
+      "loss": 1.2164,
+      "step": 2391
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.139248471859792e-05,
+      "loss": 1.0526,
+      "step": 2392
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.136010943801989e-05,
+      "loss": 1.0412,
+      "step": 2393
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.1327733586765205e-05,
+      "loss": 0.9778,
+      "step": 2394
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.129535717841818e-05,
+      "loss": 1.132,
+      "step": 2395
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.126298022656333e-05,
+      "loss": 0.7555,
+      "step": 2396
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.123060274478545e-05,
+      "loss": 0.8566,
+      "step": 2397
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.11982247466695e-05,
+      "loss": 0.9214,
+      "step": 2398
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.116584624580072e-05,
+      "loss": 1.1422,
+      "step": 2399
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.11334672557645e-05,
+      "loss": 1.0209,
+      "step": 2400
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.110108779014649e-05,
+      "loss": 1.0402,
+      "step": 2401
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.1068707862532485e-05,
+      "loss": 0.9373,
+      "step": 2402
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.103632748650852e-05,
+      "loss": 0.8724,
+      "step": 2403
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.100394667566079e-05,
+      "loss": 1.0661,
+      "step": 2404
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.097156544357567e-05,
+      "loss": 0.937,
+      "step": 2405
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.0939183803839764e-05,
+      "loss": 1.0883,
+      "step": 2406
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.090680177003976e-05,
+      "loss": 0.6866,
+      "step": 2407
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.087441935576258e-05,
+      "loss": 1.12,
+      "step": 2408
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.084203657459529e-05,
+      "loss": 0.8331,
+      "step": 2409
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.080965344012508e-05,
+      "loss": 1.161,
+      "step": 2410
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.077726996593933e-05,
+      "loss": 0.9043,
+      "step": 2411
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.074488616562555e-05,
+      "loss": 0.926,
+      "step": 2412
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.071250205277137e-05,
+      "loss": 0.9766,
+      "step": 2413
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.068011764096455e-05,
+      "loss": 0.9738,
+      "step": 2414
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.0647732943793025e-05,
+      "loss": 1.122,
+      "step": 2415
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.061534797484476e-05,
+      "loss": 1.0445,
+      "step": 2416
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.058296274770794e-05,
+      "loss": 1.303,
+      "step": 2417
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.055057727597078e-05,
+      "loss": 0.9487,
+      "step": 2418
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.051819157322161e-05,
+      "loss": 0.9254,
+      "step": 2419
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.048580565304887e-05,
+      "loss": 1.0798,
+      "step": 2420
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.0453419529041125e-05,
+      "loss": 0.9673,
+      "step": 2421
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.042103321478696e-05,
+      "loss": 1.3701,
+      "step": 2422
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.038864672387509e-05,
+      "loss": 0.9734,
+      "step": 2423
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.035626006989425e-05,
+      "loss": 0.8036,
+      "step": 2424
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.0323873266433306e-05,
+      "loss": 1.4521,
+      "step": 2425
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.029148632708117e-05,
+      "loss": 1.0397,
+      "step": 2426
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.025909926542679e-05,
+      "loss": 1.0536,
+      "step": 2427
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.022671209505916e-05,
+      "loss": 1.0426,
+      "step": 2428
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.019432482956734e-05,
+      "loss": 1.0238,
+      "step": 2429
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.0161937482540446e-05,
+      "loss": 1.151,
+      "step": 2430
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.0129550067567586e-05,
+      "loss": 0.991,
+      "step": 2431
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.0097162598237925e-05,
+      "loss": 1.2062,
+      "step": 2432
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.0064775088140636e-05,
+      "loss": 1.0617,
+      "step": 2433
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.0032387550864926e-05,
+      "loss": 1.0333,
+      "step": 2434
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5e-05,
+      "loss": 0.7442,
+      "step": 2435
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.996761244913508e-05,
+      "loss": 0.7844,
+      "step": 2436
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.9935224911859375e-05,
+      "loss": 0.952,
+      "step": 2437
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.990283740176208e-05,
+      "loss": 0.9996,
+      "step": 2438
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.987044993243244e-05,
+      "loss": 0.8146,
+      "step": 2439
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.983806251745957e-05,
+      "loss": 0.9054,
+      "step": 2440
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.9805675170432666e-05,
+      "loss": 1.01,
+      "step": 2441
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.977328790494086e-05,
+      "loss": 1.0729,
+      "step": 2442
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.974090073457322e-05,
+      "loss": 0.9991,
+      "step": 2443
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.970851367291885e-05,
+      "loss": 1.1022,
+      "step": 2444
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.9676126733566705e-05,
+      "loss": 0.8794,
+      "step": 2445
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.964373993010576e-05,
+      "loss": 1.1476,
+      "step": 2446
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.961135327612494e-05,
+      "loss": 1.089,
+      "step": 2447
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.957896678521305e-05,
+      "loss": 1.2671,
+      "step": 2448
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.95465804709589e-05,
+      "loss": 0.8109,
+      "step": 2449
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.9514194346951144e-05,
+      "loss": 0.9909,
+      "step": 2450
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.948180842677841e-05,
+      "loss": 0.8925,
+      "step": 2451
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.944942272402925e-05,
+      "loss": 1.2115,
+      "step": 2452
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.941703725229207e-05,
+      "loss": 1.0232,
+      "step": 2453
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.9384652025155236e-05,
+      "loss": 1.1159,
+      "step": 2454
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.9352267056206994e-05,
+      "loss": 0.9132,
+      "step": 2455
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.9319882359035446e-05,
+      "loss": 0.8885,
+      "step": 2456
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.928749794722865e-05,
+      "loss": 1.1686,
+      "step": 2457
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.9255113834374464e-05,
+      "loss": 1.0252,
+      "step": 2458
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.922273003406067e-05,
+      "loss": 0.9102,
+      "step": 2459
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.919034655987493e-05,
+      "loss": 0.9062,
+      "step": 2460
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.9157963425404725e-05,
+      "loss": 1.2012,
+      "step": 2461
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.912558064423744e-05,
+      "loss": 0.837,
+      "step": 2462
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.909319822996026e-05,
+      "loss": 0.7328,
+      "step": 2463
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.9060816196160254e-05,
+      "loss": 1.1989,
+      "step": 2464
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.902843455642434e-05,
+      "loss": 0.8209,
+      "step": 2465
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.899605332433922e-05,
+      "loss": 0.8433,
+      "step": 2466
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.89636725134915e-05,
+      "loss": 0.8688,
+      "step": 2467
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.8931292137467527e-05,
+      "loss": 1.134,
+      "step": 2468
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.889891220985351e-05,
+      "loss": 1.1648,
+      "step": 2469
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.88665327442355e-05,
+      "loss": 1.061,
+      "step": 2470
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.883415375419929e-05,
+      "loss": 1.0664,
+      "step": 2471
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.880177525333051e-05,
+      "loss": 0.9933,
+      "step": 2472
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.8769397255214564e-05,
+      "loss": 0.9785,
+      "step": 2473
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.873701977343667e-05,
+      "loss": 1.0179,
+      "step": 2474
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.870464282158184e-05,
+      "loss": 1.1873,
+      "step": 2475
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.867226641323481e-05,
+      "loss": 0.8953,
+      "step": 2476
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.863989056198013e-05,
+      "loss": 0.9771,
+      "step": 2477
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.860751528140209e-05,
+      "loss": 0.9102,
+      "step": 2478
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.857514058508478e-05,
+      "loss": 0.9156,
+      "step": 2479
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.854276648661203e-05,
+      "loss": 1.0701,
+      "step": 2480
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.851039299956737e-05,
+      "loss": 1.1675,
+      "step": 2481
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.847802013753414e-05,
+      "loss": 1.0337,
+      "step": 2482
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.84456479140954e-05,
+      "loss": 1.0585,
+      "step": 2483
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.8413276342833916e-05,
+      "loss": 1.0247,
+      "step": 2484
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.838090543733222e-05,
+      "loss": 1.1254,
+      "step": 2485
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.8348535211172505e-05,
+      "loss": 0.8266,
+      "step": 2486
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.831616567793673e-05,
+      "loss": 0.9389,
+      "step": 2487
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.828379685120659e-05,
+      "loss": 1.0788,
+      "step": 2488
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.8251428744563394e-05,
+      "loss": 1.0959,
+      "step": 2489
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.821906137158822e-05,
+      "loss": 0.936,
+      "step": 2490
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.81866947458618e-05,
+      "loss": 1.0485,
+      "step": 2491
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.815432888096459e-05,
+      "loss": 0.8398,
+      "step": 2492
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.8121963790476716e-05,
+      "loss": 1.352,
+      "step": 2493
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.808959948797793e-05,
+      "loss": 0.9432,
+      "step": 2494
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.805723598704772e-05,
+      "loss": 0.983,
+      "step": 2495
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.8024873301265185e-05,
+      "loss": 0.9668,
+      "step": 2496
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.799251144420913e-05,
+      "loss": 1.2321,
+      "step": 2497
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.796015042945801e-05,
