diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,7914 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 3.0,
+  "eval_steps": 500,
+  "global_step": 1314,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 5e-06,
+      "loss": 1.6819,
+      "step": 1
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1e-05,
+      "loss": 1.3456,
+      "step": 2
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.5e-05,
+      "loss": 1.6309,
+      "step": 3
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2e-05,
+      "loss": 1.5144,
+      "step": 4
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.5e-05,
+      "loss": 1.5809,
+      "step": 5
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3e-05,
+      "loss": 1.4337,
+      "step": 6
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.5e-05,
+      "loss": 1.4476,
+      "step": 7
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4e-05,
+      "loss": 1.5057,
+      "step": 8
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.5e-05,
+      "loss": 1.2802,
+      "step": 9
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5e-05,
+      "loss": 1.4187,
+      "step": 10
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.500000000000001e-05,
+      "loss": 1.3839,
+      "step": 11
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6e-05,
+      "loss": 1.3517,
+      "step": 12
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.500000000000001e-05,
+      "loss": 1.1437,
+      "step": 13
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7e-05,
+      "loss": 1.432,
+      "step": 14
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.500000000000001e-05,
+      "loss": 1.3195,
+      "step": 15
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8e-05,
+      "loss": 1.2114,
+      "step": 16
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.5e-05,
+      "loss": 1.1665,
+      "step": 17
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9e-05,
+      "loss": 1.4073,
+      "step": 18
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.5e-05,
+      "loss": 1.2103,
+      "step": 19
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0001,
+      "loss": 1.3239,
+      "step": 20
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.000105,
+      "loss": 1.2707,
+      "step": 21
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00011000000000000002,
+      "loss": 1.1545,
+      "step": 22
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00011499999999999999,
+      "loss": 1.3175,
+      "step": 23
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00012,
+      "loss": 1.0579,
+      "step": 24
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.000125,
+      "loss": 1.3129,
+      "step": 25
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00013000000000000002,
+      "loss": 1.165,
+      "step": 26
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00013500000000000003,
+      "loss": 1.2197,
+      "step": 27
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00014,
+      "loss": 1.2123,
+      "step": 28
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.000145,
+      "loss": 1.2125,
+      "step": 29
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00015000000000000001,
+      "loss": 1.1781,
+      "step": 30
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.000155,
+      "loss": 1.2041,
+      "step": 31
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00016,
+      "loss": 1.1366,
+      "step": 32
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.000165,
+      "loss": 1.146,
+      "step": 33
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00017,
+      "loss": 1.1651,
+      "step": 34
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.000175,
+      "loss": 1.239,
+      "step": 35
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00018,
+      "loss": 1.2837,
+      "step": 36
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00018500000000000002,
+      "loss": 1.1846,
+      "step": 37
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019,
+      "loss": 1.3183,
+      "step": 38
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.000195,
+      "loss": 1.2054,
+      "step": 39
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0002,
+      "loss": 1.2084,
+      "step": 40
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019999969596003515,
+      "loss": 1.1971,
+      "step": 41
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0001999987838419894,
+      "loss": 1.202,
+      "step": 42
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019999726365140918,
+      "loss": 1.4853,
+      "step": 43
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0001999951353975384,
+      "loss": 1.1747,
+      "step": 44
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0001999923990933186,
+      "loss": 1.3454,
+      "step": 45
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019998905475538873,
+      "loss": 1.2324,
+      "step": 46
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019998510240408496,
+      "loss": 1.1027,
+      "step": 47
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019998054206344076,
+      "loss": 1.1632,
+      "step": 48
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019997537376118668,
+      "loss": 1.1752,
+      "step": 49
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001999695975287501,
+      "loss": 1.2875,
+      "step": 50
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019996321340125519,
+      "loss": 1.1448,
+      "step": 51
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019995622141752244,
+      "loss": 1.1917,
+      "step": 52
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001999486216200688,
+      "loss": 1.3177,
+      "step": 53
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019994041405510705,
+      "loss": 1.272,
+      "step": 54
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019993159877254577,
+      "loss": 1.1357,
+      "step": 55
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019992217582598894,
+      "loss": 1.2682,
+      "step": 56
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019991214527273558,
+      "loss": 1.1612,
+      "step": 57
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019990150717377948,
+      "loss": 1.1252,
+      "step": 58
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0001998902615938088,
+      "loss": 1.125,
+      "step": 59
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019987840860120562,
+      "loss": 1.2295,
+      "step": 60
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0001998659482680456,
+      "loss": 1.2334,
+      "step": 61
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019985288067009762,
+      "loss": 1.2692,
+      "step": 62
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019983920588682302,
+      "loss": 1.2275,
+      "step": 63
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019982492400137544,
+      "loss": 1.2909,
+      "step": 64
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019981003510060016,
+      "loss": 1.1595,
+      "step": 65
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019979453927503364,
+      "loss": 1.1614,
+      "step": 66
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019977843661890283,
+      "loss": 1.1937,
+      "step": 67
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001997617272301248,
+      "loss": 1.1761,
+      "step": 68
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001997444112103059,
+      "loss": 1.1969,
+      "step": 69
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019972648866474146,
+      "loss": 1.1487,
+      "step": 70
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019970795970241482,
+      "loss": 1.1874,
+      "step": 71
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019968882443599696,
+      "loss": 1.1191,
+      "step": 72
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019966908298184552,
+      "loss": 1.0698,
+      "step": 73
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019964873546000436,
+      "loss": 1.2601,
+      "step": 74
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019962778199420265,
+      "loss": 1.0591,
+      "step": 75
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019960622271185426,
+      "loss": 1.2135,
+      "step": 76
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019958405774405676,
+      "loss": 1.0993,
+      "step": 77
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019956128722559098,
+      "loss": 1.2044,
+      "step": 78
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019953791129491983,
+      "loss": 1.2691,
+      "step": 79
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019951393009418763,
+      "loss": 1.2178,
+      "step": 80
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001994893437692193,
+      "loss": 1.1314,
+      "step": 81
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0001994641524695193,
+      "loss": 1.1364,
+      "step": 82
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00019943835634827085,
+      "loss": 1.2325,
+      "step": 83
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00019941195556233505,
+      "loss": 1.1242,
+      "step": 84
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00019938495027224974,
+      "loss": 1.2255,
+      "step": 85
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001993573406422287,
+      "loss": 1.253,
+      "step": 86
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00019932912684016051,
+      "loss": 1.2712,
+      "step": 87
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00019930030903760764,
+      "loss": 1.3722,
+      "step": 88
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001992708874098054,
+      "loss": 1.2131,
+      "step": 89
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00019924086213566076,
+      "loss": 1.2516,
+      "step": 90
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00019921023339775147,
+      "loss": 1.1578,
+      "step": 91
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0001991790013823246,
+      "loss": 1.2367,
+      "step": 92
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0001991471662792959,
+      "loss": 1.2715,
+      "step": 93
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00019911472828224817,
+      "loss": 1.0971,
+      "step": 94
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00019908168758843035,
+      "loss": 1.1773,
+      "step": 95
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00019904804439875633,
+      "loss": 1.1332,
+      "step": 96
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00019901379891780355,
+      "loss": 1.0965,
+      "step": 97
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00019897895135381188,
+      "loss": 1.2134,
+      "step": 98
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00019894350191868246,
+      "loss": 1.0705,
+      "step": 99
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00019890745082797606,
+      "loss": 1.2053,
+      "step": 100
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00019887079830091224,
+      "loss": 1.1043,
+      "step": 101
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00019883354456036757,
+      "loss": 1.216,
+      "step": 102
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00019879568983287467,
+      "loss": 1.2416,
+      "step": 103
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00019875723434862043,
+      "loss": 1.1489,
+      "step": 104
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00019871817834144504,
+      "loss": 1.2676,
+      "step": 105
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00019867852204884015,
+      "loss": 1.298,
+      "step": 106
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00019863826571194776,
+      "loss": 1.15,
+      "step": 107
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00019859740957555856,
+      "loss": 1.2133,
+      "step": 108
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00019855595388811053,
+      "loss": 1.4566,
+      "step": 109
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0001985138989016874,
+      "loss": 1.3723,
+      "step": 110
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00019847124487201704,
+      "loss": 1.0748,
+      "step": 111
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00019842799205847014,
+      "loss": 1.1737,
+      "step": 112
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00019838414072405826,
+      "loss": 1.3032,
+      "step": 113
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00019833969113543267,
+      "loss": 1.1347,
+      "step": 114
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00019829464356288233,
+      "loss": 1.306,
+      "step": 115
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00019824899828033252,
+      "loss": 1.0725,
+      "step": 116
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00019820275556534304,
+      "loss": 1.3514,
+      "step": 117
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00019815591569910654,
+      "loss": 1.0601,
+      "step": 118
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00019810847896644686,
+      "loss": 1.1273,
+      "step": 119
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00019806044565581727,
+      "loss": 1.0795,
+      "step": 120
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00019801181605929864,
+      "loss": 1.3234,
+      "step": 121
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00019796259047259783,
+      "loss": 1.1698,
+      "step": 122
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00019791276919504572,
+      "loss": 1.1879,
+      "step": 123
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00019786235252959553,
+      "loss": 1.1941,
+      "step": 124
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00019781134078282087,
+      "loss": 1.2291,
+      "step": 125
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00019775973426491394,
+      "loss": 1.1777,
+      "step": 126
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0001977075332896836,
+      "loss": 1.2625,
+      "step": 127
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00019765473817455356,
+      "loss": 1.133,
+      "step": 128
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00019760134924056024,
+      "loss": 1.2002,
+      "step": 129
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0001975473668123511,
+      "loss": 1.2641,
+      "step": 130
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00019749279121818235,
+      "loss": 1.205,
+      "step": 131
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0001974376227899174,
+      "loss": 1.1622,
+      "step": 132
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0001973818618630242,
+      "loss": 1.0307,
+      "step": 133
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00019732550877657386,
+      "loss": 1.3657,
+      "step": 134
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00019726856387323815,
+      "loss": 1.0734,
+      "step": 135
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00019721102749928763,
+      "loss": 1.2233,
+      "step": 136
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00019715290000458946,
+      "loss": 1.101,
+      "step": 137
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0001970941817426052,
+      "loss": 1.2623,
+      "step": 138
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0001970348730703889,
+      "loss": 1.1674,
+      "step": 139
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0001969749743485846,
+      "loss": 1.0925,
+      "step": 140
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00019691448594142446,
+      "loss": 1.2044,
+      "step": 141
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00019685340821672633,
+      "loss": 1.1814,
+      "step": 142
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00019679174154589162,
+      "loss": 1.1966,
+      "step": 143
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00019672948630390294,
+      "loss": 1.1292,
+      "step": 144
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00019666664286932198,
+      "loss": 1.1505,
+      "step": 145
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00019660321162428697,
+      "loss": 1.0732,
+      "step": 146
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00019653919295451063,
+      "loss": 0.9373,
+      "step": 147
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00019647458724927766,
+      "loss": 1.1033,
+      "step": 148
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0001964093949014423,
+      "loss": 1.1025,
+      "step": 149
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0001963436163074263,
+      "loss": 1.2571,
+      "step": 150
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0001962772518672159,
+      "loss": 1.2958,
+      "step": 151
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00019621030198436006,
+      "loss": 1.1826,
+      "step": 152
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00019614276706596755,
+      "loss": 1.0773,
+      "step": 153
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00019607464752270464,
+      "loss": 1.0419,
+      "step": 154
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0001960059437687926,
+      "loss": 1.1271,
+      "step": 155
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00019593665622200522,
+      "loss": 1.338,
+      "step": 156
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00019586678530366606,
+      "loss": 1.1358,
+      "step": 157
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00019579633143864626,
+      "loss": 1.2576,
+      "step": 158
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0001957252950553616,
+      "loss": 1.3208,
+      "step": 159
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00019565367658577003,
+      "loss": 1.2977,
+      "step": 160
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00019558147646536912,
+      "loss": 1.1039,
+      "step": 161
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00019550869513319333,
+      "loss": 1.1451,
+      "step": 162
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00019543533303181132,
+      "loss": 1.2214,
+      "step": 163
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00019536139060732328,
+      "loss": 1.172,
+      "step": 164
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00019528686830935828,
+      "loss": 1.2827,
+      "step": 165
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00019521176659107142,
+      "loss": 1.1607,
+      "step": 166
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00019513608590914123,
+      "loss": 1.1205,
+      "step": 167
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0001950598267237667,
+      "loss": 1.1986,
+      "step": 168
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00019498298949866468,
+      "loss": 1.26,
+      "step": 169
