diff --git "a/checkpoint-13509/trainer_state.json" "b/checkpoint-13509/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/checkpoint-13509/trainer_state.json"
@@ -0,0 +1,81171 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 2.967804607271718,
+  "eval_steps": 1126,
+  "global_step": 13509,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 2e-05,
+      "loss": 1.2902,
+      "step": 1
+    },
+    {
+      "epoch": 0.0,
+      "eval_loss": 0.9115545749664307,
+      "eval_runtime": 322.1856,
+      "eval_samples_per_second": 56.3,
+      "eval_steps_per_second": 28.151,
+      "step": 1
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4e-05,
+      "loss": 1.4118,
+      "step": 2
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6e-05,
+      "loss": 1.256,
+      "step": 3
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8e-05,
+      "loss": 1.3199,
+      "step": 4
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0001,
+      "loss": 1.4236,
+      "step": 5
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00012,
+      "loss": 1.3817,
+      "step": 6
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00014,
+      "loss": 1.3202,
+      "step": 7
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00016,
+      "loss": 1.2792,
+      "step": 8
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00018,
+      "loss": 1.2805,
+      "step": 9
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0002,
+      "loss": 1.2195,
+      "step": 10
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999999847725134,
+      "loss": 1.2479,
+      "step": 11
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0001999999939090053,
+      "loss": 1.305,
+      "step": 12
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0001999999862952621,
+      "loss": 1.1611,
+      "step": 13
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999997563602194,
+      "loss": 1.0523,
+      "step": 14
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999996193128514,
+      "loss": 1.1309,
+      "step": 15
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999994518105213,
+      "loss": 1.2117,
+      "step": 16
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0001999999253853234,
+      "loss": 1.3288,
+      "step": 17
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0001999999025440996,
+      "loss": 1.2277,
+      "step": 18
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999987665738137,
+      "loss": 1.2257,
+      "step": 19
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0001999998477251695,
+      "loss": 1.1263,
+      "step": 20
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0001999998157474649,
+      "loss": 1.1887,
+      "step": 21
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019999978072426857,
+      "loss": 1.2123,
+      "step": 22
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999974265558153,
+      "loss": 1.2075,
+      "step": 23
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999970154140493,
+      "loss": 1.0792,
+      "step": 24
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999965738174007,
+      "loss": 1.2648,
+      "step": 25
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999961017658823,
+      "loss": 1.1978,
+      "step": 26
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999955992595095,
+      "loss": 1.1447,
+      "step": 27
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999950662982964,
+      "loss": 1.2405,
+      "step": 28
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.000199999450288226,
+      "loss": 1.2258,
+      "step": 29
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999939090114172,
+      "loss": 1.1205,
+      "step": 30
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999932846857864,
+      "loss": 1.1179,
+      "step": 31
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999926299053863,
+      "loss": 1.2437,
+      "step": 32
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999919446702367,
+      "loss": 1.0822,
+      "step": 33
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999912289803587,
+      "loss": 1.1449,
+      "step": 34
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999990482835774,
+      "loss": 1.3301,
+      "step": 35
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999897062365055,
+      "loss": 0.9776,
+      "step": 36
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999888991825772,
+      "loss": 1.0474,
+      "step": 37
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999880616740123,
+      "loss": 1.1361,
+      "step": 38
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999871937108378,
+      "loss": 1.115,
+      "step": 39
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999862952930797,
+      "loss": 1.089,
+      "step": 40
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999853664207648,
+      "loss": 1.1859,
+      "step": 41
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999844070939222,
+      "loss": 1.0527,
+      "step": 42
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999834173125804,
+      "loss": 1.015,
+      "step": 43
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.000199998239707677,
+      "loss": 1.2429,
+      "step": 44
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999813463865217,
+      "loss": 1.1172,
+      "step": 45
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999802652418682,
+      "loss": 1.134,
+      "step": 46
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999791536428417,
+      "loss": 1.1653,
+      "step": 47
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999780115894766,
+      "loss": 1.098,
+      "step": 48
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999976839081807,
+      "loss": 1.0869,
+      "step": 49
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999756361198693,
+      "loss": 1.0181,
+      "step": 50
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999744027036998,
+      "loss": 1.1695,
+      "step": 51
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999973138833336,
+      "loss": 1.1122,
+      "step": 52
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999718445088167,
+      "loss": 1.2223,
+      "step": 53
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999705197301808,
+      "loss": 1.0845,
+      "step": 54
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999691644974693,
+      "loss": 1.2205,
+      "step": 55
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999967778810723,
+      "loss": 1.0687,
+      "step": 56
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999966362669984,
+      "loss": 1.2313,
+      "step": 57
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999964916075296,
+      "loss": 1.1096,
+      "step": 58
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999634390267026,
+      "loss": 1.2174,
+      "step": 59
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999619315242488,
+      "loss": 1.2244,
+      "step": 60
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001999960393567981,
+      "loss": 1.1469,
+      "step": 61
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999588251579453,
+      "loss": 1.0302,
+      "step": 62
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.000199995722629419,
+      "loss": 1.3699,
+      "step": 63
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999555969767635,
+      "loss": 1.0839,
+      "step": 64
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999539372057157,
+      "loss": 1.0683,
+      "step": 65
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999522469810969,
+      "loss": 1.1492,
+      "step": 66
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019999505263029586,
+      "loss": 1.0818,
+      "step": 67
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019999487751713531,
+      "loss": 1.1244,
+      "step": 68
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019999469935863342,
+      "loss": 1.0086,
+      "step": 69
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019999451815479557,
+      "loss": 1.2438,
+      "step": 70
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019999433390562732,
+      "loss": 1.1498,
+      "step": 71
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019999414661113423,
+      "loss": 1.1199,
+      "step": 72
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019999395627132204,
+      "loss": 1.1543,
+      "step": 73
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019999376288619652,
+      "loss": 1.2054,
+      "step": 74
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0001999935664557636,
+      "loss": 1.0414,
+      "step": 75
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019999336698002922,
+      "loss": 1.2056,
+      "step": 76
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019999316445899949,
+      "loss": 1.1078,
+      "step": 77
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019999295889268056,
+      "loss": 1.167,
+      "step": 78
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019999275028107869,
+      "loss": 1.0823,
+      "step": 79
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019999253862420022,
+      "loss": 1.2409,
+      "step": 80
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0001999923239220516,
+      "loss": 1.0577,
+      "step": 81
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019999210617463938,
+      "loss": 1.1682,
+      "step": 82
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019999188538197022,
+      "loss": 1.0984,
+      "step": 83
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019999166154405083,
+      "loss": 1.1261,
+      "step": 84
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019999143466088796,
+      "loss": 1.1321,
+      "step": 85
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0001999912047324886,
+      "loss": 0.9803,
+      "step": 86
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019999097175885973,
+      "loss": 1.1959,
+      "step": 87
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019999073574000843,
+      "loss": 1.1837,
+      "step": 88
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0001999904966759419,
+      "loss": 1.0425,
+      "step": 89
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019999025456666743,
+      "loss": 1.0678,
+      "step": 90
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019999000941219234,
+      "loss": 1.1834,
+      "step": 91
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019998976121252417,
+      "loss": 1.178,
+      "step": 92
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019998950996767044,
+      "loss": 1.1079,
+      "step": 93
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0001999892556776388,
+      "loss": 1.1642,
+      "step": 94
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019998899834243702,
+      "loss": 1.1114,
+      "step": 95
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019998873796207294,
+      "loss": 1.2399,
+      "step": 96
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019998847453655443,
+      "loss": 1.1991,
+      "step": 97
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019998820806588956,
+      "loss": 1.2608,
+      "step": 98
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019998793855008644,
+      "loss": 1.1684,
+      "step": 99
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019998766598915328,
+      "loss": 1.0807,
+      "step": 100
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019998739038309837,
+      "loss": 1.1495,
+      "step": 101
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019998711173193013,
+      "loss": 1.1358,
+      "step": 102
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019998683003565702,
+      "loss": 1.0595,
+      "step": 103
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0001999865452942876,
+      "loss": 1.101,
+      "step": 104
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019998625750783058,
+      "loss": 1.1221,
+      "step": 105
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019998596667629472,
+      "loss": 1.152,
+      "step": 106
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019998567279968886,
+      "loss": 1.0205,
+      "step": 107
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019998537587802198,
+      "loss": 1.0357,
+      "step": 108
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019998507591130306,
+      "loss": 1.1513,
+      "step": 109
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019998477289954133,
+      "loss": 1.1591,
+      "step": 110
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0001999844668427459,
+      "loss": 1.1536,
+      "step": 111
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019998415774092623,
+      "loss": 1.0139,
+      "step": 112
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019998384559409161,
+      "loss": 1.1909,
+      "step": 113
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019998353040225162,
+      "loss": 1.0827,
+      "step": 114
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019998321216541584,
+      "loss": 1.1355,
+      "step": 115
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019998289088359394,
+      "loss": 1.0757,
+      "step": 116
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019998256655679572,
+      "loss": 1.0737,
+      "step": 117
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0001999822391850311,
+      "loss": 1.1843,
+      "step": 118
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019998190876830996,
+      "loss": 1.1194,
+      "step": 119
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019998157530664242,
+      "loss": 1.053,
+      "step": 120
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019998123880003864,
+      "loss": 1.122,
+      "step": 121
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019998089924850887,
+      "loss": 0.9631,
+      "step": 122
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019998055665206344,
+      "loss": 1.0605,
+      "step": 123
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019998021101071275,
+      "loss": 0.9988,
+      "step": 124
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019997986232446737,
+      "loss": 1.0981,
+      "step": 125
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0001999795105933379,
+      "loss": 1.1284,
+      "step": 126
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019997915581733509,
+      "loss": 1.1916,
+      "step": 127
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019997879799646967,
+      "loss": 1.1193,
+      "step": 128
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019997843713075262,
+      "loss": 1.0561,
+      "step": 129
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0001999780732201949,
+      "loss": 1.1072,
+      "step": 130
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019997770626480751,
+      "loss": 1.039,
+      "step": 131
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019997733626460175,
+      "loss": 1.0322,
+      "step": 132
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019997696321958884,
+      "loss": 1.1176,
+      "step": 133
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019997658712978016,
+      "loss": 1.0842,
+      "step": 134
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0001999762079951871,
+      "loss": 1.2144,
+      "step": 135
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019997582581582127,