+      "loss": 0.7766,
+      "step": 2498
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.792779027058985e-05,
+      "loss": 0.941,
+      "step": 2499
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.789543098118241e-05,
+      "loss": 1.0049,
+      "step": 2500
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.786307257481303e-05,
+      "loss": 1.0374,
+      "step": 2501
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.78307150650587e-05,
+      "loss": 1.0219,
+      "step": 2502
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.7798358465496055e-05,
+      "loss": 0.9021,
+      "step": 2503
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.776600278970127e-05,
+      "loss": 1.0242,
+      "step": 2504
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.773364805125025e-05,
+      "loss": 0.9642,
+      "step": 2505
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.770129426371838e-05,
+      "loss": 1.1935,
+      "step": 2506
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.766894144068074e-05,
+      "loss": 0.9951,
+      "step": 2507
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.763658959571199e-05,
+      "loss": 0.9971,
+      "step": 2508
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.760423874238633e-05,
+      "loss": 1.1111,
+      "step": 2509
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.7571888894277604e-05,
+      "loss": 1.1699,
+      "step": 2510
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.753954006495918e-05,
+      "loss": 1.0258,
+      "step": 2511
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.750719226800403e-05,
+      "loss": 1.0544,
+      "step": 2512
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.747484551698473e-05,
+      "loss": 1.0652,
+      "step": 2513
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.744249982547332e-05,
+      "loss": 1.0185,
+      "step": 2514
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.741015520704148e-05,
+      "loss": 0.9599,
+      "step": 2515
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.737781167526043e-05,
+      "loss": 1.0214,
+      "step": 2516
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.7345469243700863e-05,
+      "loss": 1.0392,
+      "step": 2517
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.731312792593311e-05,
+      "loss": 1.0073,
+      "step": 2518
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.7280787735526956e-05,
+      "loss": 1.228,
+      "step": 2519
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.724844868605176e-05,
+      "loss": 1.2696,
+      "step": 2520
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.721611079107639e-05,
+      "loss": 0.9742,
+      "step": 2521
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.718377406416921e-05,
+      "loss": 1.1313,
+      "step": 2522
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.7151438518898145e-05,
+      "loss": 0.853,
+      "step": 2523
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.711910416883054e-05,
+      "loss": 1.02,
+      "step": 2524
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.708677102753331e-05,
+      "loss": 1.0701,
+      "step": 2525
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.7054439108572854e-05,
+      "loss": 1.0198,
+      "step": 2526
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.7022108425515024e-05,
+      "loss": 1.1708,
+      "step": 2527
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.69897789919252e-05,
+      "loss": 0.8504,
+      "step": 2528
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.695745082136817e-05,
+      "loss": 1.046,
+      "step": 2529
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.692512392740827e-05,
+      "loss": 1.2396,
+      "step": 2530
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.689279832360925e-05,
+      "loss": 1.2745,
+      "step": 2531
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.6860474023534335e-05,
+      "loss": 0.8293,
+      "step": 2532
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.6828151040746214e-05,
+      "loss": 1.1401,
+      "step": 2533
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.6795829388806984e-05,
+      "loss": 0.7452,
+      "step": 2534
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.676350908127822e-05,
+      "loss": 0.8046,
+      "step": 2535
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.673119013172093e-05,
+      "loss": 1.2207,
+      "step": 2536
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.6698872553695535e-05,
+      "loss": 1.0086,
+      "step": 2537
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.666655636076193e-05,
+      "loss": 1.1429,
+      "step": 2538
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.6634241566479305e-05,
+      "loss": 1.0025,
+      "step": 2539
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.660192818440641e-05,
+      "loss": 1.1426,
+      "step": 2540
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.656961622810136e-05,
+      "loss": 1.15,
+      "step": 2541
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.653730571112159e-05,
+      "loss": 0.9654,
+      "step": 2542
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.6504996647024025e-05,
+      "loss": 0.9159,
+      "step": 2543
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.6472689049364954e-05,
+      "loss": 1.0335,
+      "step": 2544
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.644038293170003e-05,
+      "loss": 1.0088,
+      "step": 2545
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.640807830758432e-05,
+      "loss": 0.9027,
+      "step": 2546
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.637577519057221e-05,
+      "loss": 0.7735,
+      "step": 2547
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.634347359421751e-05,
+      "loss": 0.8535,
+      "step": 2548
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.631117353207338e-05,
+      "loss": 1.2806,
+      "step": 2549
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.627887501769231e-05,
+      "loss": 1.1618,
+      "step": 2550
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.624657806462617e-05,
+      "loss": 0.9432,
+      "step": 2551
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.6214282686426135e-05,
+      "loss": 1.0044,
+      "step": 2552
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.6181988896642766e-05,
+      "loss": 1.1269,
+      "step": 2553
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.614969670882594e-05,
+      "loss": 1.0606,
+      "step": 2554
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.611740613652484e-05,
+      "loss": 0.8635,
+      "step": 2555
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.608511719328802e-05,
+      "loss": 1.033,
+      "step": 2556
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.6052829892663296e-05,
+      "loss": 1.0018,
+      "step": 2557
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.602054424819782e-05,
+      "loss": 0.9161,
+      "step": 2558
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.598826027343809e-05,
+      "loss": 0.9595,
+      "step": 2559
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.59559779819298e-05,
+      "loss": 1.152,
+      "step": 2560
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.5923697387218036e-05,
+      "loss": 1.1055,
+      "step": 2561
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.589141850284712e-05,
+      "loss": 1.0313,
+      "step": 2562
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.585914134236068e-05,
+      "loss": 1.0387,
+      "step": 2563
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.582686591930164e-05,
+      "loss": 1.0771,
+      "step": 2564
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.579459224721212e-05,
+      "loss": 1.4499,
+      "step": 2565
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.576232033963358e-05,
+      "loss": 0.9623,
+      "step": 2566
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.5730050210106707e-05,
+      "loss": 1.0098,
+      "step": 2567
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.569778187217144e-05,
+      "loss": 0.997,
+      "step": 2568
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.566551533936701e-05,
+      "loss": 0.8912,
+      "step": 2569
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.5633250625231806e-05,
+      "loss": 1.0987,
+      "step": 2570
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.5600987743303534e-05,
+      "loss": 1.1296,
+      "step": 2571
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.5568726707119084e-05,
+      "loss": 1.0279,
+      "step": 2572
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.553646753021459e-05,
+      "loss": 1.0285,
+      "step": 2573
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.550421022612542e-05,
+      "loss": 0.8252,
+      "step": 2574
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.5471954808386115e-05,
+      "loss": 1.204,
+      "step": 2575
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.5439701290530465e-05,
+      "loss": 0.9056,
+      "step": 2576
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.540744968609146e-05,
+      "loss": 1.0851,
+      "step": 2577
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.537520000860124e-05,
+      "loss": 1.0119,
+      "step": 2578
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.534295227159119e-05,
+      "loss": 0.9178,
+      "step": 2579
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.5310706488591856e-05,
+      "loss": 0.8572,
+      "step": 2580
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.527846267313297e-05,
+      "loss": 1.062,
+      "step": 2581
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.524622083874347e-05,
+      "loss": 1.1814,
+      "step": 2582
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.521398099895137e-05,
+      "loss": 1.1199,
+      "step": 2583
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.518174316728396e-05,
+      "loss": 0.9779,
+      "step": 2584
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.51495073572676e-05,
+      "loss": 0.9973,
+      "step": 2585
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.511727358242785e-05,
+      "loss": 0.9824,
+      "step": 2586
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.508504185628944e-05,
+      "loss": 1.076,
+      "step": 2587
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.505281219237613e-05,
+      "loss": 0.8061,
+      "step": 2588
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.502058460421094e-05,
+      "loss": 1.043,
+      "step": 2589
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.4988359105315956e-05,
+      "loss": 1.0717,
+      "step": 2590
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.495613570921239e-05,
+      "loss": 1.3573,
+      "step": 2591
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.4923914429420595e-05,
+      "loss": 0.7881,
+      "step": 2592
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.4891695279460004e-05,
+      "loss": 1.0007,
+      "step": 2593
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.485947827284921e-05,
+      "loss": 1.039,
+      "step": 2594
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.482726342310582e-05,
+      "loss": 1.0969,
+      "step": 2595
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.4795050743746614e-05,
+      "loss": 1.1014,
+      "step": 2596
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.476284024828744e-05,
+      "loss": 1.1211,
+      "step": 2597
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.47306319502432e-05,
+      "loss": 1.019,
+      "step": 2598
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.469842586312794e-05,
+      "loss": 1.0055,
+      "step": 2599
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.4666222000454683e-05,
+      "loss": 0.9934,
+      "step": 2600
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.46340203757356e-05,
+      "loss": 1.1093,
+      "step": 2601
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.4601821002481896e-05,
+      "loss": 1.0057,
+      "step": 2602
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.456962389420382e-05,
+      "loss": 0.8975,
+      "step": 2603
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.453742906441069e-05,
+      "loss": 1.0177,
+      "step": 2604
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.450523652661086e-05,