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00019490557470106686,
+      "loss": 1.198,
+      "step": 170
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00019482758280171716,
+      "loss": 1.3148,
+      "step": 171
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00019474901427486856,
+      "loss": 1.0349,
+      "step": 172
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0001946698695982806,
+      "loss": 1.277,
+      "step": 173
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00019459014925321614,
+      "loss": 1.0323,
+      "step": 174
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00019450985372443857,
+      "loss": 1.2584,
+      "step": 175
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00019442898350020892,
+      "loss": 1.3343,
+      "step": 176
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00019434753907228279,
+      "loss": 1.1134,
+      "step": 177
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0001942655209359074,
+      "loss": 1.0455,
+      "step": 178
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00019418292958981855,
+      "loss": 1.2267,
+      "step": 179
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00019409976553623766,
+      "loss": 1.0701,
+      "step": 180
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00019401602928086864,
+      "loss": 1.1027,
+      "step": 181
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0001939317213328949,
+      "loss": 1.1847,
+      "step": 182
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00019384684220497605,
+      "loss": 1.177,
+      "step": 183
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0001937613924132451,
+      "loss": 1.1607,
+      "step": 184
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0001936753724773051,
+      "loss": 1.3098,
+      "step": 185
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.000193588782920226,
+      "loss": 1.0301,
+      "step": 186
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0001935016242685415,
+      "loss": 1.2378,
+      "step": 187
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00019341389705224585,
+      "loss": 1.1978,
+      "step": 188
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0001933256018047907,
+      "loss": 1.0652,
+      "step": 189
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00019323673906308168,
+      "loss": 1.1856,
+      "step": 190
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00019314730936747533,
+      "loss": 1.1039,
+      "step": 191
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00019305731326177566,
+      "loss": 1.2479,
+      "step": 192
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0001929667512932309,
+      "loss": 1.0944,
+      "step": 193
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00019287562401253022,
+      "loss": 1.1903,
+      "step": 194
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00019278393197380038,
+      "loss": 1.1799,
+      "step": 195
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0001926916757346022,
+      "loss": 1.0663,
+      "step": 196
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00019259885585592735,
+      "loss": 1.0972,
+      "step": 197
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00019250547290219493,
+      "loss": 0.9872,
+      "step": 198
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00019241152744124787,
+      "loss": 1.2429,
+      "step": 199
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00019231702004434976,
+      "loss": 1.1506,
+      "step": 200
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00019222195128618106,
+      "loss": 1.2485,
+      "step": 201
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00019212632174483578,
+      "loss": 1.2173,
+      "step": 202
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.000192030132001818,
+      "loss": 1.1646,
+      "step": 203
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00019193338264203823,
+      "loss": 1.2385,
+      "step": 204
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00019183607425380994,
+      "loss": 1.0444,
+      "step": 205
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00019173820742884588,
+      "loss": 1.1717,
+      "step": 206
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00019163978276225455,
+      "loss": 1.1795,
+      "step": 207
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00019154080085253666,
+      "loss": 1.1373,
+      "step": 208
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00019144126230158127,
+      "loss": 0.9912,
+      "step": 209
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00019134116771466234,
+      "loss": 1.1482,
+      "step": 210
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.000191240517700435,
+      "loss": 1.1779,
+      "step": 211
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00019113931287093177,
+      "loss": 1.1486,
+      "step": 212
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0001910375538415589,
+      "loss": 1.2069,
+      "step": 213
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00019093524123109262,
+      "loss": 1.2227,
+      "step": 214
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0001908323756616754,
+      "loss": 1.1186,
+      "step": 215
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0001907289577588121,
+      "loss": 0.9993,
+      "step": 216
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00019062498815136626,
+      "loss": 1.037,
+      "step": 217
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00019052046747155616,
+      "loss": 1.1127,
+      "step": 218
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0001904153963549511,
+      "loss": 1.0655,
+      "step": 219
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00019030977544046742,
+      "loss": 1.0131,
+      "step": 220
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0001902036053703647,
+      "loss": 1.1767,
+      "step": 221
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0001900968867902419,
+      "loss": 1.1909,
+      "step": 222
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00018998962034903326,
+      "loss": 1.1264,
+      "step": 223
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00018988180669900442,
+      "loss": 1.0607,
+      "step": 224
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00018977344649574862,
+      "loss": 0.9862,
+      "step": 225
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00018966454039818245,
+      "loss": 1.2387,
+      "step": 226
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00018955508906854206,
+      "loss": 1.1928,
+      "step": 227
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00018944509317237905,
+      "loss": 1.1949,
+      "step": 228
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00018933455337855632,
+      "loss": 1.2275,
+      "step": 229
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0001892234703592442,
+      "loss": 1.1021,
+      "step": 230
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00018911184478991627,
+      "loss": 1.4385,
+      "step": 231
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00018899967734934515,
+      "loss": 1.1383,
+      "step": 232
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0001888869687195986,
+      "loss": 1.0239,
+      "step": 233
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00018877371958603514,
+      "loss": 1.2352,
+      "step": 234
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00018865993063730004,
+      "loss": 1.1515,
+      "step": 235
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.000188545602565321,
+      "loss": 1.0593,
+      "step": 236
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00018843073606530415,
+      "loss": 1.096,
+      "step": 237
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0001883153318357296,
+      "loss": 1.2894,
+      "step": 238
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0001881993905783473,
+      "loss": 1.0143,
+      "step": 239
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00018808291299817278,
+      "loss": 1.2318,
+      "step": 240
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00018796589980348282,
+      "loss": 1.0275,
+      "step": 241
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00018784835170581116,
+      "loss": 1.1465,
+      "step": 242
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0001877302694199442,
+      "loss": 1.147,
+      "step": 243
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00018761165366391663,
+      "loss": 1.0989,
+      "step": 244
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00018749250515900706,
+      "loss": 1.0905,
+      "step": 245
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00018737282462973363,
+      "loss": 0.9672,
+      "step": 246
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00018725261280384958,
+      "loss": 1.1728,
+      "step": 247
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00018713187041233896,
+      "loss": 1.1453,
+      "step": 248
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00018701059818941195,
+      "loss": 1.1036,
+      "step": 249
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00018688879687250067,
+      "loss": 1.0362,
+      "step": 250
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00018676646720225444,
+      "loss": 1.1521,
+      "step": 251
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00018664360992253542,
+      "loss": 0.9757,
+      "step": 252
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00018652022578041414,
+      "loss": 1.1481,
+      "step": 253
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00018639631552616468,
+      "loss": 1.0764,
+      "step": 254
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00018627187991326056,
+      "loss": 1.153,
+      "step": 255
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00018614691969836967,
+      "loss": 1.1499,
+      "step": 256
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0001860214356413501,
+      "loss": 1.0191,
+      "step": 257
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0001858954285052451,
+      "loss": 1.1723,
+      "step": 258
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00018576889905627883,
+      "loss": 1.0602,
+      "step": 259
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00018564184806385146,
+      "loss": 1.1461,
+      "step": 260
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00018551427630053463,
+      "loss": 1.1267,
+      "step": 261
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00018538618454206655,
+      "loss": 1.0706,
+      "step": 262
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00018525757356734754,
+      "loss": 1.1054,
+      "step": 263
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00018512844415843514,
+      "loss": 1.2113,
+      "step": 264
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00018499879710053932,
+      "loss": 1.0667,
+      "step": 265
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00018486863318201784,
+      "loss": 1.1547,
+      "step": 266
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00018473795319437136,
+      "loss": 1.1461,
+      "step": 267
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00018460675793223867,
+      "loss": 1.1663,
+      "step": 268
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00018447504819339182,
+      "loss": 1.0636,
+      "step": 269
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00018434282477873133,
+      "loss": 1.2808,
+      "step": 270
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00018421008849228118,
+      "loss": 1.1668,
+      "step": 271
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00018407684014118414,
+      "loss": 1.0716,
+      "step": 272
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00018394308053569665,
+      "loss": 1.1568,
+      "step": 273
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00018380881048918405,
+      "loss": 1.1537,
+      "step": 274
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00018367403081811557,
+      "loss": 1.1852,
+      "step": 275
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00018353874234205931,
+      "loss": 1.164,
+      "step": 276
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00018340294588367734,
+      "loss": 1.2054,
+      "step": 277
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00018326664226872065,
+      "loss": 1.1812,
+      "step": 278
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00018312983232602418,
+      "loss": 1.1312,
+      "step": 279
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00018299251688750175,
+      "loss": 1.1017,
+      "step": 280
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.000182854696788141,
+      "loss": 1.1522,
+      "step": 281
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00018271637286599823,
+      "loss": 1.0512,
+      "step": 282
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0001825775459621935,
+      "loss": 1.1148,
+      "step": 283
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00018243821692090533,
+      "loss": 1.0796,
+      "step": 284
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00018229838658936564,
+      "loss": 1.2327,
+      "step": 285
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00018215805581785465,
+      "loss": 1.1625,
+      "step": 286
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0001820172254596956,
+      "loss": 1.09,
+      "step": 287
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00018187589637124962,
+      "loss": 1.2245,
+      "step": 288
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00018173406941191056,
+      "loss": 1.1396,
+      "step": 289
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00018159174544409964,
+      "loss": 1.1925,
+      "step": 290
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00018144892533326042,
+      "loss": 1.194,
+      "step": 291
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00018130560994785325,
+      "loss": 0.9423,
+      "step": 292
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00018116180015935029,
+      "loss": 0.9744,
+      "step": 293
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00018101749684222996,
+      "loss": 1.059,
+      "step": 294
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00018087270087397181,
+      "loss": 1.0716,
+      "step": 295
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.000180727413135051,
+      "loss": 1.1569,
+      "step": 296
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00018058163450893319,
+      "loss": 0.9845,
+      "step": 297
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00018043536588206887,
+      "loss": 1.1119,
+      "step": 298
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00018028860814388827,
+      "loss": 1.2018,
+      "step": 299
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00018014136218679567,
+      "loss": 1.1423,
+      "step": 300
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00017999362890616424,
+      "loss": 1.0031,
+      "step": 301
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0001798454092003304,
+      "loss": 0.9347,
+      "step": 302
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0001796967039705884,
+      "loss": 1.4172,
+      "step": 303
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00017954751412118495,
+      "loss": 1.0023,
+      "step": 304
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00017939784055931354,
+      "loss": 1.1181,
+      "step": 305
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00017924768419510904,
+      "loss": 0.972,
+      "step": 306
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00017909704594164222,
+      "loss": 1.0207,
+      "step": 307
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00017894592671491405,
+      "loss": 0.9375,
+      "step": 308
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0001787943274338502,
+      "loss": 1.1102,
+      "step": 309
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00017864224902029547,
+      "loss": 1.2381,
+      "step": 310
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0001784896923990082,
+      "loss": 1.0415,
+      "step": 311
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00017833665849765457,
+      "loss": 1.0233,
+      "step": 312
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.000178183148246803,
+      "loss": 1.0763,
+      "step": 313
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00017802916257991853,
+      "loss": 1.2626,
+      "step": 314
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0001778747024333571,
+      "loss": 0.9793,
+      "step": 315
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00017771976874635985,
+      "loss": 1.0299,
+      "step": 316
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00017756436246104743,
+      "loss": 1.1348,
+      "step": 317
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0001774084845224143,
+      "loss": 1.0651,
+      "step": 318
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00017725213587832292,
+      "loss": 1.0831,
+      "step": 319
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00017709531747949796,
+      "loss": 1.0118,
+      "step": 320
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0001769380302795207,
+      "loss": 1.0746,
+      "step": 321
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00017678027523482297,
+      "loss": 1.2129,
+      "step": 322
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00017662205330468158,
+      "loss": 1.0312,
+      "step": 323
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0001764633654512123,
+      "loss": 1.2271,
+      "step": 324
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0001763042126393642,
+      "loss": 1.1998,
+      "step": 325
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00017614459583691346,
+      "loss": 1.1123,
+      "step": 326
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0001759845160144579,
+      "loss": 1.0907,
+      "step": 327
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00017582397414541077,
+      "loss": 1.1867,
+      "step": 328
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00017566297120599495,
+      "loss": 1.0127,
+      "step": 329
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00017550150817523702,
+      "loss": 1.0484,
+      "step": 330
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00017533958603496125,
+      "loss": 0.9896,
+      "step": 331
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00017517720576978367,
+      "loss": 1.0358,
+      "step": 332
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0001750143683671061,
+      "loss": 1.1482,
+      "step": 333
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00017485107481711012,
+      "loss": 1.1702,
+      "step": 334
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00017468732611275098,