+      "loss": 1.004,
+      "step": 136
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0001999754405916943,
+      "loss": 1.1733,
+      "step": 137
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019997505232281787,
+      "loss": 1.1078,
+      "step": 138
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0001999746610092039,
+      "loss": 1.0405,
+      "step": 139
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019997426665086424,
+      "loss": 1.2085,
+      "step": 140
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0001999738692478109,
+      "loss": 1.2088,
+      "step": 141
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019997346880005598,
+      "loss": 1.3012,
+      "step": 142
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0001999730653076117,
+      "loss": 1.0184,
+      "step": 143
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019997265877049034,
+      "loss": 1.1067,
+      "step": 144
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0001999722491887043,
+      "loss": 1.0186,
+      "step": 145
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019997183656226603,
+      "loss": 1.1453,
+      "step": 146
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019997142089118807,
+      "loss": 1.0723,
+      "step": 147
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019997100217548317,
+      "loss": 1.1119,
+      "step": 148
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019997058041516395,
+      "loss": 1.0785,
+      "step": 149
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019997015561024338,
+      "loss": 1.0005,
+      "step": 150
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019996972776073434,
+      "loss": 1.0229,
+      "step": 151
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019996929686664987,
+      "loss": 1.065,
+      "step": 152
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019996886292800305,
+      "loss": 1.0459,
+      "step": 153
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019996842594480717,
+      "loss": 1.1116,
+      "step": 154
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0001999679859170755,
+      "loss": 1.0312,
+      "step": 155
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019996754284482143,
+      "loss": 1.1563,
+      "step": 156
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019996709672805846,
+      "loss": 1.0179,
+      "step": 157
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0001999666475668002,
+      "loss": 1.0774,
+      "step": 158
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019996619536106027,
+      "loss": 1.0422,
+      "step": 159
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019996574011085251,
+      "loss": 1.0424,
+      "step": 160
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019996528181619077,
+      "loss": 1.0828,
+      "step": 161
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019996482047708903,
+      "loss": 1.0423,
+      "step": 162
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019996435609356125,
+      "loss": 1.0867,
+      "step": 163
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019996388866562164,
+      "loss": 1.1106,
+      "step": 164
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019996341819328444,
+      "loss": 1.0064,
+      "step": 165
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019996294467656392,
+      "loss": 1.1037,
+      "step": 166
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0001999624681154746,
+      "loss": 1.0777,
+      "step": 167
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019996198851003093,
+      "loss": 0.9905,
+      "step": 168
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019996150586024748,
+      "loss": 1.1627,
+      "step": 169
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019996102016613904,
+      "loss": 1.1721,
+      "step": 170
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019996053142772033,
+      "loss": 1.1808,
+      "step": 171
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019996003964500628,
+      "loss": 1.1152,
+      "step": 172
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019995954481801182,
+      "loss": 1.2394,
+      "step": 173
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019995904694675206,
+      "loss": 1.149,
+      "step": 174
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019995854603124214,
+      "loss": 1.0496,
+      "step": 175
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0001999580420714973,
+      "loss": 0.9757,
+      "step": 176
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019995753506753296,
+      "loss": 1.0415,
+      "step": 177
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0001999570250193645,
+      "loss": 1.0024,
+      "step": 178
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019995651192700745,
+      "loss": 1.002,
+      "step": 179
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019995599579047748,
+      "loss": 0.9617,
+      "step": 180
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019995547660979024,
+      "loss": 1.103,
+      "step": 181
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019995495438496162,
+      "loss": 1.0558,
+      "step": 182
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019995442911600747,
+      "loss": 1.049,
+      "step": 183
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0001999539008029438,
+      "loss": 1.0462,
+      "step": 184
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019995336944578674,
+      "loss": 1.1051,
+      "step": 185
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0001999528350445524,
+      "loss": 0.9253,
+      "step": 186
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0001999522975992571,
+      "loss": 1.1318,
+      "step": 187
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0001999517571099172,
+      "loss": 1.0436,
+      "step": 188
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019995121357654918,
+      "loss": 0.9919,
+      "step": 189
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019995066699916958,
+      "loss": 1.0236,
+      "step": 190
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.000199950117377795,
+      "loss": 1.0733,
+      "step": 191
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019994956471244225,
+      "loss": 1.0112,
+      "step": 192
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0001999490090031281,
+      "loss": 0.9201,
+      "step": 193
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0001999484502498695,
+      "loss": 1.0793,
+      "step": 194
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019994788845268354,
+      "loss": 1.0102,
+      "step": 195
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0001999473236115872,
+      "loss": 1.1175,
+      "step": 196
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019994675572659776,
+      "loss": 1.0691,
+      "step": 197
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019994618479773248,
+      "loss": 0.9213,
+      "step": 198
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019994561082500878,
+      "loss": 0.8912,
+      "step": 199
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019994503380844415,
+      "loss": 1.1143,
+      "step": 200
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0001999444537480561,
+      "loss": 1.0468,
+      "step": 201
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019994387064386232,
+      "loss": 1.2166,
+      "step": 202
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019994328449588064,
+      "loss": 1.1189,
+      "step": 203
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019994269530412883,
+      "loss": 1.0865,
+      "step": 204
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0001999421030686248,
+      "loss": 1.0032,
+      "step": 205
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019994150778938674,
+      "loss": 1.0802,
+      "step": 206
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019994090946643262,
+      "loss": 1.0913,
+      "step": 207
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019994030809978075,
+      "loss": 1.11,
+      "step": 208
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019993970368944943,
+      "loss": 1.0943,
+      "step": 209
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019993909623545705,
+      "loss": 1.0812,
+      "step": 210
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0001999384857378221,
+      "loss": 0.9891,
+      "step": 211
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0001999378721965632,
+      "loss": 0.9012,
+      "step": 212
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.000199937255611699,
+      "loss": 1.1478,
+      "step": 213
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019993663598324833,
+      "loss": 0.9896,
+      "step": 214
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019993601331123005,
+      "loss": 1.0891,
+      "step": 215
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019993538759566305,
+      "loss": 1.0028,
+      "step": 216
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0001999347588365665,
+      "loss": 1.0153,
+      "step": 217
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0001999341270339595,
+      "loss": 1.0581,
+      "step": 218
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019993349218786123,
+      "loss": 1.05,
+      "step": 219
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019993285429829112,
+      "loss": 0.9599,
+      "step": 220
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019993221336526854,
+      "loss": 1.1683,
+      "step": 221
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019993156938881304,
+      "loss": 1.1041,
+      "step": 222
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0001999309223689442,
+      "loss": 1.1458,
+      "step": 223
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019993027230568174,
+      "loss": 1.2516,
+      "step": 224
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019992961919904547,
+      "loss": 1.067,
+      "step": 225
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019992896304905526,
+      "loss": 1.0111,
+      "step": 226
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019992830385573112,
+      "loss": 1.1215,
+      "step": 227
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0001999276416190931,
+      "loss": 1.1412,
+      "step": 228
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019992697633916138,
+      "loss": 1.0791,
+      "step": 229
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019992630801595623,
+      "loss": 1.141,
+      "step": 230
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019992563664949795,
+      "loss": 1.0437,
+      "step": 231
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019992496223980705,
+      "loss": 1.0436,
+      "step": 232
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019992428478690405,
+      "loss": 1.025,
+      "step": 233
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0001999236042908096,
+      "loss": 0.9699,
+      "step": 234
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019992292075154437,
+      "loss": 1.0048,
+      "step": 235
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019992223416912924,
+      "loss": 0.9851,
+      "step": 236
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019992154454358507,
+      "loss": 1.0397,
+      "step": 237
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0001999208518749329,
+      "loss": 1.0889,
+      "step": 238
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019992015616319378,
+      "loss": 1.2323,
+      "step": 239
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019991945740838895,
+      "loss": 1.0959,
+      "step": 240
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019991875561053968,
+      "loss": 1.0103,
+      "step": 241
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019991805076966733,
+      "loss": 1.0481,
+      "step": 242
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019991734288579333,
+      "loss": 0.9678,
+      "step": 243
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0001999166319589393,
+      "loss": 1.0728,
+      "step": 244
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019991591798912688,
+      "loss": 1.0046,
+      "step": 245
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019991520097637776,
+      "loss": 1.0618,
+      "step": 246
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019991448092071385,
+      "loss": 0.9787,
+      "step": 247
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019991375782215705,
+      "loss": 0.9806,
+      "step": 248
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019991303168072935,
+      "loss": 1.0656,
+      "step": 249
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0001999123024964529,
+      "loss": 1.0752,
+      "step": 250
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0001999115702693499,
+      "loss": 0.9457,
+      "step": 251
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019991083499944267,
+      "loss": 1.079,
+      "step": 252
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019991009668675355,
+      "loss": 1.1294,
+      "step": 253
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019990935533130508,
+      "loss": 1.0672,
+      "step": 254
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019990861093311982,
+      "loss": 1.0705,
+      "step": 255
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019990786349222042,
+      "loss": 1.0532,
+      "step": 256
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019990711300862967,
+      "loss": 0.9975,
+      "step": 257
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019990635948237043,
+      "loss": 1.0349,
+      "step": 258
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0001999056029134656,
+      "loss": 0.9695,
+      "step": 259
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019990484330193825,
+      "loss": 1.1312,
+      "step": 260
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019990408064781155,
+      "loss": 1.0828,
+      "step": 261
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0001999033149511087,
+      "loss": 0.9631,
+      "step": 262
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.000199902546211853,
+      "loss": 1.1944,
+      "step": 263
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019990177443006787,
+      "loss": 1.0388,
+      "step": 264
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019990099960577685,
+      "loss": 1.0376,
+      "step": 265
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0001999002217390035,
+      "loss": 1.0378,
+      "step": 266
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0001998994408297715,
+      "loss": 1.0984,
+      "step": 267
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019989865687810465,
+      "loss": 1.1549,
+      "step": 268
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019989786988402683,
+      "loss": 1.0767,
+      "step": 269
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019989707984756203,
+      "loss": 1.0515,
+      "step": 270
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019989628676873427,
+      "loss": 0.9439,
+      "step": 271
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0001998954906475677,
+      "loss": 0.9988,