+      "loss": 1.0161,
+      "step": 2605
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.447304629431169e-05,
+      "loss": 0.818,
+      "step": 2606
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.444085838101967e-05,
+      "loss": 1.0112,
+      "step": 2607
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.440867280024018e-05,
+      "loss": 0.9273,
+      "step": 2608
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.437648956547772e-05,
+      "loss": 1.0964,
+      "step": 2609
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.434430869023579e-05,
+      "loss": 0.894,
+      "step": 2610
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.4312130188016866e-05,
+      "loss": 1.1985,
+      "step": 2611
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.427995407232248e-05,
+      "loss": 1.2747,
+      "step": 2612
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.424778035665309e-05,
+      "loss": 1.009,
+      "step": 2613
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.421560905450821e-05,
+      "loss": 0.8189,
+      "step": 2614
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.4183440179386336e-05,
+      "loss": 0.9112,
+      "step": 2615
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.415127374478491e-05,
+      "loss": 1.288,
+      "step": 2616
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.4119109764200396e-05,
+      "loss": 0.9095,
+      "step": 2617
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.4086948251128156e-05,
+      "loss": 0.7539,
+      "step": 2618
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.4054789219062595e-05,
+      "loss": 0.8893,
+      "step": 2619
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.402263268149706e-05,
+      "loss": 1.1649,
+      "step": 2620
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.39904786519238e-05,
+      "loss": 1.1212,
+      "step": 2621
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.395832714383409e-05,
+      "loss": 0.9941,
+      "step": 2622
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.392617817071806e-05,
+      "loss": 1.1031,
+      "step": 2623
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.389403174606484e-05,
+      "loss": 1.114,
+      "step": 2624
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.386188788336251e-05,
+      "loss": 0.9824,
+      "step": 2625
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.382974659609798e-05,
+      "loss": 0.8644,
+      "step": 2626
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.379760789775717e-05,
+      "loss": 0.9716,
+      "step": 2627
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.376547180182486e-05,
+      "loss": 1.0063,
+      "step": 2628
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.373333832178478e-05,
+      "loss": 0.9726,
+      "step": 2629
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.3701207471119556e-05,
+      "loss": 1.0879,
+      "step": 2630
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.366907926331065e-05,
+      "loss": 1.1953,
+      "step": 2631
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.3636953711838493e-05,
+      "loss": 0.94,
+      "step": 2632
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.360483083018235e-05,
+      "loss": 1.1806,
+      "step": 2633
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.357271063182039e-05,
+      "loss": 0.7874,
+      "step": 2634
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.35405931302297e-05,
+      "loss": 1.0696,
+      "step": 2635
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.3508478338886103e-05,
+      "loss": 0.9503,
+      "step": 2636
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.3476366271264425e-05,
+      "loss": 1.0919,
+      "step": 2637
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.344425694083828e-05,
+      "loss": 0.9809,
+      "step": 2638
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.341215036108014e-05,
+      "loss": 0.9185,
+      "step": 2639
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.338004654546136e-05,
+      "loss": 0.6861,
+      "step": 2640
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.334794550745204e-05,
+      "loss": 1.0378,
+      "step": 2641
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.331584726052124e-05,
+      "loss": 1.0321,
+      "step": 2642
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.3283751818136784e-05,
+      "loss": 0.9831,
+      "step": 2643
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.3251659193765284e-05,
+      "loss": 1.1868,
+      "step": 2644
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.321956940087224e-05,
+      "loss": 0.7603,
+      "step": 2645
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.3187482452921924e-05,
+      "loss": 1.0332,
+      "step": 2646
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.315539836337742e-05,
+      "loss": 0.971,
+      "step": 2647
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.312331714570064e-05,
+      "loss": 0.9704,
+      "step": 2648
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.309123881335222e-05,
+      "loss": 1.0755,
+      "step": 2649
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.305916337979168e-05,
+      "loss": 0.8322,
+      "step": 2650
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.3027090858477236e-05,
+      "loss": 1.0722,
+      "step": 2651
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.299502126286595e-05,
+      "loss": 1.0043,
+      "step": 2652
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.2962954606413644e-05,
+      "loss": 1.0281,
+      "step": 2653
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.293089090257484e-05,
+      "loss": 0.8273,
+      "step": 2654
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.289883016480291e-05,
+      "loss": 0.9445,
+      "step": 2655
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.286677240654993e-05,
+      "loss": 1.1923,
+      "step": 2656
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.283471764126674e-05,
+      "loss": 0.998,
+      "step": 2657
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.2802665882402936e-05,
+      "loss": 1.0562,
+      "step": 2658
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.277061714340683e-05,
+      "loss": 0.9594,
+      "step": 2659
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.27385714377255e-05,
+      "loss": 1.1575,
+      "step": 2660
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.2706528778804686e-05,
+      "loss": 0.9445,
+      "step": 2661
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.267448918008892e-05,
+      "loss": 1.1008,
+      "step": 2662
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.2642452655021426e-05,
+      "loss": 0.9678,
+      "step": 2663
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.2610419217044115e-05,
+      "loss": 0.9733,
+      "step": 2664
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.2578388879597636e-05,
+      "loss": 1.1183,
+      "step": 2665
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.2546361656121346e-05,
+      "loss": 1.1691,
+      "step": 2666
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.2514337560053226e-05,
+      "loss": 0.9407,
+      "step": 2667
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.2482316604830024e-05,
+      "loss": 1.1837,
+      "step": 2668
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.2450298803887126e-05,
+      "loss": 1.1867,
+      "step": 2669
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.24182841706586e-05,
+      "loss": 0.9851,
+      "step": 2670
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.238627271857723e-05,
+      "loss": 1.1773,
+      "step": 2671
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.235426446107437e-05,
+      "loss": 0.9925,
+      "step": 2672
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.2322259411580134e-05,
+      "loss": 0.9938,
+      "step": 2673
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.229025758352322e-05,
+      "loss": 0.8204,
+      "step": 2674
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.225825899033101e-05,
+      "loss": 0.8986,
+      "step": 2675
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.222626364542953e-05,
+      "loss": 0.9621,
+      "step": 2676
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.219427156224341e-05,
+      "loss": 1.2089,
+      "step": 2677
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.216228275419597e-05,
+      "loss": 0.9594,
+      "step": 2678
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.213029723470908e-05,
+      "loss": 1.1297,
+      "step": 2679
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.2098315017203284e-05,
+      "loss": 0.9541,
+      "step": 2680
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.2066336115097736e-05,
+      "loss": 0.9271,
+      "step": 2681
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.203436054181017e-05,
+      "loss": 0.9514,
+      "step": 2682
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.200238831075698e-05,
+      "loss": 1.0583,
+      "step": 2683
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.197041943535307e-05,
+      "loss": 0.8911,
+      "step": 2684
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.193845392901201e-05,
+      "loss": 0.9732,
+      "step": 2685
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.1906491805145946e-05,
+      "loss": 1.0649,
+      "step": 2686
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.1874533077165566e-05,
+      "loss": 0.8922,
+      "step": 2687
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.18425777584802e-05,
+      "loss": 0.8388,
+      "step": 2688
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.181062586249765e-05,
+      "loss": 0.9507,
+      "step": 2689
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.177867740262437e-05,
+      "loss": 1.2437,
+      "step": 2690
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.174673239226534e-05,
+      "loss": 1.0117,
+      "step": 2691
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.1714790844824084e-05,
+      "loss": 1.1674,
+      "step": 2692
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.1682852773702685e-05,
+      "loss": 0.8529,
+      "step": 2693
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.165091819230178e-05,
+      "loss": 1.0927,
+      "step": 2694
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.1618987114020494e-05,
+      "loss": 0.9817,
+      "step": 2695
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.1587059552256566e-05,
+      "loss": 0.9288,
+      "step": 2696
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.155513552040614e-05,
+      "loss": 0.9775,
+      "step": 2697
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.152321503186399e-05,
+      "loss": 1.1528,
+      "step": 2698
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.149129810002336e-05,
+      "loss": 1.0658,
+      "step": 2699
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.145938473827598e-05,
+      "loss": 1.0609,
+      "step": 2700
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.1427474960012145e-05,
+      "loss": 0.9573,
+      "step": 2701
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.1395568778620556e-05,
+      "loss": 1.1577,
+      "step": 2702
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.1363666207488464e-05,
+      "loss": 0.9598,
+      "step": 2703
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.133176726000163e-05,
+      "loss": 0.9178,
+      "step": 2704
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.129987194954421e-05,
+      "loss": 0.9664,
+      "step": 2705
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.1267980289498945e-05,
+      "loss": 0.8524,
+      "step": 2706
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.123609229324693e-05,
+      "loss": 0.8735,
+      "step": 2707
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.120420797416777e-05,
+      "loss": 0.9323,
+      "step": 2708
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.117232734563959e-05,
+      "loss": 1.1836,
+      "step": 2709
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.114045042103887e-05,
+      "loss": 0.8528,
+      "step": 2710
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.110857721374058e-05,
+      "loss": 0.8746,
+      "step": 2711