+      "loss": 1.0627,
+      "step": 335
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00017452312324975174,
+      "loss": 1.1448,
+      "step": 336
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.000174358467226597,
+      "loss": 1.0747,
+      "step": 337
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00017419335904452705,
+      "loss": 1.0825,
+      "step": 338
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00017402779970753155,
+      "loss": 1.1648,
+      "step": 339
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00017386179022234365,
+      "loss": 1.0696,
+      "step": 340
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00017369533159843369,
+      "loss": 1.0231,
+      "step": 341
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00017352842484800313,
+      "loss": 1.1968,
+      "step": 342
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00017336107098597846,
+      "loss": 0.805,
+      "step": 343
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00017319327103000492,
+      "loss": 1.0623,
+      "step": 344
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00017302502600044032,
+      "loss": 1.1143,
+      "step": 345
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00017285633692034897,
+      "loss": 1.0105,
+      "step": 346
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00017268720481549527,
+      "loss": 1.0811,
+      "step": 347
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00017251763071433765,
+      "loss": 1.0987,
+      "step": 348
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00017234761564802214,
+      "loss": 1.1191,
+      "step": 349
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00017217716065037628,
+      "loss": 1.3072,
+      "step": 350
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00017200626675790268,
+      "loss": 0.9639,
+      "step": 351
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00017183493500977278,
+      "loss": 1.0101,
+      "step": 352
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00017166316644782055,
+      "loss": 1.2726,
+      "step": 353
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00017149096211653618,
+      "loss": 1.1479,
+      "step": 354
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00017131832306305965,
+      "loss": 0.9289,
+      "step": 355
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00017114525033717435,
+      "loss": 1.0264,
+      "step": 356
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.0001709717449913008,
+      "loss": 0.9703,
+      "step": 357
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00017079780808049023,
+      "loss": 1.0139,
+      "step": 358
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.000170623440662418,
+      "loss": 1.2518,
+      "step": 359
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.0001704486437973775,
+      "loss": 1.2049,
+      "step": 360
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00017027341854827327,
+      "loss": 1.1969,
+      "step": 361
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00017009776598061495,
+      "loss": 1.0092,
+      "step": 362
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00016992168716251053,
+      "loss": 0.9765,
+      "step": 363
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00016974518316465996,
+      "loss": 0.9931,
+      "step": 364
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00016956825506034867,
+      "loss": 1.0219,
+      "step": 365
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00016939090392544085,
+      "loss": 1.0712,
+      "step": 366
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00016921313083837327,
+      "loss": 1.0711,
+      "step": 367
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.0001690349368801483,
+      "loss": 1.2672,
+      "step": 368
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.0001688563231343277,
+      "loss": 1.2043,
+      "step": 369
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.0001686772906870258,
+      "loss": 1.0933,
+      "step": 370
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00016849784062690298,
+      "loss": 1.3335,
+      "step": 371
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00016831797404515898,
+      "loss": 1.0658,
+      "step": 372
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00016813769203552647,
+      "loss": 0.9472,
+      "step": 373
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.0001679569956942641,
+      "loss": 1.0334,
+      "step": 374
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00016777588612015013,
+      "loss": 1.0628,
+      "step": 375
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00016759436441447545,
+      "loss": 1.1018,
+      "step": 376
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00016741243168103716,
+      "loss": 1.043,
+      "step": 377
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.0001672300890261317,
+      "loss": 1.1859,
+      "step": 378
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00016704733755854815,
+      "loss": 1.1158,
+      "step": 379
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00016686417838956147,
+      "loss": 1.1174,
+      "step": 380
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00016668061263292589,
+      "loss": 1.1842,
+      "step": 381
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00016649664140486786,
+      "loss": 1.1987,
+      "step": 382
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00016631226582407952,
+      "loss": 1.0134,
+      "step": 383
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.0001661274870117118,
+      "loss": 1.1397,
+      "step": 384
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.0001659423060913675,
+      "loss": 1.129,
+      "step": 385
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.0001657567241890947,
+      "loss": 1.1101,
+      "step": 386
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.0001655707424333797,
+      "loss": 1.1611,
+      "step": 387
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00016538436195514015,
+      "loss": 1.0567,
+      "step": 388
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00016519758388771843,
+      "loss": 1.1743,
+      "step": 389
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00016501040936687443,
+      "loss": 1.2523,
+      "step": 390
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00016482283953077887,
+      "loss": 1.2478,
+      "step": 391
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00016463487552000622,
+      "loss": 1.2612,
+      "step": 392
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.00016444651847752798,
+      "loss": 1.0391,
+      "step": 393
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.00016425776954870545,
+      "loss": 1.2208,
+      "step": 394
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.00016406862988128303,
+      "loss": 1.1316,
+      "step": 395
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.00016387910062538107,
+      "loss": 1.1503,
+      "step": 396
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00016368918293348892,
+      "loss": 1.022,
+      "step": 397
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.000163498877960458,
+      "loss": 0.9542,
+      "step": 398
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00016330818686349457,
+      "loss": 1.096,
+      "step": 399
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00016311711080215297,
+      "loss": 0.8671,
+      "step": 400
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.0001629256509383284,
+      "loss": 1.2355,
+      "step": 401
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00016273380843624983,
+      "loss": 1.0471,
+      "step": 402
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.000162541584462473,
+      "loss": 1.0837,
+      "step": 403
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00016234898018587337,
+      "loss": 1.1093,
+      "step": 404
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00016215599677763884,
+      "loss": 1.2198,
+      "step": 405
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.0001619626354112628,
+      "loss": 1.1788,
+      "step": 406
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00016176889726253688,
+      "loss": 1.0887,
+      "step": 407
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00016157478350954394,
+      "loss": 1.0198,
+      "step": 408
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.0001613802953326507,
+      "loss": 1.1771,
+      "step": 409
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.00016118543391450076,
+      "loss": 0.9436,
+      "step": 410
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.00016099020044000727,
+      "loss": 1.0344,
+      "step": 411
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.00016079459609634586,
+      "loss": 1.0577,
+      "step": 412
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.00016059862207294721,
+      "loss": 0.9137,
+      "step": 413
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.0001604022795614901,
+      "loss": 1.1897,
+      "step": 414
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.00016020556975589382,
+      "loss": 1.1896,
+      "step": 415
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.00016000849385231134,
+      "loss": 1.0696,
+      "step": 416
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.00015981105304912162,
+      "loss": 0.9894,
+      "step": 417
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.00015961324854692254,
+      "loss": 1.0098,
+      "step": 418
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.00015941508154852362,
+      "loss": 1.0692,
+      "step": 419
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.0001592165532589386,
+      "loss": 1.0551,
+      "step": 420
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.00015901766488537813,
+      "loss": 1.1279,
+      "step": 421
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.00015881841763724252,
+      "loss": 0.9259,
+      "step": 422
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.0001586188127261143,
+      "loss": 1.0716,
+      "step": 423
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00015841885136575086,
+      "loss": 1.0828,
+      "step": 424
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00015821853477207708,
+      "loss": 0.9241,
+      "step": 425
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.000158017864163178,
+      "loss": 0.9474,
+      "step": 426
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.0001578168407592913,
+      "loss": 1.1444,
+      "step": 427
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.00015761546578279995,
+      "loss": 1.0687,
+      "step": 428
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.00015741374045822476,
+      "loss": 1.2261,
+      "step": 429
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.00015721166601221698,
+      "loss": 1.0513,
+      "step": 430
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.0001570092436735507,
+      "loss": 1.0648,
+      "step": 431
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.00015680647467311557,
+      "loss": 1.0259,
+      "step": 432
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.0001566033602439092,
+      "loss": 1.0417,
+      "step": 433
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.00015639990162102965,
+      "loss": 1.0093,
+      "step": 434
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.00015619610004166798,
+      "loss": 1.0372,
+      "step": 435
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00015599195674510064,
+      "loss": 0.9534,
+      "step": 436
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00015578747297268213,
+      "loss": 0.9691,
+      "step": 437
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00015558264996783712,
+      "loss": 1.0506,
+      "step": 438
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0001553774889760533,
+      "loss": 0.8641,
+      "step": 439
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00015517199124487346,
+      "loss": 0.5738,
+      "step": 440
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.000154966158023888,
+      "loss": 0.6726,
+      "step": 441
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00015475999056472752,
+      "loss": 0.6682,
+      "step": 442
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00015455349012105486,
+      "loss": 0.5419,
+      "step": 443
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00015434665794855786,
+      "loss": 0.5841,
+      "step": 444
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.0001541394953049414,
+      "loss": 0.6176,
+      "step": 445
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.00015393200344991995,
+      "loss": 0.6603,
+      "step": 446
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.0001537241836452098,
+      "loss": 0.5174,
+      "step": 447
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.00015351603715452154,
+      "loss": 0.6903,
+      "step": 448
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.00015330756524355217,
+      "loss": 0.5471,
+      "step": 449
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.0001530987691799775,
+      "loss": 0.6601,
+      "step": 450
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.00015288965023344458,
+      "loss": 0.4233,
+      "step": 451
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.00015268020967556368,
+      "loss": 0.5764,
+      "step": 452
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.0001524704487799008,
+      "loss": 0.6435,
+      "step": 453
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.00015226036882196988,
+      "loss": 0.5094,
+      "step": 454
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.00015204997107922497,
+      "loss": 0.6163,
+      "step": 455
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.00015183925683105254,
+      "loss": 0.5219,
+      "step": 456
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.00015162822735876358,
+      "loss": 0.5448,
+      "step": 457
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.00015141688394558597,
+      "loss": 0.6222,
+      "step": 458
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.0001512052278766566,
+      "loss": 0.6122,
+      "step": 459
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.0001509932604390136,
+      "loss": 0.6443,
+      "step": 460
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.00015078098292158835,
+      "loss": 0.5781,
+      "step": 461
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.00015056839661519784,
+      "loss": 0.5538,
+      "step": 462
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.00015035550281253674,
+      "loss": 0.6293,
+      "step": 463
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.00015014230280816954,
+      "loss": 0.4054,
+      "step": 464
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.0001499287978985227,
+      "loss": 0.6508,
+      "step": 465
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.00014971498938187667,
+      "loss": 0.7488,
+      "step": 466
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.00014950087855835815,
+      "loss": 0.5754,
+      "step": 467
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.00014928646672993212,
+      "loss": 0.6735,
+      "step": 468
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.0001490717552003938,
+      "loss": 0.5694,
+      "step": 469
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.00014885674527536098,
+      "loss": 0.5249,
+      "step": 470
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.00014864143826226584,
+      "loss": 0.5441,
+      "step": 471
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.00014842583547034708,
+      "loss": 0.5825,
+      "step": 472
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.00014820993821064207,
+      "loss": 0.5348,
+      "step": 473
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.00014799374779597867,
+      "loss": 0.5719,
+      "step": 474
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.00014777726554096737,
+      "loss": 0.4694,
+      "step": 475
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.0001475604927619934,
+      "loss": 0.6275,
+      "step": 476
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.00014734343077720845,
+      "loss": 0.7572,
+      "step": 477
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.00014712608090652294,
+      "loss": 0.6331,
+      "step": 478
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.00014690844447159775,
+      "loss": 0.6404,
+      "step": 479
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.0001466905227958364,
+      "loss": 0.6063,
+      "step": 480
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.00014647231720437686,
+      "loss": 0.7687,
+      "step": 481
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.00014625382902408356,
+      "loss": 0.6441,
+      "step": 482
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.00014603505958353917,
+      "loss": 0.5014,
+      "step": 483
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.0001458160102130369,
+      "loss": 0.5251,
+      "step": 484
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.00014559668224457193,
+      "loss": 0.6059,
+      "step": 485
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.00014537707701183357,
+      "loss": 0.5934,
+      "step": 486
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.00014515719585019726,
+      "loss": 0.5467,
+      "step": 487
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.00014493704009671613,
+      "loss": 0.6691,
+      "step": 488
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.00014471661109011322,
+      "loss": 0.5212,
+      "step": 489
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.000144495910170773,
+      "loss": 0.7621,
+      "step": 490
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.00014427493868073346,
+      "loss": 0.5422,
+      "step": 491
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.00014405369796367795,
+      "loss": 0.4715,
+      "step": 492
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.00014383218936492683,
+      "loss": 0.5964,
+      "step": 493
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.00014361041423142942,
+      "loss": 0.472,
+      "step": 494
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.00014338837391175582,
+      "loss": 0.5631,
+      "step": 495
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.00014316606975608865,
+      "loss": 0.5553,
+      "step": 496
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.00014294350311621482,
+      "loss": 0.5384,
+      "step": 497
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.00014272067534551743,
+      "loss": 0.6568,
+      "step": 498
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.00014249758779896746,
+      "loss": 0.6185,