+      "step": 272
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0001998946914840866,
+      "loss": 1.0694,
+      "step": 273
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0001998938892783153,
+      "loss": 1.0115,
+      "step": 274
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0001998930840302782,
+      "loss": 1.0835,
+      "step": 275
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019989227573999986,
+      "loss": 1.0541,
+      "step": 276
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0001998914644075049,
+      "loss": 1.1137,
+      "step": 277
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.000199890650032818,
+      "loss": 1.0041,
+      "step": 278
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.000199889832615964,
+      "loss": 1.113,
+      "step": 279
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019988901215696773,
+      "loss": 0.9553,
+      "step": 280
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019988818865585424,
+      "loss": 1.1098,
+      "step": 281
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019988736211264856,
+      "loss": 1.1043,
+      "step": 282
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019988653252737593,
+      "loss": 1.0639,
+      "step": 283
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019988569990006153,
+      "loss": 1.0304,
+      "step": 284
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019988486423073082,
+      "loss": 1.1525,
+      "step": 285
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019988402551940912,
+      "loss": 1.0849,
+      "step": 286
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0001998831837661221,
+      "loss": 1.0178,
+      "step": 287
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0001998823389708953,
+      "loss": 1.1307,
+      "step": 288
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019988149113375452,
+      "loss": 0.9198,
+      "step": 289
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019988064025472553,
+      "loss": 1.0196,
+      "step": 290
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019987978633383426,
+      "loss": 1.1701,
+      "step": 291
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019987892937110675,
+      "loss": 1.0052,
+      "step": 292
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019987806936656903,
+      "loss": 1.1542,
+      "step": 293
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019987720632024734,
+      "loss": 1.0416,
+      "step": 294
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019987634023216794,
+      "loss": 1.0956,
+      "step": 295
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019987547110235726,
+      "loss": 0.994,
+      "step": 296
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0001998745989308417,
+      "loss": 1.0488,
+      "step": 297
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019987372371764786,
+      "loss": 1.1458,
+      "step": 298
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0001998728454628024,
+      "loss": 1.2264,
+      "step": 299
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019987196416633202,
+      "loss": 1.1417,
+      "step": 300
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0001998710798282636,
+      "loss": 1.0389,
+      "step": 301
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019987019244862406,
+      "loss": 1.0917,
+      "step": 302
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019986930202744045,
+      "loss": 1.1202,
+      "step": 303
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019986840856473984,
+      "loss": 1.1541,
+      "step": 304
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0001998675120605495,
+      "loss": 0.9585,
+      "step": 305
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0001998666125148967,
+      "loss": 1.0979,
+      "step": 306
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0001998657099278088,
+      "loss": 1.0436,
+      "step": 307
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019986480429931339,
+      "loss": 1.0013,
+      "step": 308
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019986389562943796,
+      "loss": 0.9371,
+      "step": 309
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019986298391821018,
+      "loss": 1.0413,
+      "step": 310
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019986206916565785,
+      "loss": 1.1128,
+      "step": 311
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019986115137180885,
+      "loss": 1.1177,
+      "step": 312
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019986023053669108,
+      "loss": 1.0826,
+      "step": 313
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019985930666033264,
+      "loss": 1.0164,
+      "step": 314
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0001998583797427616,
+      "loss": 1.0449,
+      "step": 315
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019985744978400627,
+      "loss": 1.1064,
+      "step": 316
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0001998565167840949,
+      "loss": 1.0844,
+      "step": 317
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019985558074305594,
+      "loss": 1.0365,
+      "step": 318
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0001998546416609179,
+      "loss": 1.1369,
+      "step": 319
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019985369953770933,
+      "loss": 1.0159,
+      "step": 320
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.000199852754373459,
+      "loss": 0.9669,
+      "step": 321
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019985180616819564,
+      "loss": 0.9857,
+      "step": 322
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019985085492194815,
+      "loss": 1.0031,
+      "step": 323
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019984990063474548,
+      "loss": 1.018,
+      "step": 324
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0001998489433066167,
+      "loss": 1.0134,
+      "step": 325
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.000199847982937591,
+      "loss": 1.099,
+      "step": 326
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019984701952769761,
+      "loss": 1.0395,
+      "step": 327
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019984605307696583,
+      "loss": 1.2129,
+      "step": 328
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019984508358542513,
+      "loss": 1.1113,
+      "step": 329
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019984411105310505,
+      "loss": 0.9209,
+      "step": 330
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019984313548003518,
+      "loss": 1.0758,
+      "step": 331
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019984215686624524,
+      "loss": 0.9869,
+      "step": 332
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019984117521176502,
+      "loss": 1.0516,
+      "step": 333
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0001998401905166244,
+      "loss": 0.9819,
+      "step": 334
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019983920278085346,
+      "loss": 1.0401,
+      "step": 335
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0001998382120044822,
+      "loss": 1.0835,
+      "step": 336
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0001998372181875408,
+      "loss": 1.0921,
+      "step": 337
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019983622133005956,
+      "loss": 1.1014,
+      "step": 338
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019983522143206882,
+      "loss": 1.0282,
+      "step": 339
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.000199834218493599,
+      "loss": 0.8881,
+      "step": 340
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0001998332125146807,
+      "loss": 0.9722,
+      "step": 341
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019983220349534454,
+      "loss": 1.0505,
+      "step": 342
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019983119143562124,
+      "loss": 1.0748,
+      "step": 343
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0001998301763355416,
+      "loss": 0.9513,
+      "step": 344
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0001998291581951366,
+      "loss": 0.9727,
+      "step": 345
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019982813701443716,
+      "loss": 1.0634,
+      "step": 346
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019982711279347445,
+      "loss": 1.011,
+      "step": 347
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019982608553227964,
+      "loss": 1.0497,
+      "step": 348
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.000199825055230884,
+      "loss": 0.9937,
+      "step": 349
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019982402188931895,
+      "loss": 1.1273,
+      "step": 350
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019982298550761594,
+      "loss": 1.0143,
+      "step": 351
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0001998219460858065,
+      "loss": 1.0841,
+      "step": 352
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019982090362392232,
+      "loss": 1.0709,
+      "step": 353
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019981985812199513,
+      "loss": 0.9525,
+      "step": 354
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019981880958005682,
+      "loss": 1.1914,
+      "step": 355
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019981775799813925,
+      "loss": 1.0414,
+      "step": 356
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019981670337627447,
+      "loss": 1.0814,
+      "step": 357
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019981564571449462,
+      "loss": 1.2062,
+      "step": 358
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019981458501283191,
+      "loss": 1.088,
+      "step": 359
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019981352127131864,
+      "loss": 1.1215,
+      "step": 360
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019981245448998718,
+      "loss": 1.0735,
+      "step": 361
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019981138466887007,
+      "loss": 1.0002,
+      "step": 362
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019981031180799983,
+      "loss": 1.0581,
+      "step": 363
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019980923590740918,
+      "loss": 0.995,
+      "step": 364
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019980815696713086,
+      "loss": 1.0963,
+      "step": 365
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019980707498719772,
+      "loss": 1.0282,
+      "step": 366
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019980598996764272,
+      "loss": 0.895,
+      "step": 367
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019980490190849896,
+      "loss": 1.1265,
+      "step": 368
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019980381080979953,
+      "loss": 0.9927,
+      "step": 369
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019980271667157764,
+      "loss": 1.0643,
+      "step": 370
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019980161949386665,
+      "loss": 1.0123,
+      "step": 371
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019980051927669993,
+      "loss": 1.0476,
+      "step": 372
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019979941602011106,
+      "loss": 1.0658,
+      "step": 373
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019979830972413355,
+      "loss": 1.1077,
+      "step": 374
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019979720038880116,
+      "loss": 0.9249,
+      "step": 375
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019979608801414762,
+      "loss": 1.0955,
+      "step": 376
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019979497260020688,
+      "loss": 1.1134,
+      "step": 377
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019979385414701283,
+      "loss": 1.2121,
+      "step": 378
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019979273265459963,
+      "loss": 1.219,
+      "step": 379
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019979160812300134,
+      "loss": 1.0623,
+      "step": 380
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019979048055225222,
+      "loss": 1.0029,
+      "step": 381
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019978934994238667,
+      "loss": 1.0378,
+      "step": 382
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019978821629343906,
+      "loss": 1.0174,
+      "step": 383
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019978707960544397,
+      "loss": 0.9817,
+      "step": 384
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.000199785939878436,
+      "loss": 1.0317,
+      "step": 385
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0001997847971124498,
+      "loss": 0.9705,
+      "step": 386
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019978365130752027,
+      "loss": 0.9227,
+      "step": 387
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019978250246368223,
+      "loss": 0.9849,
+      "step": 388
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0001997813505809707,
+      "loss": 1.087,
+      "step": 389
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019978019565942076,
+      "loss": 1.0972,
+      "step": 390
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019977903769906758,
+      "loss": 0.974,
+      "step": 391
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019977787669994644,
+      "loss": 1.089,
+      "step": 392
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019977671266209265,
+      "loss": 1.0408,
+      "step": 393
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019977554558554171,
+      "loss": 1.0783,
+      "step": 394
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019977437547032917,
+      "loss": 0.99,
+      "step": 395
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019977320231649061,
+      "loss": 1.1073,
+      "step": 396
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019977202612406184,
+      "loss": 1.1415,
+      "step": 397
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019977084689307858,
+      "loss": 1.0875,
+      "step": 398
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019976966462357685,
+      "loss": 0.9568,
+      "step": 399
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019976847931559255,
+      "loss": 1.0158,
+      "step": 400
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019976729096916188,
+      "loss": 1.0753,
+      "step": 401
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019976609958432094,
+      "loss": 0.934,
+      "step": 402
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019976490516110607,
+      "loss": 0.9132,
+      "step": 403
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019976370769955363,
+      "loss": 1.198,
+      "step": 404
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019976250719970008,
+      "loss": 1.0733,
+      "step": 405
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.000199761303661582,
+      "loss": 1.0232,
+      "step": 406
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019976009708523606,
+      "loss": 0.993,
+      "step": 407
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019975888747069895,
+      "loss": 1.0855,
+      "step": 408