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.107670773711812e-05,
+      "loss": 0.8405,
+      "step": 2712
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.104484200454334e-05,
+      "loss": 1.0768,
+      "step": 2713
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.101298002938653e-05,
+      "loss": 1.0092,
+      "step": 2714
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.0981121825016335e-05,
+      "loss": 0.8549,
+      "step": 2715
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.0949267404799906e-05,
+      "loss": 0.9853,
+      "step": 2716
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.0917416782102735e-05,
+      "loss": 0.9765,
+      "step": 2717
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.088556997028877e-05,
+      "loss": 0.9717,
+      "step": 2718
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.0853726982720364e-05,
+      "loss": 0.9902,
+      "step": 2719
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.08218878327582e-05,
+      "loss": 0.9733,
+      "step": 2720
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.079005253376142e-05,
+      "loss": 1.0013,
+      "step": 2721
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.07582210990875e-05,
+      "loss": 1.0034,
+      "step": 2722
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.072639354209234e-05,
+      "loss": 0.9269,
+      "step": 2723
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.069456987613022e-05,
+      "loss": 1.0579,
+      "step": 2724
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.066275011455369e-05,
+      "loss": 0.9617,
+      "step": 2725
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.063093427071376e-05,
+      "loss": 1.0329,
+      "step": 2726
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.0599122357959806e-05,
+      "loss": 1.1651,
+      "step": 2727
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.0567314389639464e-05,
+      "loss": 0.8938,
+      "step": 2728
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.0535510379098804e-05,
+      "loss": 0.8532,
+      "step": 2729
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.050371033968216e-05,
+      "loss": 1.0527,
+      "step": 2730
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.047191428473226e-05,
+      "loss": 1.0868,
+      "step": 2731
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.044012222759016e-05,
+      "loss": 1.0004,
+      "step": 2732
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.040833418159517e-05,
+      "loss": 1.1331,
+      "step": 2733
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.0376550160085e-05,
+      "loss": 0.8801,
+      "step": 2734
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.0344770176395606e-05,
+      "loss": 0.7707,
+      "step": 2735
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.03129942438613e-05,
+      "loss": 1.0529,
+      "step": 2736
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.0281222375814685e-05,
+      "loss": 0.9163,
+      "step": 2737
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.024945458558661e-05,
+      "loss": 1.0947,
+      "step": 2738
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.0217690886506274e-05,
+      "loss": 1.0359,
+      "step": 2739
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.018593129190113e-05,
+      "loss": 0.9732,
+      "step": 2740
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.0154175815096906e-05,
+      "loss": 1.1289,
+      "step": 2741
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.012242446941764e-05,
+      "loss": 1.011,
+      "step": 2742
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.009067726818556e-05,
+      "loss": 1.1049,
+      "step": 2743
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.005893422472125e-05,
+      "loss": 1.1609,
+      "step": 2744
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.002719535234346e-05,
+      "loss": 0.7746,
+      "step": 2745
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.999546066436925e-05,
+      "loss": 0.8284,
+      "step": 2746
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.996373017411392e-05,
+      "loss": 1.26,
+      "step": 2747
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.9932003894890964e-05,
+      "loss": 0.7771,
+      "step": 2748
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.990028184001218e-05,
+      "loss": 0.8529,
+      "step": 2749
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.9868564022787505e-05,
+      "loss": 1.2466,
+      "step": 2750
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.983685045652516e-05,
+      "loss": 0.9921,
+      "step": 2751
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.98051411545316e-05,
+      "loss": 1.0407,
+      "step": 2752
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.9773436130111416e-05,
+      "loss": 1.2728,
+      "step": 2753
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.974173539656747e-05,
+      "loss": 1.0832,
+      "step": 2754
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.971003896720082e-05,
+      "loss": 0.9996,
+      "step": 2755
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.967834685531066e-05,
+      "loss": 0.8677,
+      "step": 2756
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.9646659074194444e-05,
+      "loss": 0.8954,
+      "step": 2757
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.961497563714774e-05,
+      "loss": 1.23,
+      "step": 2758
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.958329655746435e-05,
+      "loss": 1.1009,
+      "step": 2759
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.955162184843625e-05,
+      "loss": 0.7399,
+      "step": 2760
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.951995152335352e-05,
+      "loss": 1.1266,
+      "step": 2761
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.948828559550448e-05,
+      "loss": 1.0491,
+      "step": 2762
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.945662407817552e-05,
+      "loss": 1.2812,
+      "step": 2763
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.942496698465125e-05,
+      "loss": 1.2086,
+      "step": 2764
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.9393314328214404e-05,
+      "loss": 1.0243,
+      "step": 2765
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.936166612214584e-05,
+      "loss": 1.2307,
+      "step": 2766
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.933002237972456e-05,
+      "loss": 0.8199,
+      "step": 2767
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.9298383114227675e-05,
+      "loss": 0.9756,
+      "step": 2768
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.9266748338930455e-05,
+      "loss": 0.8159,
+      "step": 2769
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.9235118067106255e-05,
+      "loss": 0.9273,
+      "step": 2770
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.920349231202655e-05,
+      "loss": 0.9831,
+      "step": 2771
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.917187108696094e-05,
+      "loss": 0.8362,
+      "step": 2772
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.9140254405177054e-05,
+      "loss": 1.017,
+      "step": 2773
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.91086422799407e-05,
+      "loss": 1.1989,
+      "step": 2774
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.9077034724515735e-05,
+      "loss": 0.9456,
+      "step": 2775
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.9045431752164094e-05,
+      "loss": 1.0585,
+      "step": 2776
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.901383337614582e-05,
+      "loss": 1.1008,
+      "step": 2777
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.898223960971896e-05,
+      "loss": 1.0721,
+      "step": 2778
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.8950650466139717e-05,
+      "loss": 0.959,
+      "step": 2779
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.89190659586623e-05,
+      "loss": 0.8531,
+      "step": 2780
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.8887486100538953e-05,
+      "loss": 1.1931,
+      "step": 2781
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.8855910905020026e-05,
+      "loss": 0.9284,
+      "step": 2782
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.882434038535386e-05,
+      "loss": 0.9847,
+      "step": 2783
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.879277455478689e-05,
+      "loss": 1.0155,
+      "step": 2784
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.876121342656355e-05,
+      "loss": 1.0554,
+      "step": 2785
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.872965701392626e-05,
+      "loss": 1.1031,
+      "step": 2786
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.869810533011554e-05,
+      "loss": 0.8375,
+      "step": 2787
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.86665583883699e-05,
+      "loss": 0.9582,
+      "step": 2788
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.863501620192581e-05,
+      "loss": 1.0163,
+      "step": 2789
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.860347878401785e-05,
+      "loss": 1.0581,
+      "step": 2790
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.8571946147878456e-05,
+      "loss": 1.0692,
+      "step": 2791
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.8540418306738186e-05,
+      "loss": 1.1383,
+      "step": 2792
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.850889527382554e-05,
+      "loss": 1.0688,
+      "step": 2793
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.847737706236696e-05,
+      "loss": 0.9576,
+      "step": 2794
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.8445863685586946e-05,
+      "loss": 1.0339,
+      "step": 2795
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.841435515670789e-05,
+      "loss": 1.1233,
+      "step": 2796
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.83828514889502e-05,
+      "loss": 1.0035,
+      "step": 2797
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.835135269553226e-05,
+      "loss": 0.7064,
+      "step": 2798
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.8319858789670315e-05,
+      "loss": 0.9573,
+      "step": 2799
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.828836978457867e-05,
+      "loss": 0.9949,
+      "step": 2800
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.8256885693469515e-05,
+      "loss": 1.0964,
+      "step": 2801
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.8225406529552976e-05,
+      "loss": 1.027,
+      "step": 2802
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.819393230603716e-05,
+      "loss": 1.1996,
+      "step": 2803
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.816246303612802e-05,
+      "loss": 1.1125,
+      "step": 2804
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.8130998733029515e-05,
+      "loss": 0.9718,
+      "step": 2805
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.809953940994346e-05,
+      "loss": 0.8992,
+      "step": 2806
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.8068085080069606e-05,
+      "loss": 1.1009,
+      "step": 2807
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.803663575660563e-05,
+      "loss": 0.9912,
+      "step": 2808
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.800519145274705e-05,
+      "loss": 0.8015,
+      "step": 2809
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.7973752181687335e-05,
+      "loss": 1.1669,
+      "step": 2810
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.79423179566178e-05,
+      "loss": 1.1344,
+      "step": 2811
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.791088879072766e-05,
+      "loss": 1.165,
+      "step": 2812
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.787946469720405e-05,
+      "loss": 1.1316,
+      "step": 2813
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.784804568923188e-05,
+      "loss": 0.8742,
+      "step": 2814
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.7816631779994014e-05,
+      "loss": 1.0251,
+      "step": 2815
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.778522298267117e-05,
+      "loss": 1.2363,
+      "step": 2816
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.775381931044184e-05,
+      "loss": 1.0583,
+      "step": 2817
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.772242077648246e-05,