+      "step": 499
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.00014227424183311545,
+      "loss": 0.5833,
+      "step": 500
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.00014205063880608345,
+      "loss": 0.6853,
+      "step": 501
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.0001418267800775565,
+      "loss": 0.5071,
+      "step": 502
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.00014160266700877471,
+      "loss": 0.5802,
+      "step": 503
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.00014137830096252462,
+      "loss": 0.4858,
+      "step": 504
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.0001411536833031311,
+      "loss": 0.6866,
+      "step": 505
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.00014092881539644908,
+      "loss": 0.5416,
+      "step": 506
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.00014070369860985518,
+      "loss": 0.7359,
+      "step": 507
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.00014047833431223938,
+      "loss": 0.6102,
+      "step": 508
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.00014025272387399674,
+      "loss": 0.4416,
+      "step": 509
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.0001400268686670191,
+      "loss": 0.4473,
+      "step": 510
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.0001398007700646866,
+      "loss": 0.4096,
+      "step": 511
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.00013957442944185946,
+      "loss": 0.6765,
+      "step": 512
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.00013934784817486956,
+      "loss": 0.4677,
+      "step": 513
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.00013912102764151215,
+      "loss": 0.5784,
+      "step": 514
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.0001388939692210374,
+      "loss": 0.6791,
+      "step": 515
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.0001386666742941419,
+      "loss": 0.5077,
+      "step": 516
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.00013843914424296053,
+      "loss": 0.6087,
+      "step": 517
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.0001382113804510579,
+      "loss": 0.5096,
+      "step": 518
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.0001379833843034199,
+      "loss": 0.5513,
+      "step": 519
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.0001377551571864453,
+      "loss": 0.6205,
+      "step": 520
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.00013752670048793744,
+      "loss": 0.5468,
+      "step": 521
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.0001372980155970956,
+      "loss": 0.6585,
+      "step": 522
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.00013706910390450677,
+      "loss": 0.6355,
+      "step": 523
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.00013683996680213695,
+      "loss": 0.7864,
+      "step": 524
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.00013661060568332292,
+      "loss": 0.651,
+      "step": 525
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.00013638102194276356,
+      "loss": 0.5124,
+      "step": 526
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.00013615121697651156,
+      "loss": 0.7163,
+      "step": 527
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.0001359211921819648,
+      "loss": 0.5803,
+      "step": 528
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.00013569094895785782,
+      "loss": 0.541,
+      "step": 529
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.00013546048870425356,
+      "loss": 0.6183,
+      "step": 530
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.00013522981282253452,
+      "loss": 0.4994,
+      "step": 531
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.0001349989227153944,
+      "loss": 0.6614,
+      "step": 532
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.00013476781978682967,
+      "loss": 0.5829,
+      "step": 533
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.00013453650544213076,
+      "loss": 0.6755,
+      "step": 534
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.00013430498108787389,
+      "loss": 0.4746,
+      "step": 535
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.00013407324813191211,
+      "loss": 0.6413,
+      "step": 536
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.00013384130798336705,
+      "loss": 0.6338,
+      "step": 537
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.00013360916205262018,
+      "loss": 0.5983,
+      "step": 538
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.00013337681175130438,
+      "loss": 0.5958,
+      "step": 539
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.00013314425849229509,
+      "loss": 0.6361,
+      "step": 540
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.00013291150368970208,
+      "loss": 0.4946,
+      "step": 541
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.00013267854875886057,
+      "loss": 0.5745,
+      "step": 542
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.0001324453951163227,
+      "loss": 0.6291,
+      "step": 543
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.00013221204417984908,
+      "loss": 0.6195,
+      "step": 544
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.00013197849736839982,
+      "loss": 0.5461,
+      "step": 545
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.00013174475610212624,
+      "loss": 0.5355,
+      "step": 546
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.0001315108218023621,
+      "loss": 0.7082,
+      "step": 547
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.00013127669589161484,
+      "loss": 0.5511,
+      "step": 548
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.00013104237979355722,
+      "loss": 0.5279,
+      "step": 549
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.00013080787493301838,
+      "loss": 0.6608,
+      "step": 550
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.0001305731827359753,
+      "loss": 0.7352,
+      "step": 551
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.00013033830462954414,
+      "loss": 0.5717,
+      "step": 552
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.0001301032420419715,
+      "loss": 0.6129,
+      "step": 553
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.0001298679964026258,
+      "loss": 0.6058,
+      "step": 554
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.0001296325691419886,
+      "loss": 0.6566,
+      "step": 555
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.00012939696169164577,
+      "loss": 0.597,
+      "step": 556
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.00012916117548427897,
+      "loss": 0.6844,
+      "step": 557
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.00012892521195365678,
+      "loss": 0.6438,
+      "step": 558
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.00012868907253462607,
+      "loss": 0.6251,
+      "step": 559
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.00012845275866310324,
+      "loss": 0.5319,
+      "step": 560
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.00012821627177606555,
+      "loss": 0.581,
+      "step": 561
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.0001279796133115423,
+      "loss": 0.5434,
+      "step": 562
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.0001277427847086061,
+      "loss": 0.3277,
+      "step": 563
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.0001275057874073641,
+      "loss": 0.5906,
+      "step": 564
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.00012726862284894938,
+      "loss": 0.7878,
+      "step": 565
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.00012703129247551204,
+      "loss": 0.682,
+      "step": 566
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.00012679379773021038,
+      "loss": 0.7769,
+      "step": 567
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.00012655614005720233,
+      "loss": 0.4693,
+      "step": 568
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.0001263183209016365,
+      "loss": 0.6305,
+      "step": 569
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.00012608034170964342,
+      "loss": 0.7404,
+      "step": 570
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.0001258422039283268,
+      "loss": 0.5719,
+      "step": 571
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.0001256039090057547,
+      "loss": 0.6054,
+      "step": 572
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.00012536545839095074,
+      "loss": 0.4711,
+      "step": 573
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.00012512685353388523,
+      "loss": 0.5795,
+      "step": 574
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.0001248880958854664,
+      "loss": 0.6066,
+      "step": 575
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.0001246491868975316,
+      "loss": 0.5995,
+      "step": 576
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.00012441012802283843,
+      "loss": 0.4882,
+      "step": 577
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.00012417092071505595,
+      "loss": 0.5523,
+      "step": 578
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.0001239315664287558,
+      "loss": 0.6729,
+      "step": 579
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.0001236920666194033,
+      "loss": 0.6734,
+      "step": 580
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.00012345242274334876,
+      "loss": 0.4987,
+      "step": 581
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.0001232126362578185,
+      "loss": 0.5735,
+      "step": 582
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.000122972708620906,
+      "loss": 0.6132,
+      "step": 583
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.00012273264129156306,
+      "loss": 0.4757,
+      "step": 584
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.00012249243572959095,
+      "loss": 0.5593,
+      "step": 585
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.00012225209339563145,
+      "loss": 0.652,
+      "step": 586
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.00012201161575115809,
+      "loss": 0.655,
+      "step": 587
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.00012177100425846712,
+      "loss": 0.64,
+      "step": 588
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.00012153026038066876,
+      "loss": 0.5883,
+      "step": 589
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.00012128938558167824,
+      "loss": 0.6466,
+      "step": 590
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.00012104838132620684,
+      "loss": 0.5959,
+      "step": 591
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.00012080724907975302,
+      "loss": 0.4044,
+      "step": 592
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.00012056599030859366,
+      "loss": 0.5271,
+      "step": 593
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.0001203246064797749,
+      "loss": 0.6267,
+      "step": 594
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.00012008309906110331,
+      "loss": 0.5708,
+      "step": 595
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.00011984146952113708,
+      "loss": 0.6444,
+      "step": 596
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.00011959971932917696,
+      "loss": 0.5842,
+      "step": 597
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.00011935784995525732,
+      "loss": 0.5752,
+      "step": 598
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.00011911586287013725,
+      "loss": 0.4599,
+      "step": 599
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.00011887375954529168,
+      "loss": 0.5428,
+      "step": 600
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.00011863154145290234,
+      "loss": 0.6214,
+      "step": 601
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.00011838921006584884,
+      "loss": 0.5716,
+      "step": 602
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.00011814676685769967,
+      "loss": 0.5751,
+      "step": 603
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.00011790421330270338,
+      "loss": 0.4826,
+      "step": 604
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.00011766155087577944,
+      "loss": 0.7021,
+      "step": 605
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.0001174187810525093,
+      "loss": 0.4642,
+      "step": 606
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.00011717590530912763,
+      "loss": 0.4898,
+      "step": 607
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.00011693292512251299,
+      "loss": 0.6179,
+      "step": 608
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.00011668984197017919,
+      "loss": 0.5781,
+      "step": 609
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.00011644665733026607,
+      "loss": 0.6675,
+      "step": 610
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.00011620337268153061,
+      "loss": 0.5463,
+      "step": 611
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.00011595998950333793,
+      "loss": 0.6657,
+      "step": 612
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.00011571650927565232,
+      "loss": 0.6032,
+      "step": 613
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.00011547293347902812,
+      "loss": 0.7151,
+      "step": 614
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.00011522926359460093,
+      "loss": 0.593,
+      "step": 615
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.00011498550110407836,
+      "loss": 0.6023,
+      "step": 616
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.00011474164748973122,
+      "loss": 0.6489,
+      "step": 617
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.00011449770423438442,
+      "loss": 0.5674,
+      "step": 618
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.00011425367282140786,
+      "loss": 0.5911,
+      "step": 619
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.00011400955473470764,
+      "loss": 0.6187,
+      "step": 620
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.00011376535145871684,
+      "loss": 0.5259,
+      "step": 621
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.00011352106447838655,
+      "loss": 0.6097,
+      "step": 622
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.0001132766952791769,
+      "loss": 0.4309,
+      "step": 623
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.00011303224534704793,
+      "loss": 0.6894,
+      "step": 624
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.00011278771616845061,
+      "loss": 0.4132,
+      "step": 625
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.00011254310923031781,
+      "loss": 0.5026,
+      "step": 626
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.0001122984260200552,
+      "loss": 0.4602,
+      "step": 627
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.0001120536680255323,
+      "loss": 0.5325,
+      "step": 628
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.00011180883673507338,
+      "loss": 0.3364,
+      "step": 629
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.0001115639336374483,
+      "loss": 0.7636,
+      "step": 630
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.0001113189602218637,
+      "loss": 0.6035,
+      "step": 631
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.00011107391797795373,
+      "loss": 0.5702,
+      "step": 632
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.00011082880839577112,
+      "loss": 0.6767,
+      "step": 633
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.00011058363296577803,
+      "loss": 0.5896,
+      "step": 634
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.00011033839317883701,
+      "loss": 0.6034,
+      "step": 635
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.00011009309052620204,
+      "loss": 0.5635,
+      "step": 636
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.00010984772649950932,
+      "loss": 0.6818,
+      "step": 637
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.00010960230259076818,
+      "loss": 0.4099,
+      "step": 638
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.00010935682029235222,
+      "loss": 0.6667,
+      "step": 639
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.00010911128109699002,
+      "loss": 0.664,
+      "step": 640
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.00010886568649775608,
+      "loss": 0.4609,
+      "step": 641
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.00010862003798806196,
+      "loss": 0.5718,
+      "step": 642
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.00010837433706164688,
+      "loss": 0.5682,
+      "step": 643
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.00010812858521256888,
+      "loss": 0.5714,
+      "step": 644
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.00010788278393519564,
+      "loss": 0.6165,
+      "step": 645
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.00010763693472419538,
+      "loss": 0.5231,
+      "step": 646
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.00010739103907452784,
+      "loss": 0.5929,
+      "step": 647
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.00010714509848143508,
+      "loss": 0.6055,
+      "step": 648
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.00010689911444043248,
+      "loss": 0.6752,
+      "step": 649
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.00010665308844729967,
+      "loss": 0.6143,
+      "step": 650
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.0001064070219980713,
+      "loss": 0.603,
+      "step": 651
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.00010616091658902807,
+      "loss": 0.6289,
+      "step": 652
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.00010591477371668756,
+      "loss": 0.62,
+      "step": 653
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.0001056685948777952,
+      "loss": 0.6476,
+      "step": 654
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.00010542238156931509,
+      "loss": 0.6157,
+      "step": 655
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.00010517613528842097,
+      "loss": 0.7172,
+      "step": 656
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.00010492985753248697,
+      "loss": 0.6018,
+      "step": 657
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.0001046835497990788,
+      "loss": 0.5698,
+      "step": 658
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.00010443721358594429,
+      "loss": 0.5755,
+      "step": 659
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.00010419085039100451,
+      "loss": 0.5586,
+      "step": 660
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.00010394446171234466,
+      "loss": 0.6374,
+      "step": 661
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.00010369804904820474,
+      "loss": 0.4623,
+      "step": 662
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.00010345161389697082,