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019975767481800755,
+      "loss": 0.9751,
+      "step": 409
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019975645912719881,
+      "loss": 1.0736,
+      "step": 410
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019975524039830969,
+      "loss": 0.9777,
+      "step": 411
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019975401863137732,
+      "loss": 0.9479,
+      "step": 412
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019975279382643897,
+      "loss": 1.0235,
+      "step": 413
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0001997515659835319,
+      "loss": 1.0427,
+      "step": 414
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0001997503351026935,
+      "loss": 1.0613,
+      "step": 415
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019974910118396126,
+      "loss": 1.0254,
+      "step": 416
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019974786422737278,
+      "loss": 1.041,
+      "step": 417
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019974662423296571,
+      "loss": 0.983,
+      "step": 418
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019974538120077776,
+      "loss": 1.0705,
+      "step": 419
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0001997441351308469,
+      "loss": 0.9402,
+      "step": 420
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019974288602321103,
+      "loss": 0.9562,
+      "step": 421
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019974163387790817,
+      "loss": 0.9401,
+      "step": 422
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019974037869497643,
+      "loss": 1.1495,
+      "step": 423
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019973912047445414,
+      "loss": 0.8879,
+      "step": 424
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0001997378592163795,
+      "loss": 1.0497,
+      "step": 425
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.000199736594920791,
+      "loss": 1.0893,
+      "step": 426
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019973532758772714,
+      "loss": 0.9018,
+      "step": 427
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019973405721722648,
+      "loss": 1.105,
+      "step": 428
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019973278380932768,
+      "loss": 0.9295,
+      "step": 429
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019973150736406962,
+      "loss": 0.9876,
+      "step": 430
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0001997302278814911,
+      "loss": 1.0989,
+      "step": 431
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0001997289453616311,
+      "loss": 1.1693,
+      "step": 432
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0001997276598045287,
+      "loss": 1.0528,
+      "step": 433
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019972637121022305,
+      "loss": 1.1274,
+      "step": 434
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019972507957875335,
+      "loss": 1.0387,
+      "step": 435
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019972378491015902,
+      "loss": 1.0284,
+      "step": 436
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0001997224872044794,
+      "loss": 1.0907,
+      "step": 437
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019972118646175403,
+      "loss": 1.0149,
+      "step": 438
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019971988268202258,
+      "loss": 1.091,
+      "step": 439
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0001997185758653247,
+      "loss": 0.9837,
+      "step": 440
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019971726601170024,
+      "loss": 1.0079,
+      "step": 441
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019971595312118904,
+      "loss": 1.0189,
+      "step": 442
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019971463719383107,
+      "loss": 1.0573,
+      "step": 443
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0001997133182296665,
+      "loss": 0.9789,
+      "step": 444
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0001997119962287354,
+      "loss": 1.0299,
+      "step": 445
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0001997106711910781,
+      "loss": 1.105,
+      "step": 446
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0001997093431167349,
+      "loss": 1.1845,
+      "step": 447
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019970801200574632,
+      "loss": 1.0333,
+      "step": 448
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0001997066778581528,
+      "loss": 0.9603,
+      "step": 449
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019970534067399504,
+      "loss": 1.0152,
+      "step": 450
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019970400045331375,
+      "loss": 1.1743,
+      "step": 451
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019970265719614972,
+      "loss": 0.9309,
+      "step": 452
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019970131090254393,
+      "loss": 1.0735,
+      "step": 453
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0001996999615725373,
+      "loss": 1.1109,
+      "step": 454
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019969860920617097,
+      "loss": 1.1276,
+      "step": 455
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0001996972538034861,
+      "loss": 1.124,
+      "step": 456
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.000199695895364524,
+      "loss": 1.1048,
+      "step": 457
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.000199694533889326,
+      "loss": 1.1206,
+      "step": 458
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019969316937793358,
+      "loss": 1.1663,
+      "step": 459
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019969180183038832,
+      "loss": 1.0457,
+      "step": 460
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019969043124673185,
+      "loss": 1.088,
+      "step": 461
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019968905762700594,
+      "loss": 1.0235,
+      "step": 462
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019968768097125236,
+      "loss": 1.0731,
+      "step": 463
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0001996863012795131,
+      "loss": 1.0018,
+      "step": 464
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019968491855183013,
+      "loss": 0.8522,
+      "step": 465
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019968353278824558,
+      "loss": 1.097,
+      "step": 466
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019968214398880167,
+      "loss": 0.9954,
+      "step": 467
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019968075215354066,
+      "loss": 1.0728,
+      "step": 468
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.000199679357282505,
+      "loss": 0.9683,
+      "step": 469
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0001996779593757371,
+      "loss": 1.0178,
+      "step": 470
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019967655843327955,
+      "loss": 0.8394,
+      "step": 471
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019967515445517506,
+      "loss": 1.0278,
+      "step": 472
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019967374744146634,
+      "loss": 0.9896,
+      "step": 473
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019967233739219626,
+      "loss": 1.117,
+      "step": 474
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019967092430740776,
+      "loss": 1.0357,
+      "step": 475
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001996695081871439,
+      "loss": 1.1898,
+      "step": 476
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019966808903144773,
+      "loss": 1.0399,
+      "step": 477
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019966666684036256,
+      "loss": 1.1488,
+      "step": 478
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001996652416139317,
+      "loss": 0.9991,
+      "step": 479
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019966381335219846,
+      "loss": 1.0787,
+      "step": 480
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019966238205520646,
+      "loss": 0.951,
+      "step": 481
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019966094772299917,
+      "loss": 1.0184,
+      "step": 482
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019965951035562035,
+      "loss": 0.8474,
+      "step": 483
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019965806995311377,
+      "loss": 1.0127,
+      "step": 484
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001996566265155233,
+      "loss": 1.0143,
+      "step": 485
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019965518004289287,
+      "loss": 0.9819,
+      "step": 486
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019965373053526655,
+      "loss": 1.0461,
+      "step": 487
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019965227799268847,
+      "loss": 0.9182,
+      "step": 488
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001996508224152029,
+      "loss": 0.9934,
+      "step": 489
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019964936380285412,
+      "loss": 0.9373,
+      "step": 490
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019964790215568662,
+      "loss": 1.1639,
+      "step": 491
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019964643747374483,
+      "loss": 0.9973,
+      "step": 492
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019964496975707343,
+      "loss": 1.0917,
+      "step": 493
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001996434990057171,
+      "loss": 1.1485,
+      "step": 494
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001996420252197206,
+      "loss": 0.9961,
+      "step": 495
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019964054839912885,
+      "loss": 1.1573,
+      "step": 496
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001996390685439868,
+      "loss": 1.1322,
+      "step": 497
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019963758565433955,
+      "loss": 0.9241,
+      "step": 498
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019963609973023224,
+      "loss": 1.1385,
+      "step": 499
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001996346107717101,
+      "loss": 0.9839,
+      "step": 500
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019963311877881855,
+      "loss": 1.0313,
+      "step": 501
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019963162375160297,
+      "loss": 1.0274,
+      "step": 502
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019963012569010889,
+      "loss": 1.0905,
+      "step": 503
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019962862459438195,
+      "loss": 1.0738,
+      "step": 504
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019962712046446785,
+      "loss": 1.0839,
+      "step": 505
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019962561330041242,
+      "loss": 1.1587,
+      "step": 506
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019962410310226156,
+      "loss": 1.088,
+      "step": 507
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019962258987006126,
+      "loss": 1.1314,
+      "step": 508
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001996210736038576,
+      "loss": 1.0311,
+      "step": 509
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019961955430369675,
+      "loss": 1.0237,
+      "step": 510
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019961803196962498,
+      "loss": 1.0049,
+      "step": 511
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019961650660168867,
+      "loss": 1.065,
+      "step": 512
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019961497819993428,
+      "loss": 1.0469,
+      "step": 513
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001996134467644083,
+      "loss": 0.8723,
+      "step": 514
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019961191229515745,
+      "loss": 0.9619,
+      "step": 515
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019961037479222844,
+      "loss": 1.0213,
+      "step": 516
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019960883425566806,
+      "loss": 1.0291,
+      "step": 517
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019960729068552323,
+      "loss": 1.0326,
+      "step": 518
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019960574408184098,
+      "loss": 1.0212,
+      "step": 519
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019960419444466844,
+      "loss": 0.9718,
+      "step": 520
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019960264177405277,
+      "loss": 1.0177,
+      "step": 521
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019960108607004122,
+      "loss": 1.0039,
+      "step": 522
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019959952733268122,
+      "loss": 1.0062,
+      "step": 523
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019959796556202024,
+      "loss": 0.9761,
+      "step": 524
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019959640075810584,
+      "loss": 1.0121,
+      "step": 525
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019959483292098567,
+      "loss": 0.9528,
+      "step": 526
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019959326205070744,
+      "loss": 1.0403,
+      "step": 527
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019959168814731907,
+      "loss": 1.0744,
+      "step": 528
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019959011121086845,
+      "loss": 1.0021,
+      "step": 529
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001995885312414036,
+      "loss": 1.1523,
+      "step": 530
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019958694823897264,
+      "loss": 1.1838,
+      "step": 531
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001995853622036238,
+      "loss": 1.0433,
+      "step": 532
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019958377313540534,
+      "loss": 0.9414,
+      "step": 533
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001995821810343657,
+      "loss": 1.0579,
+      "step": 534
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019958058590055332,
+      "loss": 1.2848,
+      "step": 535
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019957898773401684,
+      "loss": 1.1579,
+      "step": 536
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001995773865348049,
+      "loss": 1.1551,
+      "step": 537
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019957578230296626,
+      "loss": 0.9735,
+      "step": 538
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019957417503854981,
+      "loss": 0.9358,
+      "step": 539
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019957256474160447,
+      "loss": 1.0008,
+      "step": 540
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019957095141217925,
+      "loss": 1.138,
+      "step": 541
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019956933505032334,
+      "loss": 0.9996,
+      "step": 542
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019956771565608593,
+      "loss": 1.0492,
+      "step": 543
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019956609322951637,
+      "loss": 1.0338,
+      "step": 544
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019956446777066403,