+      "loss": 1.0508,
+      "step": 2818
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.769102739396724e-05,
+      "loss": 0.8923,
+      "step": 2819
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.765963917606828e-05,
+      "loss": 1.0916,
+      "step": 2820
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.7628256135955523e-05,
+      "loss": 0.8359,
+      "step": 2821
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.7596878286796654e-05,
+      "loss": 1.0308,
+      "step": 2822
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.756550564175727e-05,
+      "loss": 1.0812,
+      "step": 2823
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.753413821400073e-05,
+      "loss": 1.1334,
+      "step": 2824
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.7502776016688233e-05,
+      "loss": 0.903,
+      "step": 2825
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.7471419062978804e-05,
+      "loss": 0.8743,
+      "step": 2826
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.7440067366029185e-05,
+      "loss": 1.2649,
+      "step": 2827
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.7408720938994e-05,
+      "loss": 1.0018,
+      "step": 2828
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.7377379795025614e-05,
+      "loss": 1.1274,
+      "step": 2829
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.734604394727419e-05,
+      "loss": 1.0822,
+      "step": 2830
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.731471340888768e-05,
+      "loss": 0.9648,
+      "step": 2831
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.7283388193011775e-05,
+      "loss": 1.2036,
+      "step": 2832
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.725206831278999e-05,
+      "loss": 1.0058,
+      "step": 2833
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.722075378136352e-05,
+      "loss": 1.0304,
+      "step": 2834
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.718944461187138e-05,
+      "loss": 1.1126,
+      "step": 2835
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.715814081745033e-05,
+      "loss": 1.3473,
+      "step": 2836
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.712684241123484e-05,
+      "loss": 0.8351,
+      "step": 2837
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.709554940635716e-05,
+      "loss": 1.2308,
+      "step": 2838
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.7064261815947235e-05,
+      "loss": 0.9551,
+      "step": 2839
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.703297965313275e-05,
+      "loss": 1.2281,
+      "step": 2840
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.700170293103915e-05,
+      "loss": 1.0139,
+      "step": 2841
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.697043166278953e-05,
+      "loss": 1.1036,
+      "step": 2842
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.693916586150479e-05,
+      "loss": 0.9166,
+      "step": 2843
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.69079055403034e-05,
+      "loss": 0.8696,
+      "step": 2844
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.687665071230165e-05,
+      "loss": 0.8284,
+      "step": 2845
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.684540139061353e-05,
+      "loss": 1.1746,
+      "step": 2846
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.68141575883506e-05,
+      "loss": 0.6988,
+      "step": 2847
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.678291931862221e-05,
+      "loss": 1.5432,
+      "step": 2848
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.675168659453537e-05,
+      "loss": 0.8461,
+      "step": 2849
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.672045942919474e-05,
+      "loss": 1.0592,
+      "step": 2850
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.668923783570267e-05,
+      "loss": 0.9594,
+      "step": 2851
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.6658021827159126e-05,
+      "loss": 1.2007,
+      "step": 2852
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.66268114166618e-05,
+      "loss": 1.0474,
+      "step": 2853
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.6595606617305986e-05,
+      "loss": 0.9635,
+      "step": 2854
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.656440744218464e-05,
+      "loss": 1.1289,
+      "step": 2855
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.6533213904388374e-05,
+      "loss": 1.1273,
+      "step": 2856
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.650202601700538e-05,
+      "loss": 0.9284,
+      "step": 2857
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.647084379312154e-05,
+      "loss": 0.9698,
+      "step": 2858
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.6439667245820345e-05,
+      "loss": 0.9711,
+      "step": 2859
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.640849638818286e-05,
+      "loss": 0.8591,
+      "step": 2860
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.637733123328785e-05,
+      "loss": 1.0968,
+      "step": 2861
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.634617179421157e-05,
+      "loss": 1.2488,
+      "step": 2862
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.631501808402799e-05,
+      "loss": 1.0191,
+      "step": 2863
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.628387011580864e-05,
+      "loss": 0.9224,
+      "step": 2864
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.6252727902622574e-05,
+      "loss": 0.8331,
+      "step": 2865
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.622159145753654e-05,
+      "loss": 1.1438,
+      "step": 2866
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.619046079361477e-05,
+      "loss": 1.0236,
+      "step": 2867
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.615933592391913e-05,
+      "loss": 1.0305,
+      "step": 2868
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.612821686150906e-05,
+      "loss": 1.2629,
+      "step": 2869
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.60971036194415e-05,
+      "loss": 0.7922,
+      "step": 2870
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.6065996210771016e-05,
+      "loss": 1.0649,
+      "step": 2871
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.6034894648549675e-05,
+      "loss": 1.1305,
+      "step": 2872
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.600379894582714e-05,
+      "loss": 1.1212,
+      "step": 2873
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.597270911565059e-05,
+      "loss": 1.0264,
+      "step": 2874
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.594162517106472e-05,
+      "loss": 1.0269,
+      "step": 2875
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.5910547125111784e-05,
+      "loss": 1.0209,
+      "step": 2876
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.587947499083155e-05,
+      "loss": 1.0367,
+      "step": 2877
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.584840878126132e-05,
+      "loss": 1.1174,
+      "step": 2878
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.58173485094359e-05,
+      "loss": 1.2679,
+      "step": 2879
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.578629418838757e-05,
+      "loss": 1.0679,
+      "step": 2880
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.575524583114618e-05,
+      "loss": 0.8457,
+      "step": 2881
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.5724203450739056e-05,
+      "loss": 1.1141,
+      "step": 2882
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.569316706019097e-05,
+      "loss": 1.1065,
+      "step": 2883
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.566213667252423e-05,
+      "loss": 0.9091,
+      "step": 2884
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.5631112300758595e-05,
+      "loss": 0.8599,
+      "step": 2885
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.560009395791133e-05,
+      "loss": 1.0901,
+      "step": 2886
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.5569081656997175e-05,
+      "loss": 1.1213,
+      "step": 2887
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.553807541102827e-05,
+      "loss": 0.7791,
+      "step": 2888
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.55070752330143e-05,
+      "loss": 1.1022,
+      "step": 2889
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.547608113596233e-05,
+      "loss": 1.0213,
+      "step": 2890
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.5445093132876936e-05,
+      "loss": 1.0787,
+      "step": 2891
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.5414111236760115e-05,
+      "loss": 0.8626,
+      "step": 2892
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.538313546061126e-05,
+      "loss": 0.7648,
+      "step": 2893
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.535216581742725e-05,
+      "loss": 1.1356,
+      "step": 2894
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.532120232020236e-05,
+      "loss": 0.9011,
+      "step": 2895
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.529024498192832e-05,
+      "loss": 1.2007,
+      "step": 2896
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.525929381559425e-05,
+      "loss": 1.0856,
+      "step": 2897
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.5228348834186663e-05,
+      "loss": 0.9948,
+      "step": 2898
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.519741005068954e-05,
+      "loss": 0.8977,
+      "step": 2899
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.516647747808417e-05,
+      "loss": 1.0367,
+      "step": 2900
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.5135551129349306e-05,
+      "loss": 1.1069,
+      "step": 2901
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.510463101746109e-05,
+      "loss": 0.9735,
+      "step": 2902
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.507371715539299e-05,
+      "loss": 0.8624,
+      "step": 2903
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.504280955611593e-05,
+      "loss": 1.1063,
+      "step": 2904
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.5011908232598125e-05,
+      "loss": 1.0889,
+      "step": 2905
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.4981013197805204e-05,
+      "loss": 0.9633,
+      "step": 2906
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.495012446470018e-05,
+      "loss": 0.7723,
+      "step": 2907
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.4919242046243355e-05,
+      "loss": 1.1177,
+      "step": 2908
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.488836595539243e-05,
+      "loss": 1.0899,
+      "step": 2909
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.4857496205102474e-05,
+      "loss": 0.8084,
+      "step": 2910
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.48266328083258e-05,
+      "loss": 1.0301,
+      "step": 2911
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.4795775778012164e-05,
+      "loss": 1.234,
+      "step": 2912
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.476492512710857e-05,
+      "loss": 1.0483,
+      "step": 2913
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.473408086855939e-05,
+      "loss": 0.9969,
+      "step": 2914
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.470324301530631e-05,
+      "loss": 1.2104,
+      "step": 2915
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.4672411580288313e-05,
+      "loss": 1.0199,
+      "step": 2916
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.464158657644172e-05,
+      "loss": 1.1093,
+      "step": 2917
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.461076801670008e-05,
+      "loss": 0.7393,
+      "step": 2918
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.4579955913994325e-05,
+      "loss": 0.845,
+      "step": 2919
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.4549150281252636e-05,
+      "loss": 0.874,
+      "step": 2920
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.4518351131400476e-05,
+      "loss": 0.9801,
+      "step": 2921
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.448755847736061e-05,
+      "loss": 0.8186,
+      "step": 2922
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.4456772332053045e-05,
+      "loss": 1.0099,
+      "step": 2923
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.442599270839508e-05,
+      "loss": 0.9523,
+      "step": 2924
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.4395219619301286e-05,