+      "loss": 0.5365,
+      "step": 663
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.00010320515775716555,
+      "loss": 0.6716,
+      "step": 664
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.0001029586821274392,
+      "loss": 0.5875,
+      "step": 665
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.00010271218850656066,
+      "loss": 0.4641,
+      "step": 666
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.00010246567839340817,
+      "loss": 0.6042,
+      "step": 667
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.00010221915328696021,
+      "loss": 0.6493,
+      "step": 668
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.00010197261468628654,
+      "loss": 0.5813,
+      "step": 669
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.00010172606409053886,
+      "loss": 0.4873,
+      "step": 670
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.00010147950299894185,
+      "loss": 0.6622,
+      "step": 671
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.0001012329329107841,
+      "loss": 0.5865,
+      "step": 672
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.00010098635532540872,
+      "loss": 0.5075,
+      "step": 673
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.00010073977174220459,
+      "loss": 0.6902,
+      "step": 674
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.00010049318366059697,
+      "loss": 0.5963,
+      "step": 675
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.00010024659258003848,
+      "loss": 0.6803,
+      "step": 676
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.0001,
+      "loss": 0.4835,
+      "step": 677
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.975340741996156e-05,
+      "loss": 0.5966,
+      "step": 678
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.950681633940305e-05,
+      "loss": 0.43,
+      "step": 679
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.926022825779546e-05,
+      "loss": 0.5717,
+      "step": 680
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.901364467459131e-05,
+      "loss": 0.6375,
+      "step": 681
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.876706708921595e-05,
+      "loss": 0.5587,
+      "step": 682
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.852049700105815e-05,
+      "loss": 0.5731,
+      "step": 683
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.827393590946116e-05,
+      "loss": 0.5774,
+      "step": 684
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.802738531371346e-05,
+      "loss": 0.6526,
+      "step": 685
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.77808467130398e-05,
+      "loss": 0.6426,
+      "step": 686
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.753432160659185e-05,
+      "loss": 0.5733,
+      "step": 687
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.728781149343936e-05,
+      "loss": 0.5751,
+      "step": 688
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.704131787256083e-05,
+      "loss": 0.7615,
+      "step": 689
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.679484224283449e-05,
+      "loss": 0.5689,
+      "step": 690
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.654838610302923e-05,
+      "loss": 0.6123,
+      "step": 691
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.630195095179527e-05,
+      "loss": 0.5531,
+      "step": 692
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.605553828765539e-05,
+      "loss": 0.7153,
+      "step": 693
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.58091496089955e-05,
+      "loss": 0.5583,
+      "step": 694
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.556278641405573e-05,
+      "loss": 0.5586,
+      "step": 695
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.53164502009212e-05,
+      "loss": 0.6765,
+      "step": 696
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.507014246751304e-05,
+      "loss": 0.5783,
+      "step": 697
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.482386471157904e-05,
+      "loss": 0.5004,
+      "step": 698
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.457761843068493e-05,
+      "loss": 0.5693,
+      "step": 699
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.433140512220483e-05,
+      "loss": 0.6927,
+      "step": 700
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.408522628331246e-05,
+      "loss": 0.4887,
+      "step": 701
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.383908341097197e-05,
+      "loss": 0.6689,
+      "step": 702
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.359297800192872e-05,
+      "loss": 0.5364,
+      "step": 703
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.334691155270036e-05,
+      "loss": 0.6707,
+      "step": 704
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.31008855595675e-05,
+      "loss": 0.6863,
+      "step": 705
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.285490151856493e-05,
+      "loss": 0.721,
+      "step": 706
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.260896092547216e-05,
+      "loss": 0.558,
+      "step": 707
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.236306527580464e-05,
+      "loss": 0.6502,
+      "step": 708
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.211721606480434e-05,
+      "loss": 0.7737,
+      "step": 709
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.187141478743113e-05,
+      "loss": 0.5102,
+      "step": 710
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.162566293835316e-05,
+      "loss": 0.4123,
+      "step": 711
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.137996201193805e-05,
+      "loss": 0.5964,
+      "step": 712
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.113431350224395e-05,
+      "loss": 0.4437,
+      "step": 713
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.088871890301001e-05,
+      "loss": 0.6312,
+      "step": 714
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.06431797076478e-05,
+      "loss": 0.5999,
+      "step": 715
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.039769740923183e-05,
+      "loss": 0.6335,
+      "step": 716
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.015227350049071e-05,
+      "loss": 0.5792,
+      "step": 717
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.990690947379795e-05,
+      "loss": 0.5945,
+      "step": 718
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.9661606821163e-05,
+      "loss": 0.4054,
+      "step": 719
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.941636703422197e-05,
+      "loss": 0.6032,
+      "step": 720
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.917119160422889e-05,
+      "loss": 0.5066,
+      "step": 721
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.892608202204628e-05,
+      "loss": 0.4883,
+      "step": 722
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.868103977813631e-05,
+      "loss": 0.5621,
+      "step": 723
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.843606636255174e-05,
+      "loss": 0.5431,
+      "step": 724
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.819116326492664e-05,
+      "loss": 0.6483,
+      "step": 725
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.79463319744677e-05,
+      "loss": 0.6156,
+      "step": 726
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.770157397994481e-05,
+      "loss": 0.6163,
+      "step": 727
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.745689076968222e-05,
+      "loss": 0.6626,
+      "step": 728
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.721228383154939e-05,
+      "loss": 0.3811,
+      "step": 729
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.69677546529521e-05,
+      "loss": 0.7575,
+      "step": 730
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.672330472082314e-05,
+      "loss": 0.5763,
+      "step": 731
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.647893552161347e-05,
+      "loss": 0.6615,
+      "step": 732
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.62346485412832e-05,
+      "loss": 0.6341,
+      "step": 733
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.599044526529238e-05,
+      "loss": 0.5851,
+      "step": 734
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.574632717859219e-05,
+      "loss": 0.3902,
+      "step": 735
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.55022957656156e-05,
+      "loss": 0.5242,
+      "step": 736
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.52583525102688e-05,
+      "loss": 0.5237,
+      "step": 737
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.501449889592164e-05,
+      "loss": 0.6457,
+      "step": 738
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.477073640539909e-05,
+      "loss": 0.4338,
+      "step": 739
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.452706652097186e-05,
+      "loss": 0.6235,
+      "step": 740
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.42834907243477e-05,
+      "loss": 0.5244,
+      "step": 741
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.404001049666211e-05,
+      "loss": 0.6943,
+      "step": 742
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.379662731846942e-05,
+      "loss": 0.5344,
+      "step": 743
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.355334266973397e-05,
+      "loss": 0.6037,
+      "step": 744
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.331015802982082e-05,
+      "loss": 0.4845,
+      "step": 745
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.306707487748705e-05,
+      "loss": 0.7717,
+      "step": 746
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.282409469087239e-05,
+      "loss": 0.5519,
+      "step": 747
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.258121894749073e-05,
+      "loss": 0.5284,
+      "step": 748
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.23384491242206e-05,
+      "loss": 0.5785,
+      "step": 749
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.209578669729663e-05,
+      "loss": 0.5244,
+      "step": 750
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.185323314230032e-05,
+      "loss": 0.5486,
+      "step": 751
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.161078993415117e-05,
+      "loss": 0.4737,
+      "step": 752
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.13684585470977e-05,
+      "loss": 0.5918,
+      "step": 753
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.112624045470835e-05,
+      "loss": 0.6141,
+      "step": 754
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.08841371298628e-05,
+      "loss": 0.3379,
+      "step": 755
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.064215004474272e-05,
+      "loss": 0.5141,
+      "step": 756
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.040028067082305e-05,
+      "loss": 0.4426,
+      "step": 757
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.01585304788629e-05,
+      "loss": 0.661,
+      "step": 758
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 7.991690093889671e-05,
+      "loss": 0.4921,
+      "step": 759
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.967539352022513e-05,
+      "loss": 0.4858,
+      "step": 760
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.943400969140635e-05,
+      "loss": 0.5962,
+      "step": 761
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.919275092024697e-05,
+      "loss": 0.6083,
+      "step": 762
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.89516186737932e-05,
+      "loss": 0.6439,
+      "step": 763
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.87106144183218e-05,
+      "loss": 0.5193,
+      "step": 764
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.846973961933124e-05,
+      "loss": 0.2868,
+      "step": 765
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.822899574153291e-05,
+      "loss": 0.4139,
+      "step": 766
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.798838424884195e-05,
+      "loss": 0.5663,
+      "step": 767
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.774790660436858e-05,
+      "loss": 0.5397,
+      "step": 768
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.750756427040906e-05,
+      "loss": 0.6151,
+      "step": 769
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.726735870843697e-05,
+      "loss": 0.5842,
+      "step": 770
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.702729137909401e-05,
+      "loss": 0.4999,
+      "step": 771
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.678736374218151e-05,
+      "loss": 0.5844,
+      "step": 772
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.654757725665124e-05,
+      "loss": 0.6323,
+      "step": 773
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.630793338059672e-05,
+      "loss": 0.4702,
+      "step": 774
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.606843357124426e-05,
+      "loss": 0.5288,
+      "step": 775
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.582907928494407e-05,
+      "loss": 0.5858,
+      "step": 776
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.55898719771616e-05,
+      "loss": 0.484,
+      "step": 777
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.535081310246841e-05,
+      "loss": 0.6479,
+      "step": 778
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.511190411453364e-05,
+      "loss": 0.4571,
+      "step": 779
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.487314646611479e-05,
+      "loss": 0.4951,
+      "step": 780
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.463454160904928e-05,
+      "loss": 0.5957,
+      "step": 781
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.43960909942453e-05,
+      "loss": 0.4979,
+      "step": 782
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.415779607167321e-05,
+      "loss": 0.5733,
+      "step": 783
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.391965829035659e-05,
+      "loss": 0.5862,
+      "step": 784
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.368167909836352e-05,
+      "loss": 0.4632,
+      "step": 785
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.34438599427977e-05,
+      "loss": 0.5955,
+      "step": 786
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.320620226978965e-05,
+      "loss": 0.4981,
+      "step": 787
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.296870752448801e-05,
+      "loss": 0.6223,
+      "step": 788
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.273137715105063e-05,
+      "loss": 0.5994,
+      "step": 789
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.249421259263592e-05,
+      "loss": 0.731,
+      "step": 790
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.225721529139393e-05,
+      "loss": 0.5325,
+      "step": 791
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.202038668845773e-05,
+      "loss": 0.5118,
+      "step": 792
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.178372822393444e-05,
+      "loss": 0.5064,
+      "step": 793
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.154724133689677e-05,
+      "loss": 0.4535,
+      "step": 794
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.131092746537394e-05,
+      "loss": 0.6072,
+      "step": 795
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.107478804634325e-05,
+      "loss": 0.4077,
+      "step": 796
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.083882451572107e-05,
+      "loss": 0.6195,
+      "step": 797
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.060303830835424e-05,
+      "loss": 0.5435,
+      "step": 798
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.036743085801143e-05,
+      "loss": 0.5076,
+      "step": 799
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 7.01320035973742e-05,
+      "loss": 0.496,
+      "step": 800
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.989675795802853e-05,
+      "loss": 0.7021,
+      "step": 801
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.966169537045589e-05,
+      "loss": 0.5237,
+      "step": 802
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.942681726402473e-05,
+      "loss": 0.4552,
+      "step": 803
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.919212506698163e-05,
+      "loss": 0.385,
+      "step": 804
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.895762020644279e-05,
+      "loss": 0.5672,
+      "step": 805
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.872330410838516e-05,
+      "loss": 0.6631,
+      "step": 806
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.848917819763793e-05,
+      "loss": 0.5667,
+      "step": 807
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.82552438978738e-05,
+      "loss": 0.5194,
+      "step": 808
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.80215026316002e-05,
+      "loss": 0.6219,
+      "step": 809
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.778795582015097e-05,
+      "loss": 0.6502,
+      "step": 810
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.755460488367728e-05,
+      "loss": 0.2574,
+      "step": 811
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.732145124113947e-05,
+      "loss": 0.4165,
+      "step": 812
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.708849631029793e-05,
+      "loss": 0.4889,
+      "step": 813
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.685574150770495e-05,
+      "loss": 0.5851,
+      "step": 814
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.662318824869565e-05,
+      "loss": 0.5219,
+      "step": 815
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.639083794737983e-05,
+      "loss": 0.4303,
+      "step": 816
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.615869201663296e-05,
+      "loss": 0.6366,
+      "step": 817
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.59267518680879e-05,
+      "loss": 0.5489,
+      "step": 818
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.569501891212615e-05,
+      "loss": 0.6061,
+      "step": 819
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.546349455786926e-05,
+      "loss": 0.5419,
+      "step": 820
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.52321802131704e-05,
+      "loss": 0.4464,
+      "step": 821
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.500107728460562e-05,
+      "loss": 0.613,
+      "step": 822
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.477018717746551e-05,
+      "loss": 0.3781,
+      "step": 823
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.453951129574644e-05,
+      "loss": 0.479,
+      "step": 824
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.430905104214218e-05,
+      "loss": 0.6029,
+      "step": 825
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.407880781803521e-05,
+      "loss": 0.6229,
+      "step": 826
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.384878302348845e-05,
+      "loss": 0.4271,
+      "step": 827
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.361897805723645e-05,
+      "loss": 0.6708,