+      "loss": 1.0514,
+      "step": 545
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019956283927957845,
+      "loss": 1.0691,
+      "step": 546
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001995612077563092,
+      "loss": 0.9185,
+      "step": 547
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019955957320090602,
+      "loss": 1.1187,
+      "step": 548
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019955793561341864,
+      "loss": 1.0387,
+      "step": 549
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019955629499389692,
+      "loss": 0.9267,
+      "step": 550
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019955465134239084,
+      "loss": 1.1182,
+      "step": 551
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019955300465895052,
+      "loss": 1.0413,
+      "step": 552
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.000199551354943626,
+      "loss": 1.1555,
+      "step": 553
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001995497021964676,
+      "loss": 1.0846,
+      "step": 554
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019954804641752563,
+      "loss": 1.0217,
+      "step": 555
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019954638760685055,
+      "loss": 1.0118,
+      "step": 556
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019954472576449281,
+      "loss": 1.143,
+      "step": 557
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019954306089050309,
+      "loss": 1.0524,
+      "step": 558
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019954139298493202,
+      "loss": 0.983,
+      "step": 559
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019953972204783048,
+      "loss": 1.1482,
+      "step": 560
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019953804807924927,
+      "loss": 1.0454,
+      "step": 561
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019953637107923946,
+      "loss": 1.155,
+      "step": 562
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019953469104785207,
+      "loss": 1.0556,
+      "step": 563
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019953300798513827,
+      "loss": 0.9931,
+      "step": 564
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019953132189114934,
+      "loss": 1.1463,
+      "step": 565
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019952963276593662,
+      "loss": 1.1118,
+      "step": 566
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0001995279406095515,
+      "loss": 1.1257,
+      "step": 567
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0001995262454220456,
+      "loss": 1.0753,
+      "step": 568
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0001995245472034705,
+      "loss": 1.0306,
+      "step": 569
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0001995228459538779,
+      "loss": 1.0246,
+      "step": 570
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019952114167331967,
+      "loss": 1.0318,
+      "step": 571
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0001995194343618477,
+      "loss": 0.9452,
+      "step": 572
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019951772401951393,
+      "loss": 1.0459,
+      "step": 573
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019951601064637054,
+      "loss": 1.0389,
+      "step": 574
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0001995142942424696,
+      "loss": 0.9689,
+      "step": 575
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019951257480786344,
+      "loss": 1.0065,
+      "step": 576
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019951085234260447,
+      "loss": 1.1282,
+      "step": 577
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0001995091268467451,
+      "loss": 1.1533,
+      "step": 578
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019950739832033784,
+      "loss": 1.0026,
+      "step": 579
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0001995056667634354,
+      "loss": 1.134,
+      "step": 580
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019950393217609049,
+      "loss": 1.1734,
+      "step": 581
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019950219455835595,
+      "loss": 1.0894,
+      "step": 582
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019950045391028465,
+      "loss": 1.069,
+      "step": 583
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019949871023192968,
+      "loss": 1.0832,
+      "step": 584
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019949696352334408,
+      "loss": 0.9801,
+      "step": 585
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0001994952137845811,
+      "loss": 0.9739,
+      "step": 586
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019949346101569396,
+      "loss": 1.0488,
+      "step": 587
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019949170521673607,
+      "loss": 1.1393,
+      "step": 588
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019948994638776095,
+      "loss": 1.0676,
+      "step": 589
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019948818452882207,
+      "loss": 0.9839,
+      "step": 590
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019948641963997318,
+      "loss": 1.1815,
+      "step": 591
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019948465172126798,
+      "loss": 1.1585,
+      "step": 592
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019948288077276032,
+      "loss": 1.0722,
+      "step": 593
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019948110679450414,
+      "loss": 1.0083,
+      "step": 594
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019947932978655347,
+      "loss": 1.0571,
+      "step": 595
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0001994775497489624,
+      "loss": 1.004,
+      "step": 596
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019947576668178518,
+      "loss": 1.0818,
+      "step": 597
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0001994739805850761,
+      "loss": 1.1108,
+      "step": 598
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019947219145888956,
+      "loss": 0.9509,
+      "step": 599
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019947039930328002,
+      "loss": 1.0255,
+      "step": 600
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019946860411830208,
+      "loss": 0.9485,
+      "step": 601
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019946680590401044,
+      "loss": 1.0609,
+      "step": 602
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0001994650046604598,
+      "loss": 1.1636,
+      "step": 603
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019946320038770508,
+      "loss": 0.997,
+      "step": 604
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0001994613930858012,
+      "loss": 1.0472,
+      "step": 605
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019945958275480318,
+      "loss": 0.9642,
+      "step": 606
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019945776939476618,
+      "loss": 1.034,
+      "step": 607
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019945595300574547,
+      "loss": 1.0707,
+      "step": 608
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019945413358779627,
+      "loss": 1.0383,
+      "step": 609
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019945231114097407,
+      "loss": 0.9245,
+      "step": 610
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019945048566533433,
+      "loss": 1.2035,
+      "step": 611
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019944865716093265,
+      "loss": 0.991,
+      "step": 612
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019944682562782475,
+      "loss": 1.0635,
+      "step": 613
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019944499106606638,
+      "loss": 0.9961,
+      "step": 614
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019944315347571344,
+      "loss": 1.0034,
+      "step": 615
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019944131285682182,
+      "loss": 1.0946,
+      "step": 616
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019943946920944765,
+      "loss": 1.1079,
+      "step": 617
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019943762253364706,
+      "loss": 0.8995,
+      "step": 618
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019943577282947628,
+      "loss": 1.0767,
+      "step": 619
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019943392009699165,
+      "loss": 1.0443,
+      "step": 620
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019943206433624958,
+      "loss": 1.0701,
+      "step": 621
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019943020554730664,
+      "loss": 1.0511,
+      "step": 622
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019942834373021937,
+      "loss": 1.0628,
+      "step": 623
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019942647888504452,
+      "loss": 1.0458,
+      "step": 624
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019942461101183883,
+      "loss": 1.0712,
+      "step": 625
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019942274011065924,
+      "loss": 0.9649,
+      "step": 626
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019942086618156274,
+      "loss": 1.1515,
+      "step": 627
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019941898922460634,
+      "loss": 1.05,
+      "step": 628
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019941710923984722,
+      "loss": 1.0901,
+      "step": 629
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019941522622734265,
+      "loss": 1.0932,
+      "step": 630
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019941334018715,
+      "loss": 1.0134,
+      "step": 631
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019941145111932666,
+      "loss": 0.979,
+      "step": 632
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0001994095590239302,
+      "loss": 1.1313,
+      "step": 633
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019940766390101825,
+      "loss": 0.949,
+      "step": 634
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019940576575064846,
+      "loss": 0.9083,
+      "step": 635
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0001994038645728787,
+      "loss": 1.0535,
+      "step": 636
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019940196036776685,
+      "loss": 0.9848,
+      "step": 637
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019940005313537094,
+      "loss": 0.9914,
+      "step": 638
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019939814287574898,
+      "loss": 0.994,
+      "step": 639
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019939622958895919,
+      "loss": 0.8778,
+      "step": 640
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019939431327505985,
+      "loss": 0.967,
+      "step": 641
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019939239393410932,
+      "loss": 1.0986,
+      "step": 642
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019939047156616603,
+      "loss": 0.9679,
+      "step": 643
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0001993885461712885,
+      "loss": 0.9391,
+      "step": 644
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019938661774953546,
+      "loss": 0.9589,
+      "step": 645
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019938468630096556,
+      "loss": 1.0484,
+      "step": 646
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019938275182563767,
+      "loss": 1.0489,
+      "step": 647
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019938081432361064,
+      "loss": 0.921,
+      "step": 648
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019937887379494352,
+      "loss": 1.0732,
+      "step": 649
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019937693023969546,
+      "loss": 0.9876,
+      "step": 650
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019937498365792556,
+      "loss": 1.0048,
+      "step": 651
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019937303404969314,
+      "loss": 1.0545,
+      "step": 652
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019937108141505758,
+      "loss": 0.9367,
+      "step": 653
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019936912575407837,
+      "loss": 1.1789,
+      "step": 654
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.000199367167066815,
+      "loss": 1.0056,
+      "step": 655
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019936520535332718,
+      "loss": 1.0949,
+      "step": 656
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019936324061367464,
+      "loss": 1.0397,
+      "step": 657
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019936127284791722,
+      "loss": 1.1051,
+      "step": 658
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019935930205611485,
+      "loss": 1.1602,
+      "step": 659
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019935732823832755,
+      "loss": 1.068,
+      "step": 660
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001993553513946154,
+      "loss": 1.0055,
+      "step": 661
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019935337152503864,
+      "loss": 0.9019,
+      "step": 662
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019935138862965757,
+      "loss": 1.0362,
+      "step": 663
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019934940270853254,
+      "loss": 0.9499,
+      "step": 664
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019934741376172408,
+      "loss": 1.0689,
+      "step": 665
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019934542178929275,
+      "loss": 1.0802,
+      "step": 666
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019934342679129918,
+      "loss": 0.9842,
+      "step": 667
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019934142876780418,
+      "loss": 1.0086,
+      "step": 668
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019933942771886855,
+      "loss": 1.0484,
+      "step": 669
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019933742364455328,
+      "loss": 1.0683,
+      "step": 670
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019933541654491935,
+      "loss": 0.9663,
+      "step": 671
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019933340642002795,
+      "loss": 1.0169,
+      "step": 672
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019933139326994023,
+      "loss": 1.0994,
+      "step": 673
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019932937709471758,
+      "loss": 1.1617,
+      "step": 674
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019932735789442133,
+      "loss": 1.032,
+      "step": 675
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.000199325335669113,
+      "loss": 0.9784,
+      "step": 676
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001993233104188542,
+      "loss": 1.0078,
+      "step": 677
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001993212821437066,
+      "loss": 0.9982,
+      "step": 678
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001993192508437319,
+      "loss": 1.1674,
+      "step": 679
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019931721651899204,
+      "loss": 1.0125,
+      "step": 680
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019931517916954897,
+      "loss": 1.0935,
+      "step": 681