+      "loss": 1.0111,
+      "step": 2925
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.436445307768347e-05,
+      "loss": 1.1718,
+      "step": 2926
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.4333693096450714e-05,
+      "loss": 1.2617,
+      "step": 2927
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.430293968850931e-05,
+      "loss": 0.8434,
+      "step": 2928
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.427219286676283e-05,
+      "loss": 1.1369,
+      "step": 2929
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.424145264411208e-05,
+      "loss": 0.9672,
+      "step": 2930
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.421071903345506e-05,
+      "loss": 0.8206,
+      "step": 2931
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.417999204768706e-05,
+      "loss": 1.1042,
+      "step": 2932
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.414927169970051e-05,
+      "loss": 1.1412,
+      "step": 2933
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.4118558002385124e-05,
+      "loss": 1.1504,
+      "step": 2934
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.408785096862782e-05,
+      "loss": 0.9374,
+      "step": 2935
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.405715061131264e-05,
+      "loss": 0.9614,
+      "step": 2936
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.4026456943320925e-05,
+      "loss": 0.8533,
+      "step": 2937
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.3995769977531166e-05,
+      "loss": 0.8488,
+      "step": 2938
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.3965089726819024e-05,
+      "loss": 1.018,
+      "step": 2939
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.3934416204057395e-05,
+      "loss": 1.1626,
+      "step": 2940
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.390374942211628e-05,
+      "loss": 0.8831,
+      "step": 2941
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.387308939386291e-05,
+      "loss": 0.8882,
+      "step": 2942
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.384243613216168e-05,
+      "loss": 1.0307,
+      "step": 2943
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.38117896498741e-05,
+      "loss": 1.0929,
+      "step": 2944
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.3781149959858896e-05,
+      "loss": 1.0415,
+      "step": 2945
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.375051707497187e-05,
+      "loss": 0.9601,
+      "step": 2946
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.371989100806603e-05,
+      "loss": 1.0841,
+      "step": 2947
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.368927177199154e-05,
+      "loss": 1.0234,
+      "step": 2948
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.365865937959561e-05,
+      "loss": 1.16,
+      "step": 2949
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.362805384372267e-05,
+      "loss": 0.933,
+      "step": 2950
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.359745517721421e-05,
+      "loss": 0.9498,
+      "step": 2951
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.356686339290887e-05,
+      "loss": 1.143,
+      "step": 2952
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.353627850364241e-05,
+      "loss": 0.9777,
+      "step": 2953
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.350570052224765e-05,
+      "loss": 1.1767,
+      "step": 2954
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.347512946155457e-05,
+      "loss": 0.9629,
+      "step": 2955
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.3444565334390196e-05,
+      "loss": 1.228,
+      "step": 2956
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.3414008153578685e-05,
+      "loss": 0.9536,
+      "step": 2957
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.338345793194128e-05,
+      "loss": 0.9366,
+      "step": 2958
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.335291468229624e-05,
+      "loss": 0.778,
+      "step": 2959
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.332237841745898e-05,
+      "loss": 0.9385,
+      "step": 2960
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.329184915024193e-05,
+      "loss": 1.0181,
+      "step": 2961
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.326132689345461e-05,
+      "loss": 0.777,
+      "step": 2962
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.323081165990363e-05,
+      "loss": 0.7946,
+      "step": 2963
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.3200303462392544e-05,
+      "loss": 0.9118,
+      "step": 2964
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.316980231372208e-05,
+      "loss": 1.4156,
+      "step": 2965
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.313930822668992e-05,
+      "loss": 0.9692,
+      "step": 2966
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.310882121409084e-05,
+      "loss": 0.9601,
+      "step": 2967
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.3078341288716605e-05,
+      "loss": 1.0001,
+      "step": 2968
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.304786846335604e-05,
+      "loss": 1.0192,
+      "step": 2969
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.3017402750794974e-05,
+      "loss": 1.0805,
+      "step": 2970
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.298694416381627e-05,
+      "loss": 1.2801,
+      "step": 2971
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.295649271519974e-05,
+      "loss": 0.9809,
+      "step": 2972
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.2926048417722295e-05,
+      "loss": 0.8613,
+      "step": 2973
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.2895611284157755e-05,
+      "loss": 0.9847,
+      "step": 2974
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.2865181327277e-05,
+      "loss": 0.8911,
+      "step": 2975
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.28347585598479e-05,
+      "loss": 1.0396,
+      "step": 2976
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.280434299463523e-05,
+      "loss": 1.1967,
+      "step": 2977
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.277393464440082e-05,
+      "loss": 0.945,
+      "step": 2978
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.274353352190345e-05,
+      "loss": 0.9964,
+      "step": 2979
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.271313963989886e-05,
+      "loss": 0.8902,
+      "step": 2980
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.268275301113978e-05,
+      "loss": 1.1136,
+      "step": 2981
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.2652373648375836e-05,
+      "loss": 0.9008,
+      "step": 2982
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.26220015643537e-05,
+      "loss": 0.8743,
+      "step": 2983
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.259163677181687e-05,
+      "loss": 1.1283,
+      "step": 2984
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.2561279283505883e-05,
+      "loss": 0.7668,
+      "step": 2985
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.253092911215819e-05,
+      "loss": 0.9714,
+      "step": 2986
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.250058627050813e-05,
+      "loss": 0.9755,
+      "step": 2987
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.2470250771287036e-05,
+      "loss": 1.0887,
+      "step": 2988
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.243992262722306e-05,
+      "loss": 1.0785,
+      "step": 2989
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.240960185104137e-05,
+      "loss": 1.2008,
+      "step": 2990
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.2379288455464e-05,
+      "loss": 1.0495,
+      "step": 2991
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.2348982453209866e-05,
+      "loss": 1.0105,
+      "step": 2992
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.231868385699485e-05,
+      "loss": 0.9593,
+      "step": 2993
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.228839267953161e-05,
+      "loss": 0.7784,
+      "step": 2994
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.225810893352981e-05,
+      "loss": 0.9424,
+      "step": 2995
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.222783263169594e-05,
+      "loss": 0.9519,
+      "step": 2996
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.219756378673335e-05,
+      "loss": 1.2381,
+      "step": 2997
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.216730241134232e-05,
+      "loss": 1.0224,
+      "step": 2998
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.2137048518219946e-05,
+      "loss": 0.7536,
+      "step": 2999
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.21068021200602e-05,
+      "loss": 1.016,
+      "step": 3000
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.207656322955393e-05,
+      "loss": 0.9543,
+      "step": 3001
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.2046331859388756e-05,
+      "loss": 1.201,
+      "step": 3002
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.201610802224923e-05,
+      "loss": 0.9416,
+      "step": 3003
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.1985891730816744e-05,
+      "loss": 0.9635,
+      "step": 3004
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.195568299776945e-05,
+      "loss": 1.0595,
+      "step": 3005
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.19254818357824e-05,
+      "loss": 1.004,
+      "step": 3006
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.189528825752741e-05,
+      "loss": 1.0687,
+      "step": 3007
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.186510227567317e-05,
+      "loss": 0.8487,
+      "step": 3008
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.1834923902885146e-05,
+      "loss": 1.0555,
+      "step": 3009
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.180475315182563e-05,
+      "loss": 1.1488,
+      "step": 3010
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.177459003515371e-05,
+      "loss": 1.2482,
+      "step": 3011
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.174443456552525e-05,
+      "loss": 1.0593,
+      "step": 3012
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.171428675559294e-05,
+      "loss": 0.9879,
+      "step": 3013
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.168414661800625e-05,
+      "loss": 1.0957,
+      "step": 3014
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.1654014165411396e-05,
+      "loss": 1.0137,
+      "step": 3015
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.1623889410451435e-05,
+      "loss": 0.9518,
+      "step": 3016
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.1593772365766105e-05,
+      "loss": 1.0064,
+      "step": 3017
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.156366304399199e-05,
+      "loss": 1.0084,
+      "step": 3018
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.1533561457762415e-05,
+      "loss": 0.9756,
+      "step": 3019
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.150346761970741e-05,
+      "loss": 1.0057,
+      "step": 3020
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.147338154245381e-05,
+      "loss": 1.0501,
+      "step": 3021
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.144330323862517e-05,
+      "loss": 1.1879,
+      "step": 3022
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.141323272084179e-05,
+      "loss": 0.9606,
+      "step": 3023
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.1383170001720716e-05,
+      "loss": 0.9689,
+      "step": 3024
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.1353115093875676e-05,
+      "loss": 0.9843,
+      "step": 3025
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.1323068009917174e-05,
+      "loss": 0.9303,
+      "step": 3026
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.129302876245239e-05,
+      "loss": 0.9964,
+      "step": 3027
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.126299736408525e-05,
+      "loss": 0.9683,
+      "step": 3028
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.1232973827416366e-05,
+      "loss": 1.2055,
+      "step": 3029
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.120295816504305e-05,
+      "loss": 0.7432,
+      "step": 3030
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.117295038955933e-05,
+      "loss": 1.0949,
+      "step": 3031