+      "step": 828
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.33893943166771e-05,
+      "loss": 0.5021,
+      "step": 829
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.316003319786309e-05,
+      "loss": 0.4785,
+      "step": 830
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.293089609549325e-05,
+      "loss": 0.5957,
+      "step": 831
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.270198440290443e-05,
+      "loss": 0.5599,
+      "step": 832
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.24732995120626e-05,
+      "loss": 0.4143,
+      "step": 833
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.224484281355473e-05,
+      "loss": 0.5559,
+      "step": 834
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.201661569658012e-05,
+      "loss": 0.6875,
+      "step": 835
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.178861954894212e-05,
+      "loss": 0.5779,
+      "step": 836
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.156085575703945e-05,
+      "loss": 0.3266,
+      "step": 837
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.133332570585812e-05,
+      "loss": 0.5173,
+      "step": 838
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.110603077896265e-05,
+      "loss": 0.312,
+      "step": 839
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.087897235848785e-05,
+      "loss": 0.4248,
+      "step": 840
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.0652151825130486e-05,
+      "loss": 0.4151,
+      "step": 841
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.0425570558140574e-05,
+      "loss": 0.4938,
+      "step": 842
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.019922993531345e-05,
+      "loss": 0.5815,
+      "step": 843
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.9973131332980925e-05,
+      "loss": 0.4496,
+      "step": 844
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.9747276126003257e-05,
+      "loss": 0.6376,
+      "step": 845
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.952166568776062e-05,
+      "loss": 0.6191,
+      "step": 846
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.9296301390144825e-05,
+      "loss": 0.5334,
+      "step": 847
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.90711846035509e-05,
+      "loss": 0.4456,
+      "step": 848
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.884631669686892e-05,
+      "loss": 0.635,
+      "step": 849
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.8621699037475406e-05,
+      "loss": 0.596,
+      "step": 850
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.8397332991225295e-05,
+      "loss": 0.5259,
+      "step": 851
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.817321992244351e-05,
+      "loss": 0.5855,
+      "step": 852
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.794936119391658e-05,
+      "loss": 0.4387,
+      "step": 853
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.7725758166884546e-05,
+      "loss": 0.4881,
+      "step": 854
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.750241220103254e-05,
+      "loss": 0.6042,
+      "step": 855
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.727932465448258e-05,
+      "loss": 0.5094,
+      "step": 856
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.705649688378518e-05,
+      "loss": 0.5251,
+      "step": 857
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.683393024391139e-05,
+      "loss": 0.4743,
+      "step": 858
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.6611626088244194e-05,
+      "loss": 0.7013,
+      "step": 859
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.638958576857058e-05,
+      "loss": 0.5063,
+      "step": 860
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.6167810635073207e-05,
+      "loss": 0.5708,
+      "step": 861
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.594630203632206e-05,
+      "loss": 0.4183,
+      "step": 862
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.572506131926658e-05,
+      "loss": 0.549,
+      "step": 863
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.550408982922706e-05,
+      "loss": 0.4257,
+      "step": 864
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.528338890988681e-05,
+      "loss": 0.5504,
+      "step": 865
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.506295990328385e-05,
+      "loss": 0.4364,
+      "step": 866
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.484280414980277e-05,
+      "loss": 0.587,
+      "step": 867
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.4622922988166426e-05,
+      "loss": 0.5226,
+      "step": 868
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.440331775542813e-05,
+      "loss": 0.5032,
+      "step": 869
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.418398978696312e-05,
+      "loss": 0.588,
+      "step": 870
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.396494041646081e-05,
+      "loss": 0.6562,
+      "step": 871
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.37461709759165e-05,
+      "loss": 0.558,
+      "step": 872
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.3527682795623146e-05,
+      "loss": 0.624,
+      "step": 873
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.3309477204163637e-05,
+      "loss": 0.4503,
+      "step": 874
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.309155552840228e-05,
+      "loss": 0.413,
+      "step": 875
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.28739190934771e-05,
+      "loss": 0.4639,
+      "step": 876
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.265656922279154e-05,
+      "loss": 0.2744,
+      "step": 877
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.243950723800663e-05,
+      "loss": 0.2639,
+      "step": 878
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.222273445903263e-05,
+      "loss": 0.1992,
+      "step": 879
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.200625220402139e-05,
+      "loss": 0.2826,
+      "step": 880
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.179006178935797e-05,
+      "loss": 0.2146,
+      "step": 881
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.157416452965293e-05,
+      "loss": 0.2251,
+      "step": 882
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.1358561737734214e-05,
+      "loss": 0.1948,
+      "step": 883
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.114325472463905e-05,
+      "loss": 0.1372,
+      "step": 884
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.092824479960625e-05,
+      "loss": 0.1991,
+      "step": 885
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.0713533270067936e-05,
+      "loss": 0.2826,
+      "step": 886
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 5.0499121441641864e-05,
+      "loss": 0.2505,
+      "step": 887
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 5.028501061812334e-05,
+      "loss": 0.3082,
+      "step": 888
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 5.0071202101477334e-05,
+      "loss": 0.2937,
+      "step": 889
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.985769719183045e-05,
+      "loss": 0.2785,
+      "step": 890
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.964449718746328e-05,
+      "loss": 0.2436,
+      "step": 891
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.943160338480217e-05,
+      "loss": 0.2034,
+      "step": 892
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.921901707841165e-05,
+      "loss": 0.2472,
+      "step": 893
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.900673956098644e-05,
+      "loss": 0.2935,
+      "step": 894
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.87947721233434e-05,
+      "loss": 0.0893,
+      "step": 895
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.858311605441408e-05,
+      "loss": 0.266,
+      "step": 896
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.837177264123648e-05,
+      "loss": 0.1819,
+      "step": 897
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.8160743168947496e-05,
+      "loss": 0.2738,
+      "step": 898
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.795002892077502e-05,
+      "loss": 0.2068,
+      "step": 899
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.773963117803013e-05,
+      "loss": 0.1475,
+      "step": 900
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.75295512200992e-05,
+      "loss": 0.171,
+      "step": 901
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.7319790324436367e-05,
+      "loss": 0.2322,
+      "step": 902
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.711034976655544e-05,
+      "loss": 0.1909,
+      "step": 903
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.690123082002249e-05,
+      "loss": 0.1659,
+      "step": 904
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.6692434756447876e-05,
+      "loss": 0.275,
+      "step": 905
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.648396284547848e-05,
+      "loss": 0.2653,
+      "step": 906
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.627581635479024e-05,
+      "loss": 0.22,
+      "step": 907
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.606799655008009e-05,
+      "loss": 0.17,
+      "step": 908
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.5860504695058625e-05,
+      "loss": 0.2068,
+      "step": 909
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.565334205144214e-05,
+      "loss": 0.2213,
+      "step": 910
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.544650987894514e-05,
+      "loss": 0.1543,
+      "step": 911
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.524000943527249e-05,
+      "loss": 0.2182,
+      "step": 912
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.5033841976112015e-05,
+      "loss": 0.2541,
+      "step": 913
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.482800875512656e-05,
+      "loss": 0.2914,
+      "step": 914
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.462251102394669e-05,
+      "loss": 0.1802,
+      "step": 915
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.4417350032162894e-05,
+      "loss": 0.2606,
+      "step": 916
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.421252702731791e-05,
+      "loss": 0.1971,
+      "step": 917
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.40080432548994e-05,
+      "loss": 0.2834,
+      "step": 918
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.380389995833206e-05,
+      "loss": 0.2257,
+      "step": 919
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.3600098378970365e-05,
+      "loss": 0.2955,
+      "step": 920
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.3396639756090784e-05,
+      "loss": 0.2264,
+      "step": 921
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.3193525326884435e-05,
+      "loss": 0.2673,
+      "step": 922
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.29907563264493e-05,
+      "loss": 0.1629,
+      "step": 923
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.278833398778306e-05,
+      "loss": 0.1767,
+      "step": 924
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.258625954177525e-05,
+      "loss": 0.0968,
+      "step": 925
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.238453421720006e-05,
+      "loss": 0.2329,
+      "step": 926
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.218315924070874e-05,
+      "loss": 0.2749,
+      "step": 927
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.198213583682202e-05,
+      "loss": 0.2321,
+      "step": 928
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.1781465227922957e-05,
+      "loss": 0.2391,
+      "step": 929
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.1581148634249176e-05,
+      "loss": 0.1415,
+      "step": 930
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.1381187273885726e-05,
+      "loss": 0.1352,
+      "step": 931
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.1181582362757475e-05,
+      "loss": 0.2324,
+      "step": 932
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.098233511462189e-05,
+      "loss": 0.1912,
+      "step": 933
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.078344674106141e-05,
+      "loss": 0.1645,
+      "step": 934
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.058491845147641e-05,
+      "loss": 0.2053,
+      "step": 935
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.038675145307747e-05,
+      "loss": 0.2527,
+      "step": 936
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.0188946950878404e-05,
+      "loss": 0.2433,
+      "step": 937
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.999150614768869e-05,
+      "loss": 0.1732,
+      "step": 938
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.9794430244106196e-05,
+      "loss": 0.1512,
+      "step": 939
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.959772043850998e-05,
+      "loss": 0.2331,
+      "step": 940
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.9401377927052816e-05,
+      "loss": 0.2367,
+      "step": 941
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.9205403903654156e-05,
+      "loss": 0.255,
+      "step": 942
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.900979955999271e-05,
+      "loss": 0.2429,
+      "step": 943
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.881456608549926e-05,
+      "loss": 0.1963,
+      "step": 944
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.861970466734931e-05,
+      "loss": 0.1036,
+      "step": 945
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.8425216490456095e-05,
+      "loss": 0.1508,
+      "step": 946
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.823110273746313e-05,
+      "loss": 0.1603,
+      "step": 947
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.803736458873722e-05,
+      "loss": 0.2099,
+      "step": 948
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.78440032223612e-05,
+      "loss": 0.2079,
+      "step": 949
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.7651019814126654e-05,
+      "loss": 0.1721,
+      "step": 950
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.745841553752703e-05,
+      "loss": 0.1555,
+      "step": 951
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.72661915637502e-05,
+      "loss": 0.2968,
+      "step": 952
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.707434906167161e-05,
+      "loss": 0.1047,
+      "step": 953
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.688288919784702e-05,
+      "loss": 0.1934,
+      "step": 954
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.669181313650545e-05,
+      "loss": 0.1916,
+      "step": 955
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.650112203954202e-05,
+      "loss": 0.1549,
+      "step": 956
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.6310817066511105e-05,
+      "loss": 0.1003,
+      "step": 957
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.6120899374618955e-05,
+      "loss": 0.2804,
+      "step": 958
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.593137011871698e-05,
+      "loss": 0.1772,
+      "step": 959
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.574223045129459e-05,
+      "loss": 0.1746,
+      "step": 960
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.5553481522472066e-05,
+      "loss": 0.263,
+      "step": 961
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.536512447999383e-05,
+      "loss": 0.2504,
+      "step": 962
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.517716046922118e-05,
+      "loss": 0.1442,
+      "step": 963
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.498959063312558e-05,
+      "loss": 0.1643,
+      "step": 964
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.4802416112281564e-05,
+      "loss": 0.265,
+      "step": 965
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.461563804485985e-05,
+      "loss": 0.2275,
+      "step": 966
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.442925756662032e-05,
+      "loss": 0.2501,
+      "step": 967
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.4243275810905314e-05,
+      "loss": 0.2823,
+      "step": 968
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.40576939086325e-05,
+      "loss": 0.2683,
+      "step": 969
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.387251298828822e-05,
+      "loss": 0.2341,
+      "step": 970
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.36877341759205e-05,
+      "loss": 0.1833,
+      "step": 971
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.350335859513216e-05,
+      "loss": 0.123,
+      "step": 972
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.331938736707415e-05,
+      "loss": 0.2159,
+      "step": 973
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.3135821610438545e-05,
+      "loss": 0.2506,
+      "step": 974
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.2952662441451885e-05,
+      "loss": 0.1828,
+      "step": 975
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.276991097386831e-05,
+      "loss": 0.2228,
+      "step": 976
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.2587568318962866e-05,
+      "loss": 0.1768,
+      "step": 977
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.2405635585524565e-05,
+      "loss": 0.0988,
+      "step": 978
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.2224113879849915e-05,
+      "loss": 0.1921,
+      "step": 979
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.20430043057359e-05,
+      "loss": 0.2149,
+      "step": 980
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.186230796447354e-05,
+      "loss": 0.2384,
+      "step": 981
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.168202595484103e-05,
+      "loss": 0.1974,
+      "step": 982
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.1502159373097053e-05,
+      "loss": 0.2374,
+      "step": 983
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.132270931297423e-05,
+      "loss": 0.2636,
+      "step": 984
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.114367686567228e-05,
+      "loss": 0.2145,
+      "step": 985
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.09650631198517e-05,
+      "loss": 0.1412,
+      "step": 986
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.0786869161626744e-05,
+      "loss": 0.2153,
+      "step": 987
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.060909607455916e-05,
+      "loss": 0.2129,
+      "step": 988
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.0431744939651364e-05,
+      "loss": 0.2433,
+      "step": 989
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.025481683534006e-05,
+      "loss": 0.2356,
+      "step": 990
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.0078312837489487e-05,
+      "loss": 0.2208,
+      "step": 991
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 2.9902234019385057e-05,
+      "loss": 0.2511,
+      "step": 992
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.9726581451726766e-05,