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019931313879546474,
+      "loss": 0.9946,
+      "step": 682
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019931109539680146,
+      "loss": 1.0014,
+      "step": 683
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019930904897362138,
+      "loss": 1.0328,
+      "step": 684
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001993069995259868,
+      "loss": 0.9471,
+      "step": 685
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019930494705396018,
+      "loss": 1.0626,
+      "step": 686
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.000199302891557604,
+      "loss": 0.9952,
+      "step": 687
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019930083303698086,
+      "loss": 1.1627,
+      "step": 688
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019929877149215346,
+      "loss": 0.9368,
+      "step": 689
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019929670692318458,
+      "loss": 1.0036,
+      "step": 690
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001992946393301371,
+      "loss": 0.9587,
+      "step": 691
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019929256871307398,
+      "loss": 1.0007,
+      "step": 692
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019929049507205827,
+      "loss": 0.9091,
+      "step": 693
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019928841840715316,
+      "loss": 1.0954,
+      "step": 694
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019928633871842187,
+      "loss": 0.89,
+      "step": 695
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019928425600592773,
+      "loss": 1.1111,
+      "step": 696
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001992821702697342,
+      "loss": 1.0708,
+      "step": 697
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019928008150990479,
+      "loss": 1.1184,
+      "step": 698
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019927798972650305,
+      "loss": 0.9428,
+      "step": 699
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019927589491959278,
+      "loss": 1.0359,
+      "step": 700
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019927379708923771,
+      "loss": 0.9094,
+      "step": 701
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019927169623550176,
+      "loss": 0.9286,
+      "step": 702
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019926959235844893,
+      "loss": 0.8808,
+      "step": 703
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001992674854581433,
+      "loss": 1.0611,
+      "step": 704
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019926537553464893,
+      "loss": 1.0535,
+      "step": 705
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019926326258803018,
+      "loss": 1.0009,
+      "step": 706
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001992611466183514,
+      "loss": 1.0123,
+      "step": 707
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.000199259027625677,
+      "loss": 1.0991,
+      "step": 708
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019925690561007148,
+      "loss": 1.0561,
+      "step": 709
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019925478057159954,
+      "loss": 1.0876,
+      "step": 710
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019925265251032585,
+      "loss": 1.0716,
+      "step": 711
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019925052142631523,
+      "loss": 0.9723,
+      "step": 712
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001992483873196326,
+      "loss": 1.0343,
+      "step": 713
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019924625019034293,
+      "loss": 1.0375,
+      "step": 714
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019924411003851132,
+      "loss": 1.1172,
+      "step": 715
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019924196686420294,
+      "loss": 1.0154,
+      "step": 716
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019923982066748303,
+      "loss": 1.0258,
+      "step": 717
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019923767144841703,
+      "loss": 1.0255,
+      "step": 718
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019923551920707032,
+      "loss": 0.8622,
+      "step": 719
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019923336394350852,
+      "loss": 1.0193,
+      "step": 720
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019923120565779718,
+      "loss": 1.0418,
+      "step": 721
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019922904435000212,
+      "loss": 0.8839,
+      "step": 722
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019922688002018906,
+      "loss": 1.0048,
+      "step": 723
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019922471266842405,
+      "loss": 0.9342,
+      "step": 724
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019922254229477298,
+      "loss": 0.9768,
+      "step": 725
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019922036889930195,
+      "loss": 0.9184,
+      "step": 726
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019921819248207724,
+      "loss": 1.0755,
+      "step": 727
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019921601304316507,
+      "loss": 1.1098,
+      "step": 728
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019921383058263184,
+      "loss": 1.0164,
+      "step": 729
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.000199211645100544,
+      "loss": 0.9153,
+      "step": 730
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019920945659696808,
+      "loss": 0.9595,
+      "step": 731
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001992072650719708,
+      "loss": 1.0414,
+      "step": 732
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019920507052561886,
+      "loss": 1.0516,
+      "step": 733
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001992028729579791,
+      "loss": 1.0021,
+      "step": 734
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019920067236911843,
+      "loss": 0.9898,
+      "step": 735
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001991984687591039,
+      "loss": 0.9922,
+      "step": 736
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001991962621280026,
+      "loss": 1.0281,
+      "step": 737
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019919405247588174,
+      "loss": 1.0822,
+      "step": 738
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019919183980280863,
+      "loss": 1.0445,
+      "step": 739
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019918962410885063,
+      "loss": 0.9972,
+      "step": 740
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019918740539407522,
+      "loss": 1.065,
+      "step": 741
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019918518365855,
+      "loss": 0.9678,
+      "step": 742
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019918295890234263,
+      "loss": 1.048,
+      "step": 743
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0001991807311255208,
+      "loss": 0.9492,
+      "step": 744
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019917850032815247,
+      "loss": 1.0895,
+      "step": 745
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019917626651030546,
+      "loss": 0.9889,
+      "step": 746
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019917402967204787,
+      "loss": 1.018,
+      "step": 747
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019917178981344785,
+      "loss": 0.9904,
+      "step": 748
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019916954693457352,
+      "loss": 1.0342,
+      "step": 749
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019916730103549326,
+      "loss": 0.9015,
+      "step": 750
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019916505211627547,
+      "loss": 1.0607,
+      "step": 751
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0001991628001769886,
+      "loss": 0.9832,
+      "step": 752
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019916054521770127,
+      "loss": 1.1477,
+      "step": 753
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0001991582872384821,
+      "loss": 0.9907,
+      "step": 754
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019915602623939992,
+      "loss": 1.0465,
+      "step": 755
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019915376222052358,
+      "loss": 1.0868,
+      "step": 756
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.000199151495181922,
+      "loss": 1.1518,
+      "step": 757
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019914922512366423,
+      "loss": 1.0705,
+      "step": 758
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0001991469520458194,
+      "loss": 1.0143,
+      "step": 759
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019914467594845677,
+      "loss": 1.0003,
+      "step": 760
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019914239683164564,
+      "loss": 1.1539,
+      "step": 761
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0001991401146954554,
+      "loss": 0.9252,
+      "step": 762
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019913782953995555,
+      "loss": 1.1063,
+      "step": 763
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019913554136521575,
+      "loss": 1.2113,
+      "step": 764
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019913325017130565,
+      "loss": 0.9578,
+      "step": 765
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019913095595829497,
+      "loss": 1.0236,
+      "step": 766
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019912865872625363,
+      "loss": 1.0173,
+      "step": 767
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019912635847525162,
+      "loss": 0.8697,
+      "step": 768
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019912405520535896,
+      "loss": 0.9959,
+      "step": 769
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019912174891664576,
+      "loss": 1.0841,
+      "step": 770
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019911943960918232,
+      "loss": 1.1033,
+      "step": 771
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019911712728303896,
+      "loss": 1.0996,
+      "step": 772
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0001991148119382861,
+      "loss": 1.0417,
+      "step": 773
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019911249357499423,
+      "loss": 1.105,
+      "step": 774
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.000199110172193234,
+      "loss": 1.0663,
+      "step": 775
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019910784779307602,
+      "loss": 1.0074,
+      "step": 776
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019910552037459117,
+      "loss": 0.9711,
+      "step": 777
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019910318993785029,
+      "loss": 1.113,
+      "step": 778
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019910085648292437,
+      "loss": 1.017,
+      "step": 779
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019909852000988445,
+      "loss": 1.0867,
+      "step": 780
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019909618051880174,
+      "loss": 1.0108,
+      "step": 781
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019909383800974738,
+      "loss": 0.932,
+      "step": 782
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019909149248279286,
+      "loss": 0.8733,
+      "step": 783
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0001990891439380095,
+      "loss": 1.0042,
+      "step": 784
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019908679237546885,
+      "loss": 1.0357,
+      "step": 785
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019908443779524258,
+      "loss": 0.9597,
+      "step": 786
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019908208019740232,
+      "loss": 1.018,
+      "step": 787
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019907971958201993,
+      "loss": 0.8547,
+      "step": 788
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001990773559491673,
+      "loss": 1.0137,
+      "step": 789
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019907498929891637,
+      "loss": 1.0264,
+      "step": 790
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019907261963133922,
+      "loss": 1.0669,
+      "step": 791
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019907024694650809,
+      "loss": 1.0481,
+      "step": 792
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019906787124449515,
+      "loss": 1.1006,
+      "step": 793
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019906549252537281,
+      "loss": 1.0673,
+      "step": 794
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019906311078921347,
+      "loss": 0.9853,
+      "step": 795
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001990607260360897,
+      "loss": 0.9805,
+      "step": 796
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019905833826607415,
+      "loss": 1.1128,
+      "step": 797
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019905594747923947,
+      "loss": 1.0133,
+      "step": 798
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019905355367565855,
+      "loss": 1.0214,
+      "step": 799
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001990511568554042,
+      "loss": 0.9423,
+      "step": 800
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001990487570185495,
+      "loss": 1.049,
+      "step": 801
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001990463541651675,
+      "loss": 0.9333,
+      "step": 802
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019904394829533138,
+      "loss": 1.0125,
+      "step": 803
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019904153940911444,
+      "loss": 1.0602,
+      "step": 804
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019903912750658998,
+      "loss": 1.0257,
+      "step": 805
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001990367125878315,
+      "loss": 1.0097,
+      "step": 806
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019903429465291253,
+      "loss": 0.947,
+      "step": 807
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019903187370190673,
+      "loss": 1.0635,
+      "step": 808
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019902944973488782,
+      "loss": 1.077,
+      "step": 809
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001990270227519296,
+      "loss": 1.0346,
+      "step": 810
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019902459275310602,
+      "loss": 0.9948,
+      "step": 811
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019902215973849108,
+      "loss": 0.9556,
+      "step": 812
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019901972370815884,
+      "loss": 1.0617,
+      "step": 813
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001990172846621835,
+      "loss": 1.0169,
+      "step": 814
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001990148426006394,
+      "loss": 1.1031,
+      "step": 815
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019901239752360083,
+      "loss": 1.1524,
+      "step": 816
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001990099494311423,
+      "loss": 1.021,
+      "step": 817
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019900749832333836,
+      "loss": 0.9503,