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.11429505135559e-05,
+      "loss": 1.0589,
+      "step": 3032
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.111295854962015e-05,
+      "loss": 0.9488,
+      "step": 3033
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.108297451033616e-05,
+      "loss": 0.7517,
+      "step": 3034
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.105299840828466e-05,
+      "loss": 0.8815,
+      "step": 3035
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.1023030256043084e-05,
+      "loss": 1.3195,
+      "step": 3036
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.0993070066185526e-05,
+      "loss": 1.006,
+      "step": 3037
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.0963117851282676e-05,
+      "loss": 1.0561,
+      "step": 3038
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.093317362390197e-05,
+      "loss": 1.0058,
+      "step": 3039
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.090323739660742e-05,
+      "loss": 0.7892,
+      "step": 3040
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.087330918195972e-05,
+      "loss": 0.8969,
+      "step": 3041
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.0843388992516234e-05,
+      "loss": 0.9533,
+      "step": 3042
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.081347684083084e-05,
+      "loss": 1.1098,
+      "step": 3043
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.078357273945419e-05,
+      "loss": 0.8373,
+      "step": 3044
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.075367670093345e-05,
+      "loss": 0.9507,
+      "step": 3045
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.0723788737812446e-05,
+      "loss": 1.1202,
+      "step": 3046
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.069390886263165e-05,
+      "loss": 0.9429,
+      "step": 3047
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.066403708792805e-05,
+      "loss": 1.1512,
+      "step": 3048
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.0634173426235326e-05,
+      "loss": 0.9313,
+      "step": 3049
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.0604317890083676e-05,
+      "loss": 1.1688,
+      "step": 3050
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.057447049199997e-05,
+      "loss": 1.0196,
+      "step": 3051
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.0544631244507604e-05,
+      "loss": 1.2383,
+      "step": 3052
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.0514800160126566e-05,
+      "loss": 0.757,
+      "step": 3053
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.048497725137346e-05,
+      "loss": 0.8545,
+      "step": 3054
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.045516253076137e-05,
+      "loss": 0.7861,
+      "step": 3055
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.042535601080002e-05,
+      "loss": 0.8846,
+      "step": 3056
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.0395557703995696e-05,
+      "loss": 1.3228,
+      "step": 3057
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.0365767622851183e-05,
+      "loss": 0.8089,
+      "step": 3058
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.033598577986586e-05,
+      "loss": 0.7873,
+      "step": 3059
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.0306212187535653e-05,
+      "loss": 0.8932,
+      "step": 3060
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.0276446858352963e-05,
+      "loss": 0.9638,
+      "step": 3061
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.024668980480681e-05,
+      "loss": 1.0694,
+      "step": 3062
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.0216941039382672e-05,
+      "loss": 1.241,
+      "step": 3063
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.0187200574562602e-05,
+      "loss": 0.988,
+      "step": 3064
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.0157468422825145e-05,
+      "loss": 1.1325,
+      "step": 3065
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.0127744596645336e-05,
+      "loss": 0.9026,
+      "step": 3066
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.0098029108494764e-05,
+      "loss": 1.0294,
+      "step": 3067
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.0068321970841485e-05,
+      "loss": 0.9861,
+      "step": 3068
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.0038623196150052e-05,
+      "loss": 0.9155,
+      "step": 3069
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.000893279688155e-05,
+      "loss": 0.9264,
+      "step": 3070
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.997925078549349e-05,
+      "loss": 1.1027,
+      "step": 3071
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.9949577174439926e-05,
+      "loss": 1.0616,
+      "step": 3072
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.9919911976171304e-05,
+      "loss": 1.1396,
+      "step": 3073
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.989025520313462e-05,
+      "loss": 1.0845,
+      "step": 3074
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.986060686777332e-05,
+      "loss": 1.1891,
+      "step": 3075
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.9830966982527258e-05,
+      "loss": 0.9012,
+      "step": 3076
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.980133555983282e-05,
+      "loss": 0.8279,
+      "step": 3077
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.9771712612122767e-05,
+      "loss": 0.8941,
+      "step": 3078
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.974209815182635e-05,
+      "loss": 0.8471,
+      "step": 3079
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.9712492191369244e-05,
+      "loss": 0.9182,
+      "step": 3080
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.9682894743173562e-05,
+      "loss": 1.1198,
+      "step": 3081
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.9653305819657862e-05,
+      "loss": 1.0976,
+      "step": 3082
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.9623725433237065e-05,
+      "loss": 1.0302,
+      "step": 3083
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.9594153596322572e-05,
+      "loss": 0.8014,
+      "step": 3084
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.9564590321322207e-05,
+      "loss": 0.9739,
+      "step": 3085
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.9535035620640116e-05,
+      "loss": 0.9016,
+      "step": 3086
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.9505489506676936e-05,
+      "loss": 1.0934,
+      "step": 3087
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.9475951991829675e-05,
+      "loss": 1.2278,
+      "step": 3088
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.9446423088491693e-05,
+      "loss": 1.0895,
+      "step": 3089
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.9416902809052815e-05,
+      "loss": 0.969,
+      "step": 3090
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.938739116589916e-05,
+      "loss": 1.0923,
+      "step": 3091
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.935788817141327e-05,
+      "loss": 0.8682,
+      "step": 3092
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.9328393837974082e-05,
+      "loss": 0.9575,
+      "step": 3093
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.9298908177956842e-05,
+      "loss": 0.9004,
+      "step": 3094
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.9269431203733212e-05,
+      "loss": 0.8882,
+      "step": 3095
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.923996292767115e-05,
+      "loss": 1.1545,
+      "step": 3096
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.921050336213501e-05,
+      "loss": 1.0227,
+      "step": 3097
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.9181052519485496e-05,
+      "loss": 0.9626,
+      "step": 3098
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.9151610412079612e-05,
+      "loss": 0.9986,
+      "step": 3099
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.912217705227075e-05,
+      "loss": 1.0147,
+      "step": 3100
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.9092752452408556e-05,
+      "loss": 0.9055,
+      "step": 3101
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.9063336624839067e-05,
+      "loss": 0.8905,
+      "step": 3102
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.9033929581904617e-05,
+      "loss": 0.9656,
+      "step": 3103
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.9004531335943863e-05,
+      "loss": 1.0914,
+      "step": 3104
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.8975141899291775e-05,
+      "loss": 1.0566,
+      "step": 3105
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.894576128427958e-05,
+      "loss": 1.0899,
+      "step": 3106
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.8916389503234854e-05,
+      "loss": 0.8865,
+      "step": 3107
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.8887026568481467e-05,
+      "loss": 1.0163,
+      "step": 3108
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.8857672492339537e-05,
+      "loss": 1.1232,
+      "step": 3109
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.882832728712551e-05,
+      "loss": 1.2485,
+      "step": 3110
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.879899096515206e-05,
+      "loss": 0.9022,
+      "step": 3111
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.8769663538728176e-05,
+      "loss": 0.9626,
+      "step": 3112
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.874034502015911e-05,
+      "loss": 1.0289,
+      "step": 3113
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.8711035421746367e-05,
+      "loss": 1.0521,
+      "step": 3114
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.868173475578772e-05,
+      "loss": 1.1052,
+      "step": 3115
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.8652443034577147e-05,
+      "loss": 0.917,
+      "step": 3116
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.8623160270404943e-05,
+      "loss": 1.0256,
+      "step": 3117
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.8593886475557618e-05,
+      "loss": 0.9181,
+      "step": 3118
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.8564621662317875e-05,
+      "loss": 1.0068,
+      "step": 3119
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.853536584296471e-05,
+      "loss": 0.9599,
+      "step": 3120
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.850611902977332e-05,
+      "loss": 0.8929,
+      "step": 3121
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.8476881235015125e-05,
+      "loss": 0.9483,
+      "step": 3122
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.8447652470957777e-05,
+      "loss": 0.9922,
+      "step": 3123
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.8418432749865088e-05,
+      "loss": 1.3176,
+      "step": 3124
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.838922208399712e-05,
+      "loss": 0.9538,
+      "step": 3125
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.8360020485610163e-05,
+      "loss": 1.0204,
+      "step": 3126
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.8330827966956608e-05,
+      "loss": 1.1465,
+      "step": 3127
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.830164454028514e-05,
+      "loss": 1.0227,
+      "step": 3128
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.8272470217840542e-05,
+      "loss": 0.9857,
+      "step": 3129
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.824330501186384e-05,
+      "loss": 0.9986,
+      "step": 3130
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.8214148934592223e-05,
+      "loss": 1.1981,
+      "step": 3131
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.818500199825902e-05,
+      "loss": 1.0584,
+      "step": 3132
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.815586421509377e-05,
+      "loss": 1.1762,
+      "step": 3133
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.8126735597322107e-05,
+      "loss": 1.155,
+      "step": 3134
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.8097616157165883e-05,
+      "loss": 0.9734,
+      "step": 3135
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.806850590684309e-05,
+      "loss": 0.6439,
+      "step": 3136
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.8039404858567807e-05,
+      "loss": 1.2212,
+      "step": 3137
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.8010313024550316e-05,