+      "loss": 0.3181,
+      "step": 993
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.955135620262254e-05,
+      "loss": 0.1662,
+      "step": 994
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.937655933758202e-05,
+      "loss": 0.1374,
+      "step": 995
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.920219191950979e-05,
+      "loss": 0.1709,
+      "step": 996
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.9028255008699213e-05,
+      "loss": 0.1662,
+      "step": 997
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.8854749662825664e-05,
+      "loss": 0.2204,
+      "step": 998
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.8681676936940393e-05,
+      "loss": 0.1645,
+      "step": 999
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.850903788346383e-05,
+      "loss": 0.1947,
+      "step": 1000
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.8336833552179443e-05,
+      "loss": 0.2298,
+      "step": 1001
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.8165064990227252e-05,
+      "loss": 0.2329,
+      "step": 1002
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.799373324209734e-05,
+      "loss": 0.213,
+      "step": 1003
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.7822839349623743e-05,
+      "loss": 0.253,
+      "step": 1004
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.7652384351977856e-05,
+      "loss": 0.2194,
+      "step": 1005
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.7482369285662378e-05,
+      "loss": 0.2741,
+      "step": 1006
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.7312795184504715e-05,
+      "loss": 0.2357,
+      "step": 1007
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.714366307965105e-05,
+      "loss": 0.2567,
+      "step": 1008
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.6974973999559682e-05,
+      "loss": 0.2402,
+      "step": 1009
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.680672896999512e-05,
+      "loss": 0.0844,
+      "step": 1010
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.6638929014021552e-05,
+      "loss": 0.0055,
+      "step": 1011
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.6471575151996865e-05,
+      "loss": 0.1922,
+      "step": 1012
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.6304668401566335e-05,
+      "loss": 0.0577,
+      "step": 1013
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.6138209777656352e-05,
+      "loss": 0.1469,
+      "step": 1014
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.5972200292468464e-05,
+      "loss": 0.2158,
+      "step": 1015
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.5806640955472972e-05,
+      "loss": 0.2362,
+      "step": 1016
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.5641532773402998e-05,
+      "loss": 0.1729,
+      "step": 1017
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.5476876750248257e-05,
+      "loss": 0.2804,
+      "step": 1018
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.5312673887249027e-05,
+      "loss": 0.2605,
+      "step": 1019
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.514892518288988e-05,
+      "loss": 0.1888,
+      "step": 1020
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.4985631632893902e-05,
+      "loss": 0.1773,
+      "step": 1021
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.4822794230216327e-05,
+      "loss": 0.2871,
+      "step": 1022
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.4660413965038753e-05,
+      "loss": 0.2498,
+      "step": 1023
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.4498491824763002e-05,
+      "loss": 0.231,
+      "step": 1024
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.4337028794005057e-05,
+      "loss": 0.2382,
+      "step": 1025
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.417602585458927e-05,
+      "loss": 0.1792,
+      "step": 1026
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.401548398554213e-05,
+      "loss": 0.2354,
+      "step": 1027
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.3855404163086558e-05,
+      "loss": 0.2113,
+      "step": 1028
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.3695787360635823e-05,
+      "loss": 0.1993,
+      "step": 1029
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.3536634548787694e-05,
+      "loss": 0.1822,
+      "step": 1030
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.337794669531842e-05,
+      "loss": 0.2427,
+      "step": 1031
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.321972476517705e-05,
+      "loss": 0.2494,
+      "step": 1032
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.3061969720479326e-05,
+      "loss": 0.1765,
+      "step": 1033
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.290468252050204e-05,
+      "loss": 0.224,
+      "step": 1034
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.274786412167711e-05,
+      "loss": 0.2247,
+      "step": 1035
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.25915154775857e-05,
+      "loss": 0.118,
+      "step": 1036
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.243563753895259e-05,
+      "loss": 0.1374,
+      "step": 1037
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.2280231253640173e-05,
+      "loss": 0.2533,
+      "step": 1038
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.2125297566642922e-05,
+      "loss": 0.2346,
+      "step": 1039
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.197083742008148e-05,
+      "loss": 0.1625,
+      "step": 1040
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.181685175319702e-05,
+      "loss": 0.1928,
+      "step": 1041
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.1663341502345448e-05,
+      "loss": 0.2175,
+      "step": 1042
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.151030760099182e-05,
+      "loss": 0.2422,
+      "step": 1043
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.1357750979704528e-05,
+      "loss": 0.1991,
+      "step": 1044
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.1205672566149804e-05,
+      "loss": 0.1839,
+      "step": 1045
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.1054073285085962e-05,
+      "loss": 0.1046,
+      "step": 1046
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.0902954058357782e-05,
+      "loss": 0.2108,
+      "step": 1047
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.0752315804890977e-05,
+      "loss": 0.2033,
+      "step": 1048
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.0602159440686497e-05,
+      "loss": 0.1988,
+      "step": 1049
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.0452485878815063e-05,
+      "loss": 0.2709,
+      "step": 1050
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.0303296029411588e-05,
+      "loss": 0.2455,
+      "step": 1051
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.0154590799669627e-05,
+      "loss": 0.1649,
+      "step": 1052
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.0006371093835763e-05,
+      "loss": 0.0902,
+      "step": 1053
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.985863781320435e-05,
+      "loss": 0.1709,
+      "step": 1054
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.971139185611176e-05,
+      "loss": 0.1915,
+      "step": 1055
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9564634117931125e-05,
+      "loss": 0.1739,
+      "step": 1056
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9418365491066837e-05,
+      "loss": 0.1766,
+      "step": 1057
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.9272586864948993e-05,
+      "loss": 0.2287,
+      "step": 1058
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.912729912602822e-05,
+      "loss": 0.1076,
+      "step": 1059
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.898250315777005e-05,
+      "loss": 0.1616,
+      "step": 1060
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.8838199840649716e-05,
+      "loss": 0.1938,
+      "step": 1061
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.8694390052146737e-05,
+      "loss": 0.1587,
+      "step": 1062
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.855107466673961e-05,
+      "loss": 0.2593,
+      "step": 1063
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8408254555900363e-05,
+      "loss": 0.2158,
+      "step": 1064
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8265930588089476e-05,
+      "loss": 0.2148,
+      "step": 1065
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8124103628750388e-05,
+      "loss": 0.1833,
+      "step": 1066
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7982774540304403e-05,
+      "loss": 0.2417,
+      "step": 1067
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7841944182145365e-05,
+      "loss": 0.226,
+      "step": 1068
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7701613410634365e-05,
+      "loss": 0.2438,
+      "step": 1069
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7561783079094705e-05,
+      "loss": 0.1615,
+      "step": 1070
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.7422454037806525e-05,
+      "loss": 0.2321,
+      "step": 1071
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.728362713400178e-05,
+      "loss": 0.1068,
+      "step": 1072
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.7145303211859008e-05,
+      "loss": 0.2583,
+      "step": 1073
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.7007483112498247e-05,
+      "loss": 0.1893,
+      "step": 1074
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.687016767397581e-05,
+      "loss": 0.1601,
+      "step": 1075
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6733357731279377e-05,
+      "loss": 0.2538,
+      "step": 1076
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6597054116322687e-05,
+      "loss": 0.26,
+      "step": 1077
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.64612576579407e-05,
+      "loss": 0.1101,
+      "step": 1078
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6325969181884436e-05,
+      "loss": 0.1726,
+      "step": 1079
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.619118951081594e-05,
+      "loss": 0.1489,
+      "step": 1080
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.6056919464303367e-05,
+      "loss": 0.1859,
+      "step": 1081
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.5923159858815885e-05,
+      "loss": 0.1327,
+      "step": 1082
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.5789911507718826e-05,
+      "loss": 0.1259,
+      "step": 1083
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.5657175221268684e-05,
+      "loss": 0.1762,
+      "step": 1084
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.552495180660818e-05,
+      "loss": 0.1467,
+      "step": 1085
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5393242067761326e-05,
+      "loss": 0.1864,
+      "step": 1086
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5262046805628662e-05,
+      "loss": 0.1327,
+      "step": 1087
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5131366817982185e-05,
+      "loss": 0.1625,
+      "step": 1088
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.5001202899460697e-05,
+      "loss": 0.1634,
+      "step": 1089
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4871555841564887e-05,
+      "loss": 0.2654,
+      "step": 1090
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4742426432652457e-05,
+      "loss": 0.2414,
+      "step": 1091
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4613815457933467e-05,
+      "loss": 0.2181,
+      "step": 1092
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4485723699465392e-05,
+      "loss": 0.2221,
+      "step": 1093
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4358151936148533e-05,
+      "loss": 0.1683,
+      "step": 1094
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4231100943721175e-05,
+      "loss": 0.238,
+      "step": 1095
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4104571494754925e-05,
+      "loss": 0.2,
+      "step": 1096
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.3978564358649927e-05,
+      "loss": 0.1384,
+      "step": 1097
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3853080301630338e-05,
+      "loss": 0.1809,
+      "step": 1098
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.372812008673946e-05,
+      "loss": 0.2029,
+      "step": 1099
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.360368447383532e-05,
+      "loss": 0.2099,
+      "step": 1100
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3479774219585917e-05,
+      "loss": 0.1694,
+      "step": 1101
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.3356390077464587e-05,
+      "loss": 0.1675,
+      "step": 1102
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.3233532797745596e-05,
+      "loss": 0.1922,
+      "step": 1103
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.311120312749935e-05,
+      "loss": 0.175,
+      "step": 1104
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.2989401810588053e-05,
+      "loss": 0.2115,
+      "step": 1105
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.286812958766106e-05,
+      "loss": 0.2005,
+      "step": 1106
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.274738719615044e-05,
+      "loss": 0.2016,
+      "step": 1107
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2627175370266397e-05,
+      "loss": 0.2266,
+      "step": 1108
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2507494840992962e-05,
+      "loss": 0.1236,
+      "step": 1109
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2388346336083379e-05,
+      "loss": 0.2229,
+      "step": 1110
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.2269730580055805e-05,
+      "loss": 0.2306,
+      "step": 1111
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.2151648294188866e-05,
+      "loss": 0.2135,
+      "step": 1112
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.2034100196517206e-05,
+      "loss": 0.185,
+      "step": 1113
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.1917087001827254e-05,
+      "loss": 0.2213,
+      "step": 1114
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.180060942165272e-05,
+      "loss": 0.1765,
+      "step": 1115
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1684668164270418e-05,
+      "loss": 0.2436,
+      "step": 1116
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1569263934695851e-05,
+      "loss": 0.1658,
+      "step": 1117
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1454397434679021e-05,
+      "loss": 0.0351,
+      "step": 1118
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.134006936269999e-05,
+      "loss": 0.2288,
+      "step": 1119
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.1226280413964884e-05,
+      "loss": 0.1378,
+      "step": 1120
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.1113031280401409e-05,
+      "loss": 0.2044,
+      "step": 1121
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.1000322650654837e-05,
+      "loss": 0.1946,
+      "step": 1122
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.088815521008375e-05,
+      "loss": 0.1979,
+      "step": 1123
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0776529640755795e-05,
+      "loss": 0.2315,
+      "step": 1124
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0665446621443708e-05,
+      "loss": 0.2385,
+      "step": 1125
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0554906827620981e-05,
+      "loss": 0.1666,
+      "step": 1126
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.044491093145793e-05,
+      "loss": 0.1671,
+      "step": 1127
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.0335459601817543e-05,
+      "loss": 0.2662,
+      "step": 1128
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.0226553504251401e-05,
+      "loss": 0.1882,
+      "step": 1129
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.0118193300995582e-05,
+      "loss": 0.1017,
+      "step": 1130
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.001037965096676e-05,
+      "loss": 0.1749,
+      "step": 1131
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.903113209758096e-06,
+      "loss": 0.2557,
+      "step": 1132
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.796394629635275e-06,
+      "loss": 0.2487,
+      "step": 1133
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.690224559532612e-06,
+      "loss": 0.1444,
+      "step": 1134
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.58460364504894e-06,
+      "loss": 0.2408,
+      "step": 1135
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.47953252844388e-06,
+      "loss": 0.1939,
+      "step": 1136
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.375011848633764e-06,
+      "loss": 0.2524,
+      "step": 1137
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.2710422411879e-06,
+      "loss": 0.1851,
+      "step": 1138
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.1676243383246e-06,
+      "loss": 0.2296,
+      "step": 1139
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.064758768907377e-06,
+      "loss": 0.2288,
+      "step": 1140
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.962446158441108e-06,
+      "loss": 0.1709,
+      "step": 1141
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.860687129068245e-06,
+      "loss": 0.2003,
+      "step": 1142
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.759482299565003e-06,
+      "loss": 0.2107,
+      "step": 1143
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.65883228533766e-06,
+      "loss": 0.2051,
+      "step": 1144
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.558737698418761e-06,
+      "loss": 0.2017,
+      "step": 1145
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.45919914746337e-06,
+      "loss": 0.2266,
+      "step": 1146
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.360217237745472e-06,
+      "loss": 0.2327,
+      "step": 1147
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.261792571154148e-06,
+      "loss": 0.2191,
+      "step": 1148
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.163925746190082e-06,
+      "loss": 0.1925,
+      "step": 1149
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 8.06661735796177e-06,
+      "loss": 0.253,
+      "step": 1150
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.969867998182024e-06,
+      "loss": 0.1904,
+      "step": 1151
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.87367825516424e-06,
+      "loss": 0.2178,
+      "step": 1152
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.778048713818975e-06,
+      "loss": 0.222,
+      "step": 1153
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.682979955650249e-06,
+      "loss": 0.1368,
+      "step": 1154
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.588472558752114e-06,
+      "loss": 0.2443,
+      "step": 1155
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.494527097805104e-06,