+      "step": 818
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019900504420026366,
+      "loss": 1.0947,
+      "step": 819
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019900258706199292,
+      "loss": 1.0248,
+      "step": 820
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019900012690860104,
+      "loss": 1.1508,
+      "step": 821
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019899766374016284,
+      "loss": 1.0027,
+      "step": 822
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001989951975567534,
+      "loss": 1.0896,
+      "step": 823
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019899272835844785,
+      "loss": 0.9599,
+      "step": 824
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001989902561453213,
+      "loss": 1.0241,
+      "step": 825
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019898778091744914,
+      "loss": 1.0109,
+      "step": 826
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019898530267490672,
+      "loss": 0.882,
+      "step": 827
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019898282141776944,
+      "loss": 0.9039,
+      "step": 828
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.000198980337146113,
+      "loss": 1.1413,
+      "step": 829
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019897784986001296,
+      "loss": 1.0698,
+      "step": 830
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001989753595595451,
+      "loss": 1.032,
+      "step": 831
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019897286624478527,
+      "loss": 1.173,
+      "step": 832
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019897036991580938,
+      "loss": 1.0034,
+      "step": 833
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0001989678705726935,
+      "loss": 1.0054,
+      "step": 834
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0001989653682155137,
+      "loss": 1.1116,
+      "step": 835
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00019896286284434622,
+      "loss": 1.0241,
+      "step": 836
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00019896035445926736,
+      "loss": 1.1369,
+      "step": 837
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00019895784306035345,
+      "loss": 1.0268,
+      "step": 838
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00019895532864768107,
+      "loss": 1.2064,
+      "step": 839
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00019895281122132676,
+      "loss": 0.9441,
+      "step": 840
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00019895029078136715,
+      "loss": 1.0103,
+      "step": 841
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00019894776732787904,
+      "loss": 0.987,
+      "step": 842
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0001989452408609393,
+      "loss": 0.9868,
+      "step": 843
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0001989427113806248,
+      "loss": 1.0688,
+      "step": 844
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00019894017888701267,
+      "loss": 1.0835,
+      "step": 845
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00019893764338017997,
+      "loss": 1.0364,
+      "step": 846
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00019893510486020395,
+      "loss": 1.0863,
+      "step": 847
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00019893256332716188,
+      "loss": 1.1263,
+      "step": 848
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0001989300187811312,
+      "loss": 0.9823,
+      "step": 849
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0001989274712221894,
+      "loss": 1.0176,
+      "step": 850
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00019892492065041407,
+      "loss": 0.9843,
+      "step": 851
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00019892236706588285,
+      "loss": 1.1007,
+      "step": 852
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00019891981046867358,
+      "loss": 0.9395,
+      "step": 853
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00019891725085886404,
+      "loss": 0.9663,
+      "step": 854
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00019891468823653223,
+      "loss": 1.0504,
+      "step": 855
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0001989121226017562,
+      "loss": 1.0013,
+      "step": 856
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00019890955395461405,
+      "loss": 0.9307,
+      "step": 857
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00019890698229518405,
+      "loss": 0.9661,
+      "step": 858
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0001989044076235445,
+      "loss": 0.9907,
+      "step": 859
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0001989018299397738,
+      "loss": 0.9852,
+      "step": 860
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00019889924924395048,
+      "loss": 1.0689,
+      "step": 861
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0001988966655361531,
+      "loss": 1.0578,
+      "step": 862
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00019889407881646038,
+      "loss": 0.9968,
+      "step": 863
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0001988914890849511,
+      "loss": 0.9799,
+      "step": 864
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00019888889634170408,
+      "loss": 1.0682,
+      "step": 865
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00019888630058679832,
+      "loss": 0.9552,
+      "step": 866
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0001988837018203129,
+      "loss": 0.9412,
+      "step": 867
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00019888110004232692,
+      "loss": 1.1148,
+      "step": 868
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00019887849525291964,
+      "loss": 1.0645,
+      "step": 869
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0001988758874521704,
+      "loss": 1.0504,
+      "step": 870
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00019887327664015855,
+      "loss": 0.975,
+      "step": 871
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0001988706628169637,
+      "loss": 1.0644,
+      "step": 872
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00019886804598266537,
+      "loss": 0.9943,
+      "step": 873
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00019886542613734332,
+      "loss": 1.0266,
+      "step": 874
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00019886280328107732,
+      "loss": 1.0448,
+      "step": 875
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00019886017741394722,
+      "loss": 1.087,
+      "step": 876
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00019885754853603304,
+      "loss": 1.1222,
+      "step": 877
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0001988549166474148,
+      "loss": 0.8913,
+      "step": 878
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00019885228174817266,
+      "loss": 1.1927,
+      "step": 879
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001988496438383869,
+      "loss": 1.0019,
+      "step": 880
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00019884700291813778,
+      "loss": 1.12,
+      "step": 881
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00019884435898750584,
+      "loss": 1.0726,
+      "step": 882
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001988417120465715,
+      "loss": 1.057,
+      "step": 883
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00019883906209541542,
+      "loss": 1.015,
+      "step": 884
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00019883640913411833,
+      "loss": 1.0861,
+      "step": 885
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00019883375316276096,
+      "loss": 1.1061,
+      "step": 886
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00019883109418142422,
+      "loss": 1.0129,
+      "step": 887
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00019882843219018911,
+      "loss": 1.0328,
+      "step": 888
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00019882576718913673,
+      "loss": 1.0679,
+      "step": 889
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001988230991783482,
+      "loss": 1.0265,
+      "step": 890
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00019882042815790474,
+      "loss": 1.0207,
+      "step": 891
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00019881775412788772,
+      "loss": 0.9408,
+      "step": 892
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00019881507708837862,
+      "loss": 0.9345,
+      "step": 893
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00019881239703945896,
+      "loss": 0.9402,
+      "step": 894
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00019880971398121032,
+      "loss": 0.9383,
+      "step": 895
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00019880702791371443,
+      "loss": 1.0246,
+      "step": 896
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001988043388370531,
+      "loss": 1.1404,
+      "step": 897
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00019880164675130827,
+      "loss": 0.9312,
+      "step": 898
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00019879895165656183,
+      "loss": 0.9537,
+      "step": 899
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001987962535528959,
+      "loss": 0.9976,
+      "step": 900
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00019879355244039267,
+      "loss": 0.9807,
+      "step": 901
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00019879084831913443,
+      "loss": 1.1015,
+      "step": 902
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00019878814118920346,
+      "loss": 1.0165,
+      "step": 903
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00019878543105068225,
+      "loss": 1.0344,
+      "step": 904
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00019878271790365335,
+      "loss": 0.8769,
+      "step": 905
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00019878000174819937,
+      "loss": 1.0688,
+      "step": 906
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.000198777282584403,
+      "loss": 0.986,
+      "step": 907
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001987745604123471,
+      "loss": 0.9936,
+      "step": 908
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00019877183523211455,
+      "loss": 1.0023,
+      "step": 909
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00019876910704378834,
+      "loss": 1.1325,
+      "step": 910
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001987663758474516,
+      "loss": 0.9773,
+      "step": 911
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00019876364164318747,
+      "loss": 1.1318,
+      "step": 912
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001987609044310792,
+      "loss": 1.0471,
+      "step": 913
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001987581642112102,
+      "loss": 1.12,
+      "step": 914
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00019875542098366387,
+      "loss": 1.0638,
+      "step": 915
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001987526747485238,
+      "loss": 0.9409,
+      "step": 916
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00019874992550587362,
+      "loss": 1.2854,
+      "step": 917
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00019874717325579707,
+      "loss": 1.0848,
+      "step": 918
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001987444179983779,
+      "loss": 1.0904,
+      "step": 919
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00019874165973370012,
+      "loss": 0.9664,
+      "step": 920
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00019873889846184765,
+      "loss": 1.0406,
+      "step": 921
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00019873613418290464,
+      "loss": 0.9482,
+      "step": 922
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00019873336689695526,
+      "loss": 0.9949,
+      "step": 923
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00019873059660408378,
+      "loss": 0.9624,
+      "step": 924
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00019872782330437454,
+      "loss": 0.9115,
+      "step": 925
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00019872504699791206,
+      "loss": 0.954,
+      "step": 926
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00019872226768478086,
+      "loss": 0.9729,
+      "step": 927
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00019871948536506558,
+      "loss": 1.0564,
+      "step": 928
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00019871670003885096,
+      "loss": 0.8338,
+      "step": 929
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00019871391170622188,
+      "loss": 0.9649,
+      "step": 930
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00019871112036726316,
+      "loss": 0.9306,
+      "step": 931
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00019870832602205988,
+      "loss": 0.9399,
+      "step": 932
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00019870552867069712,
+      "loss": 1.0579,
+      "step": 933
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00019870272831326008,
+      "loss": 1.0458,
+      "step": 934
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00019869992494983403,
+      "loss": 1.0108,
+      "step": 935
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00019869711858050434,
+      "loss": 1.0837,
+      "step": 936
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0001986943092053565,
+      "loss": 1.0125,
+      "step": 937
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0001986914968244761,
+      "loss": 0.9319,
+      "step": 938
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00019868868143794872,
+      "loss": 0.8633,
+      "step": 939
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00019868586304586013,
+      "loss": 0.9914,
+      "step": 940
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0001986830416482962,
+      "loss": 1.1919,
+      "step": 941
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0001986802172453428,
+      "loss": 1.0232,
+      "step": 942
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.000198677389837086,
+      "loss": 1.1635,
+      "step": 943
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00019867455942361187,
+      "loss": 1.0275,
+      "step": 944
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00019867172600500664,
+      "loss": 1.0872,
+      "step": 945
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00019866888958135657,
+      "loss": 0.9097,
+      "step": 946
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00019866605015274807,
+      "loss": 1.047,
+      "step": 947
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00019866320771926756,
+      "loss": 0.9363,
+      "step": 948
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00019866036228100166,
+      "loss": 1.0473,
+      "step": 949
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00019865751383803707,
+      "loss": 0.9747,
+      "step": 950
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00019865466239046042,
+      "loss": 1.028,
+      "step": 951
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00019865180793835868,
+      "loss": 1.0213,
+      "step": 952
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00019864895048181866,
+      "loss": 1.0074,