+      "loss": 1.1881,
+      "step": 3138
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.7981230416997008e-05,
+      "loss": 0.9649,
+      "step": 3139
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.7952157048110406e-05,
+      "loss": 1.002,
+      "step": 3140
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.7923092930089175e-05,
+      "loss": 1.0512,
+      "step": 3141
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.7894038075128036e-05,
+      "loss": 1.0969,
+      "step": 3142
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.7864992495417912e-05,
+      "loss": 0.9828,
+      "step": 3143
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.7835956203145754e-05,
+      "loss": 0.753,
+      "step": 3144
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.780692921049465e-05,
+      "loss": 0.9142,
+      "step": 3145
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.777791152964383e-05,
+      "loss": 1.0748,
+      "step": 3146
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.774890317276852e-05,
+      "loss": 1.046,
+      "step": 3147
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.771990415204011e-05,
+      "loss": 1.0143,
+      "step": 3148
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.7690914479626058e-05,
+      "loss": 1.0499,
+      "step": 3149
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.7661934167689884e-05,
+      "loss": 0.9696,
+      "step": 3150
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.7632963228391206e-05,
+      "loss": 1.0161,
+      "step": 3151
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.760400167388566e-05,
+      "loss": 1.0094,
+      "step": 3152
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.757504951632499e-05,
+      "loss": 0.9478,
+      "step": 3153
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.7546106767857017e-05,
+      "loss": 1.012,
+      "step": 3154
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.751717344062552e-05,
+      "loss": 0.9412,
+      "step": 3155
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.7488249546770416e-05,
+      "loss": 1.1736,
+      "step": 3156
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.745933509842764e-05,
+      "loss": 1.1167,
+      "step": 3157
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.7430430107729143e-05,
+      "loss": 1.3004,
+      "step": 3158
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.7401534586802957e-05,
+      "loss": 1.0189,
+      "step": 3159
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.737264854777306e-05,
+      "loss": 0.8315,
+      "step": 3160
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.7343772002759526e-05,
+      "loss": 0.8963,
+      "step": 3161
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.7314904963878397e-05,
+      "loss": 0.9756,
+      "step": 3162
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.728604744324175e-05,
+      "loss": 0.8235,
+      "step": 3163
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.7257199452957693e-05,
+      "loss": 0.8313,
+      "step": 3164
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.722836100513027e-05,
+      "loss": 1.053,
+      "step": 3165
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.719953211185957e-05,
+      "loss": 0.9885,
+      "step": 3166
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.717071278524166e-05,
+      "loss": 1.0604,
+      "step": 3167
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.71419030373686e-05,
+      "loss": 0.8413,
+      "step": 3168
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.7113102880328445e-05,
+      "loss": 0.9377,
+      "step": 3169
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.7084312326205165e-05,
+      "loss": 1.1162,
+      "step": 3170
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.705553138707878e-05,
+      "loss": 0.8273,
+      "step": 3171
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.7026760075025192e-05,
+      "loss": 1.0876,
+      "step": 3172
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.699799840211634e-05,
+      "loss": 1.2069,
+      "step": 3173
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.696924638042009e-05,
+      "loss": 0.8893,
+      "step": 3174
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.6940504022000246e-05,
+      "loss": 0.8855,
+      "step": 3175
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.691177133891658e-05,
+      "loss": 1.1201,
+      "step": 3176
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.6883048343224772e-05,
+      "loss": 1.1347,
+      "step": 3177
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.685433504697647e-05,
+      "loss": 0.977,
+      "step": 3178
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.682563146221926e-05,
+      "loss": 1.1337,
+      "step": 3179
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.6796937600996587e-05,
+      "loss": 0.8734,
+      "step": 3180
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.676825347534788e-05,
+      "loss": 1.1409,
+      "step": 3181
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.6739579097308497e-05,
+      "loss": 1.0374,
+      "step": 3182
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.6710914478909633e-05,
+      "loss": 1.1639,
+      "step": 3183
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.668225963217844e-05,
+      "loss": 0.9071,
+      "step": 3184
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.6653614569137968e-05,
+      "loss": 0.9306,
+      "step": 3185
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.6624979301807152e-05,
+      "loss": 1.0286,
+      "step": 3186
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.6596353842200834e-05,
+      "loss": 1.1364,
+      "step": 3187
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.6567738202329683e-05,
+      "loss": 0.9494,
+      "step": 3188
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.653913239420034e-05,
+      "loss": 1.0738,
+      "step": 3189
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.6510536429815226e-05,
+      "loss": 1.0879,
+      "step": 3190
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.6481950321172693e-05,
+      "loss": 0.8485,
+      "step": 3191
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.6453374080266945e-05,
+      "loss": 0.9579,
+      "step": 3192
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.6424807719088042e-05,
+      "loss": 1.171,
+      "step": 3193
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.6396251249621917e-05,
+      "loss": 1.0407,
+      "step": 3194
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.636770468385029e-05,
+      "loss": 1.0687,
+      "step": 3195
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.63391680337508e-05,
+      "loss": 1.0646,
+      "step": 3196
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.63106413112969e-05,
+      "loss": 1.0558,
+      "step": 3197
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.628212452845785e-05,
+      "loss": 1.1647,
+      "step": 3198
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.6253617697198794e-05,
+      "loss": 0.9389,
+      "step": 3199
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.622512082948063e-05,
+      "loss": 0.9841,
+      "step": 3200
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.619663393726014e-05,
+      "loss": 0.8578,
+      "step": 3201
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.616815703248988e-05,
+      "loss": 0.9348,
+      "step": 3202
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.6139690127118238e-05,
+      "loss": 0.9104,
+      "step": 3203
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.611123323308943e-05,
+      "loss": 0.9738,
+      "step": 3204
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.6082786362343377e-05,
+      "loss": 0.9925,
+      "step": 3205
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.6054349526815892e-05,
+      "loss": 1.16,
+      "step": 3206
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.6025922738438568e-05,
+      "loss": 0.9211,
+      "step": 3207
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.5997506009138706e-05,
+      "loss": 0.758,
+      "step": 3208
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.5969099350839467e-05,
+      "loss": 0.9481,
+      "step": 3209
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.5940702775459747e-05,
+      "loss": 1.3127,
+      "step": 3210
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.591231629491423e-05,
+      "loss": 1.0442,
+      "step": 3211
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.588393992111337e-05,
+      "loss": 1.0113,
+      "step": 3212
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.5855573665963334e-05,
+      "loss": 0.7794,
+      "step": 3213
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.5827217541366087e-05,
+      "loss": 0.9125,
+      "step": 3214
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.5798871559219363e-05,
+      "loss": 0.9958,
+      "step": 3215
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.5770535731416557e-05,
+      "loss": 1.0408,
+      "step": 3216
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.57422100698469e-05,
+      "loss": 1.0886,
+      "step": 3217
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.5713894586395283e-05,
+      "loss": 0.9346,
+      "step": 3218
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.5685589292942357e-05,
+      "loss": 0.9397,
+      "step": 3219
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.5657294201364522e-05,
+      "loss": 1.0153,
+      "step": 3220
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.5629009323533852e-05,
+      "loss": 1.018,
+      "step": 3221
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.560073467131819e-05,
+      "loss": 1.1835,
+      "step": 3222
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.5572470256581005e-05,
+      "loss": 0.9852,
+      "step": 3223
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.554421609118155e-05,
+      "loss": 1.276,
+      "step": 3224
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.551597218697476e-05,
+      "loss": 0.9083,
+      "step": 3225
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.5487738555811213e-05,
+      "loss": 0.9805,
+      "step": 3226
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.5459515209537242e-05,
+      "loss": 1.0745,
+      "step": 3227
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.5431302159994834e-05,
+      "loss": 0.8806,
+      "step": 3228
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.5403099419021664e-05,
+      "loss": 0.933,
+      "step": 3229
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.537490699845109e-05,
+      "loss": 1.1385,
+      "step": 3230
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.5346724910112102e-05,
+      "loss": 0.8992,
+      "step": 3231
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.5318553165829407e-05,
+      "loss": 1.2223,
+      "step": 3232
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.5290391777423315e-05,
+      "loss": 0.9406,
+      "step": 3233
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.5262240756709836e-05,
+      "loss": 0.8922,
+      "step": 3234
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.5234100115500643e-05,
+      "loss": 0.956,
+      "step": 3235
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.5205969865602973e-05,
+      "loss": 1.0574,
+      "step": 3236
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.5177850018819786e-05,
+      "loss": 1.1025,
+      "step": 3237
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.5149740586949648e-05,
+      "loss": 1.3225,
+      "step": 3238
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.5121641581786747e-05,
+      "loss": 1.0776,
+      "step": 3239
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.5093553015120935e-05,
+      "loss": 0.995,
+      "step": 3240
+    },
+    {
+      "epoch": 1.0,
+      "eval_loss": 1.0341861248016357,
+      "eval_runtime": 24.853,
+      "eval_samples_per_second": 109.806,
+      "eval_steps_per_second": 13.761,
+      "step": 3240
+    }
+  ],
+  "logging_steps": 1,
+  "max_steps": 4860,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 2,
+  "save_steps": 3240,
+  "total_flos": 2.2025315703953818e+17,
+  "train_batch_size": 2,
+  "trial_name": null,
+  "trial_params": null
+}