+      "loss": 0.1468,
+      "step": 1156
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.401144144072669e-06,
+      "loss": 0.2342,
+      "step": 1157
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.308324265397836e-06,
+      "loss": 0.1645,
+      "step": 1158
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.216068026199641e-06,
+      "loss": 0.1807,
+      "step": 1159
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.124375987469767e-06,
+      "loss": 0.1215,
+      "step": 1160
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.033248706769102e-06,
+      "loss": 0.1607,
+      "step": 1161
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 6.9426867382243684e-06,
+      "loss": 0.2165,
+      "step": 1162
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.852690632524694e-06,
+      "loss": 0.1695,
+      "step": 1163
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.763260936918325e-06,
+      "loss": 0.1586,
+      "step": 1164
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.674398195209308e-06,
+      "loss": 0.2074,
+      "step": 1165
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.586102947754147e-06,
+      "loss": 0.1315,
+      "step": 1166
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.498375731458528e-06,
+      "loss": 0.213,
+      "step": 1167
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.411217079774012e-06,
+      "loss": 0.2707,
+      "step": 1168
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.324627522694903e-06,
+      "loss": 0.2264,
+      "step": 1169
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.238607586754896e-06,
+      "loss": 0.1265,
+      "step": 1170
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.153157795023956e-06,
+      "loss": 0.1781,
+      "step": 1171
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 6.068278667105132e-06,
+      "loss": 0.1526,
+      "step": 1172
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.9839707191313706e-06,
+      "loss": 0.1672,
+      "step": 1173
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.900234463762366e-06,
+      "loss": 0.1981,
+      "step": 1174
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.81707041018148e-06,
+      "loss": 0.1957,
+      "step": 1175
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.734479064092624e-06,
+      "loss": 0.2222,
+      "step": 1176
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.652460927717218e-06,
+      "loss": 0.194,
+      "step": 1177
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.571016499791093e-06,
+      "loss": 0.2159,
+      "step": 1178
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.490146275561437e-06,
+      "loss": 0.2476,
+      "step": 1179
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.4098507467838955e-06,
+      "loss": 0.1322,
+      "step": 1180
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.3301304017194135e-06,
+      "loss": 0.2163,
+      "step": 1181
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.250985725131441e-06,
+      "loss": 0.1821,
+      "step": 1182
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.172417198282864e-06,
+      "loss": 0.1862,
+      "step": 1183
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.094425298933136e-06,
+      "loss": 0.2165,
+      "step": 1184
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 5.01701050133534e-06,
+      "loss": 0.1419,
+      "step": 1185
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.9401732762333065e-06,
+      "loss": 0.1452,
+      "step": 1186
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.86391409085879e-06,
+      "loss": 0.1948,
+      "step": 1187
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.788233408928589e-06,
+      "loss": 0.2447,
+      "step": 1188
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.713131690641759e-06,
+      "loss": 0.2473,
+      "step": 1189
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.6386093926767405e-06,
+      "loss": 0.1738,
+      "step": 1190
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.564666968188713e-06,
+      "loss": 0.1201,
+      "step": 1191
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.491304866806667e-06,
+      "loss": 0.1836,
+      "step": 1192
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.4185235346308805e-06,
+      "loss": 0.198,
+      "step": 1193
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.34632341422998e-06,
+      "loss": 0.2354,
+      "step": 1194
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.27470494463843e-06,
+      "loss": 0.205,
+      "step": 1195
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.203668561353757e-06,
+      "loss": 0.2448,
+      "step": 1196
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.133214696333942e-06,
+      "loss": 0.1934,
+      "step": 1197
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.063343777994821e-06,
+      "loss": 0.267,
+      "step": 1198
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.99405623120741e-06,
+      "loss": 0.2535,
+      "step": 1199
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.925352477295385e-06,
+      "loss": 0.2343,
+      "step": 1200
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.857232934032473e-06,
+      "loss": 0.1343,
+      "step": 1201
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.789698015639953e-06,
+      "loss": 0.1522,
+      "step": 1202
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.722748132784093e-06,
+      "loss": 0.1969,
+      "step": 1203
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.6563836925737284e-06,
+      "loss": 0.1712,
+      "step": 1204
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.5906050985576645e-06,
+      "loss": 0.1957,
+      "step": 1205
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.525412750722368e-06,
+      "loss": 0.1356,
+      "step": 1206
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.4608070454893894e-06,
+      "loss": 0.1817,
+      "step": 1207
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.3967883757130447e-06,
+      "loss": 0.2089,
+      "step": 1208
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.3333571306780497e-06,
+      "loss": 0.2297,
+      "step": 1209
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.270513696097055e-06,
+      "loss": 0.1586,
+      "step": 1210
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.208258454108393e-06,
+      "loss": 0.1958,
+      "step": 1211
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.146591783273667e-06,
+      "loss": 0.1063,
+      "step": 1212
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.085514058575556e-06,
+      "loss": 0.1937,
+      "step": 1213
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.025025651415414e-06,
+      "loss": 0.1601,
+      "step": 1214
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.9651269296111283e-06,
+      "loss": 0.1627,
+      "step": 1215
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.905818257394799e-06,
+      "loss": 0.185,
+      "step": 1216
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.8470999954105517e-06,
+      "loss": 0.2361,
+      "step": 1217
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.788972500712372e-06,
+      "loss": 0.1271,
+      "step": 1218
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.7314361267618436e-06,
+      "loss": 0.2226,
+      "step": 1219
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.6744912234261677e-06,
+      "loss": 0.1953,
+      "step": 1220
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.618138136975823e-06,
+      "loss": 0.1943,
+      "step": 1221
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.562377210082656e-06,
+      "loss": 0.2166,
+      "step": 1222
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.5072087818176382e-06,
+      "loss": 0.1295,
+      "step": 1223
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.4526331876489338e-06,
+      "loss": 0.2534,
+      "step": 1224
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.3986507594397687e-06,
+      "loss": 0.1885,
+      "step": 1225
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.3452618254464543e-06,
+      "loss": 0.1965,
+      "step": 1226
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.2924667103163878e-06,
+      "loss": 0.1939,
+      "step": 1227
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.240265735086067e-06,
+      "loss": 0.2037,
+      "step": 1228
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.1886592171791454e-06,
+      "loss": 0.1508,
+      "step": 1229
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.137647470404469e-06,
+      "loss": 0.1761,
+      "step": 1230
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.0872308049542877e-06,
+      "loss": 0.1381,
+      "step": 1231
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.03740952740219e-06,
+      "loss": 0.1882,
+      "step": 1232
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.9881839407013823e-06,
+      "loss": 0.2602,
+      "step": 1233
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.9395543441827592e-06,
+      "loss": 0.1877,
+      "step": 1234
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.8915210335531453e-06,
+      "loss": 0.2036,
+      "step": 1235
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.8440843008934561e-06,
+      "loss": 0.1789,
+      "step": 1236
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.797244434656975e-06,
+      "loss": 0.1377,
+      "step": 1237
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.7510017196674777e-06,
+      "loss": 0.1597,
+      "step": 1238
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.7053564371176777e-06,
+      "loss": 0.1465,
+      "step": 1239
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.6603088645673503e-06,
+      "loss": 0.2292,
+      "step": 1240
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.6158592759417334e-06,
+      "loss": 0.2373,
+      "step": 1241
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.572007941529896e-06,
+      "loss": 0.1416,
+      "step": 1242
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.5287551279829504e-06,
+      "loss": 0.056,
+      "step": 1243
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.48610109831262e-06,
+      "loss": 0.2153,
+      "step": 1244
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.4440461118894743e-06,
+      "loss": 0.2421,
+      "step": 1245
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.402590424441441e-06,
+      "loss": 0.2739,
+      "step": 1246
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.361734288052252e-06,
+      "loss": 0.1582,
+      "step": 1247
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.3214779511598664e-06,
+      "loss": 0.1724,
+      "step": 1248
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.2818216585549825e-06,
+      "loss": 0.2197,
+      "step": 1249
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.2427656513795739e-06,
+      "loss": 0.1686,
+      "step": 1250
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.2043101671253554e-06,
+      "loss": 0.226,
+      "step": 1251
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.166455439632419e-06,
+      "loss": 0.2176,
+      "step": 1252
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.1292016990877895e-06,
+      "loss": 0.1974,
+      "step": 1253
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.092549172023949e-06,
+      "loss": 0.1465,
+      "step": 1254
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.0564980813175807e-06,
+      "loss": 0.2335,
+      "step": 1255
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.0210486461881164e-06,
+      "loss": 0.2136,
+      "step": 1256
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.862010821964696e-07,
+      "loss": 0.2296,
+      "step": 1257
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.519556012436815e-07,
+      "loss": 0.129,
+      "step": 1258
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.183124115696439e-07,
+      "loss": 0.1272,
+      "step": 1259
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.852717177518455e-07,
+      "loss": 0.1747,
+      "step": 1260
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.52833720704116e-07,
+      "loss": 0.2419,
+      "step": 1261
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.209986176753948e-07,
+      "loss": 0.1598,
+      "step": 1262
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 7.897666022485539e-07,
+      "loss": 0.2341,
+      "step": 1263
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.591378643392211e-07,
+      "loss": 0.1968,
+      "step": 1264
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.291125901946027e-07,
+      "loss": 0.1352,
+      "step": 1265
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 6.996909623923631e-07,
+      "loss": 0.1498,
+      "step": 1266
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 6.708731598395023e-07,
+      "loss": 0.1641,
+      "step": 1267
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 6.426593577713136e-07,
+      "loss": 0.1855,
+      "step": 1268
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 6.150497277502609e-07,
+      "loss": 0.228,
+      "step": 1269
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.880444376649585e-07,
+      "loss": 0.1935,
+      "step": 1270
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.616436517291602e-07,
+      "loss": 0.2432,
+      "step": 1271
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.358475304807375e-07,
+      "loss": 0.1484,
+      "step": 1272
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 5.106562307807261e-07,
+      "loss": 0.1192,
+      "step": 1273
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.860699058123697e-07,
+      "loss": 0.2084,
+      "step": 1274
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.62088705080177e-07,
+      "loss": 0.2581,
+      "step": 1275
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.3871277440902246e-07,
+      "loss": 0.2945,
+      "step": 1276
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 4.1594225594323575e-07,
+      "loss": 0.0984,
+      "step": 1277
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.937772881457691e-07,
+      "loss": 0.145,
+      "step": 1278
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.7221800579735346e-07,
+      "loss": 0.1107,
+      "step": 1279
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.512645399956549e-07,
+      "loss": 0.1353,
+      "step": 1280
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.309170181544863e-07,
+      "loss": 0.1983,
+      "step": 1281
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.1117556400306336e-07,
+      "loss": 0.1906,
+      "step": 1282
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.9204029758517216e-07,
+      "loss": 0.1499,
+      "step": 1283
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.7351133525855833e-07,
+      "loss": 0.1461,
+      "step": 1284
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.555887896941167e-07,
+      "loss": 0.0998,
+      "step": 1285
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.382727698752474e-07,
+      "loss": 0.1744,
+      "step": 1286
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.2156338109717844e-07,
+      "loss": 0.145,
+      "step": 1287
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.054607249663665e-07,
+      "loss": 0.2109,
+      "step": 1288
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.899648993998304e-07,
+      "loss": 0.1631,
+      "step": 1289
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.7507599862456315e-07,
+      "loss": 0.0877,
+      "step": 1290
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.6079411317698745e-07,
+      "loss": 0.2158,
+      "step": 1291
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.4711932990238987e-07,
+      "loss": 0.2376,
+      "step": 1292
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.340517319543877e-07,
+      "loss": 0.1312,
+      "step": 1293
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.2159139879439618e-07,
+      "loss": 0.2186,
+      "step": 1294
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.0973840619121766e-07,
+      "loss": 0.2251,
+      "step": 1295
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 9.849282622053092e-08,
+      "loss": 0.1674,
+      "step": 1296
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 8.785472726442479e-08,
+      "loss": 0.2063,
+      "step": 1297
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 7.782417401106523e-08,
+      "loss": 0.1794,
+      "step": 1298
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 6.840122745422894e-08,
+      "loss": 0.1144,
+      "step": 1299
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 5.958594489295921e-08,
+      "loss": 0.1742,
+      "step": 1300
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 5.137837993121064e-08,
+      "loss": 0.0999,
+      "step": 1301
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 4.37785824775605e-08,
+      "loss": 0.2051,
+      "step": 1302
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 3.678659874483126e-08,
+      "loss": 0.176,
+      "step": 1303
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.0402471249890705e-08,
+      "loss": 0.1654,
+      "step": 1304
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.4626238813318936e-08,
+      "loss": 0.2154,
+      "step": 1305
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.9457936559241775e-08,
+      "loss": 0.1299,
+      "step": 1306
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.4897595915053242e-08,
+      "loss": 0.1524,
+      "step": 1307
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.0945244611293426e-08,
+      "loss": 0.2642,
+      "step": 1308
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 7.600906681382025e-09,
+      "loss": 0.1853,
+      "step": 1309
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 4.8646024615961465e-09,
+      "loss": 0.2353,
+      "step": 1310
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 2.7363485908371565e-09,
+      "loss": 0.0933,
+      "step": 1311
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.2161580105973792e-09,
+      "loss": 0.1867,
+      "step": 1312
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 3.0403996484906773e-10,
+      "loss": 0.2633,
+      "step": 1313
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 0.0,
+      "loss": 0.1789,
+      "step": 1314
+    },
+    {
+      "epoch": 3.0,
+      "step": 1314,
+      "total_flos": 4.5566885955528294e+17,
+      "train_loss": 0.6405287658054342,
+      "train_runtime": 9827.3567,
+      "train_samples_per_second": 2.137,
+      "train_steps_per_second": 0.134
+    }
+  ],
+  "logging_steps": 1.0,
+  "max_steps": 1314,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 3,
+  "save_steps": 50000,
+  "total_flos": 4.5566885955528294e+17,
+  "train_batch_size": 16,
+  "trial_name": null,
+  "trial_params": null
+}