+      "step": 953
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00019864609002092745,
+      "loss": 1.067,
+      "step": 954
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0001986432265557722,
+      "loss": 0.972,
+      "step": 955
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00019864036008644003,
+      "loss": 0.9114,
+      "step": 956
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0001986374906130183,
+      "loss": 0.9962,
+      "step": 957
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00019863461813559438,
+      "loss": 1.0276,
+      "step": 958
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00019863174265425575,
+      "loss": 1.0105,
+      "step": 959
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00019862886416909,
+      "loss": 0.9962,
+      "step": 960
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00019862598268018475,
+      "loss": 1.1166,
+      "step": 961
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0001986230981876278,
+      "loss": 1.0482,
+      "step": 962
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.000198620210691507,
+      "loss": 0.9868,
+      "step": 963
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00019861732019191028,
+      "loss": 1.0501,
+      "step": 964
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00019861442668892565,
+      "loss": 1.0953,
+      "step": 965
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00019861153018264124,
+      "loss": 0.9627,
+      "step": 966
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00019860863067314527,
+      "loss": 1.0047,
+      "step": 967
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00019860572816052602,
+      "loss": 0.9652,
+      "step": 968
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0001986028226448719,
+      "loss": 1.1034,
+      "step": 969
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00019859991412627146,
+      "loss": 1.0479,
+      "step": 970
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00019859700260481318,
+      "loss": 1.1247,
+      "step": 971
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00019859408808058578,
+      "loss": 0.9702,
+      "step": 972
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00019859117055367803,
+      "loss": 1.1214,
+      "step": 973
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00019858825002417875,
+      "loss": 0.9762,
+      "step": 974
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00019858532649217692,
+      "loss": 0.8806,
+      "step": 975
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00019858239995776153,
+      "loss": 0.9613,
+      "step": 976
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00019857947042102177,
+      "loss": 1.0543,
+      "step": 977
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00019857653788204682,
+      "loss": 1.1201,
+      "step": 978
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00019857360234092598,
+      "loss": 0.9836,
+      "step": 979
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00019857066379774867,
+      "loss": 1.064,
+      "step": 980
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00019856772225260437,
+      "loss": 1.0494,
+      "step": 981
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00019856477770558272,
+      "loss": 1.011,
+      "step": 982
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00019856183015677333,
+      "loss": 1.0167,
+      "step": 983
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.000198558879606266,
+      "loss": 1.0447,
+      "step": 984
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00019855592605415052,
+      "loss": 0.9554,
+      "step": 985
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00019855296950051696,
+      "loss": 0.9658,
+      "step": 986
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00019855000994545527,
+      "loss": 1.0216,
+      "step": 987
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00019854704738905566,
+      "loss": 0.9848,
+      "step": 988
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00019854408183140826,
+      "loss": 0.9927,
+      "step": 989
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00019854111327260342,
+      "loss": 0.9725,
+      "step": 990
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0001985381417127316,
+      "loss": 0.9753,
+      "step": 991
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00019853516715188327,
+      "loss": 0.9991,
+      "step": 992
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.000198532189590149,
+      "loss": 0.9852,
+      "step": 993
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00019852920902761946,
+      "loss": 1.029,
+      "step": 994
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00019852622546438547,
+      "loss": 1.0463,
+      "step": 995
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00019852323890053784,
+      "loss": 0.9557,
+      "step": 996
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00019852024933616756,
+      "loss": 0.8977,
+      "step": 997
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0001985172567713657,
+      "loss": 0.9762,
+      "step": 998
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00019851426120622335,
+      "loss": 0.947,
+      "step": 999
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00019851126264083178,
+      "loss": 1.0603,
+      "step": 1000
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00019850826107528226,
+      "loss": 1.1398,
+      "step": 1001
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00019850525650966626,
+      "loss": 1.0262,
+      "step": 1002
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00019850224894407527,
+      "loss": 0.9733,
+      "step": 1003
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00019849923837860086,
+      "loss": 0.9964,
+      "step": 1004
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00019849622481333471,
+      "loss": 1.0378,
+      "step": 1005
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00019849320824836863,
+      "loss": 0.9517,
+      "step": 1006
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00019849018868379448,
+      "loss": 1.1026,
+      "step": 1007
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00019848716611970421,
+      "loss": 0.9598,
+      "step": 1008
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00019848414055618992,
+      "loss": 0.9964,
+      "step": 1009
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00019848111199334366,
+      "loss": 0.9534,
+      "step": 1010
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00019847808043125774,
+      "loss": 1.042,
+      "step": 1011
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0001984750458700245,
+      "loss": 1.0867,
+      "step": 1012
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00019847200830973627,
+      "loss": 0.969,
+      "step": 1013
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00019846896775048563,
+      "loss": 0.9795,
+      "step": 1014
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00019846592419236518,
+      "loss": 0.951,
+      "step": 1015
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00019846287763546758,
+      "loss": 0.9187,
+      "step": 1016
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00019845982807988565,
+      "loss": 1.0844,
+      "step": 1017
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0001984567755257122,
+      "loss": 0.9395,
+      "step": 1018
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00019845371997304024,
+      "loss": 0.9508,
+      "step": 1019
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00019845066142196283,
+      "loss": 1.0565,
+      "step": 1020
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00019844759987257312,
+      "loss": 0.9553,
+      "step": 1021
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00019844453532496433,
+      "loss": 0.9837,
+      "step": 1022
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00019844146777922981,
+      "loss": 1.0249,
+      "step": 1023
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00019843839723546295,
+      "loss": 1.1107,
+      "step": 1024
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0001984353236937573,
+      "loss": 1.0336,
+      "step": 1025
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00019843224715420648,
+      "loss": 1.1058,
+      "step": 1026
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0001984291676169041,
+      "loss": 1.0705,
+      "step": 1027
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.000198426085081944,
+      "loss": 1.1839,
+      "step": 1028
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00019842299954942014,
+      "loss": 1.0777,
+      "step": 1029
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00019841991101942637,
+      "loss": 0.9628,
+      "step": 1030
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00019841681949205677,
+      "loss": 1.0003,
+      "step": 1031
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00019841372496740554,
+      "loss": 1.0434,
+      "step": 1032
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0001984106274455669,
+      "loss": 0.9434,
+      "step": 1033
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0001984075269266352,
+      "loss": 1.0279,
+      "step": 1034
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0001984044234107048,
+      "loss": 0.945,
+      "step": 1035
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00019840131689787032,
+      "loss": 1.0454,
+      "step": 1036
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00019839820738822627,
+      "loss": 1.0494,
+      "step": 1037
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00019839509488186742,
+      "loss": 1.0857,
+      "step": 1038
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00019839197937888855,
+      "loss": 0.9959,
+      "step": 1039
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00019838886087938451,
+      "loss": 1.0225,
+      "step": 1040
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0001983857393834503,
+      "loss": 1.0244,
+      "step": 1041
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00019838261489118098,
+      "loss": 1.0142,
+      "step": 1042
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0001983794874026717,
+      "loss": 0.9772,
+      "step": 1043
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00019837635691801774,
+      "loss": 1.0036,
+      "step": 1044
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0001983732234373144,
+      "loss": 1.0732,
+      "step": 1045
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0001983700869606571,
+      "loss": 0.9843,
+      "step": 1046
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0001983669474881414,
+      "loss": 0.8867,
+      "step": 1047
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0001983638050198629,
+      "loss": 0.8859,
+      "step": 1048
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00019836065955591728,
+      "loss": 1.0908,
+      "step": 1049
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00019835751109640036,
+      "loss": 1.028,
+      "step": 1050
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00019835435964140805,
+      "loss": 1.1344,
+      "step": 1051
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00019835120519103627,
+      "loss": 1.0504,
+      "step": 1052
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00019834804774538113,
+      "loss": 1.0095,
+      "step": 1053
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00019834488730453877,
+      "loss": 1.0085,
+      "step": 1054
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00019834172386860546,
+      "loss": 1.0874,
+      "step": 1055
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00019833855743767753,
+      "loss": 1.0661,
+      "step": 1056
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00019833538801185139,
+      "loss": 0.9789,
+      "step": 1057
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00019833221559122362,
+      "loss": 1.0536,
+      "step": 1058
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0001983290401758908,
+      "loss": 0.968,
+      "step": 1059
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00019832586176594963,
+      "loss": 1.0388,
+      "step": 1060
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0001983226803614969,
+      "loss": 1.016,
+      "step": 1061
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00019831949596262955,
+      "loss": 0.9242,
+      "step": 1062
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00019831630856944452,
+      "loss": 1.0145,
+      "step": 1063
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0001983131181820389,
+      "loss": 1.0206,
+      "step": 1064
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00019830992480050985,
+      "loss": 1.065,
+      "step": 1065
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0001983067284249546,
+      "loss": 1.1223,
+      "step": 1066
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00019830352905547054,
+      "loss": 0.961,
+      "step": 1067
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00019830032669215507,
+      "loss": 1.0477,
+      "step": 1068
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00019829712133510573,
+      "loss": 1.0822,
+      "step": 1069
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00019829391298442018,
+      "loss": 1.0325,
+      "step": 1070
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00019829070164019603,
+      "loss": 1.0082,
+      "step": 1071
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00019828748730253115,
+      "loss": 0.9571,
+      "step": 1072
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00019828426997152345,
+      "loss": 0.9834,
+      "step": 1073
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00019828104964727088,
+      "loss": 1.0196,
+      "step": 1074
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0001982778263298715,
+      "loss": 1.036,
+      "step": 1075
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00019827460001942354,
+      "loss": 1.0138,
+      "step": 1076
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0001982713707160252,
+      "loss": 1.0459,
+      "step": 1077
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00019826813841977483,
+      "loss": 0.8971,
+      "step": 1078
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00019826490313077088,
+      "loss": 0.9076,
+      "step": 1079
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0001982616648491119,
+      "loss": 1.0233,
+      "step": 1080
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0001982584235748965,
+      "loss": 1.0721,
+      "step": 1081
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00019825517930822337,
+      "loss": 1.033,
+      "step": 1082
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00019825193204919134,
+      "loss": 1.0046,
+      "step": 1083
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00019824868179789932,
+      "loss": 0.9587,
+      "step": 1084
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00019824542855444627,
+      "loss": 1.0734,
+      "step": 1085
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00019824217231893126,
+      "loss": 0.9922,
+      "step": 1086
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00019823891309145343,
+      "loss": 0.9472,
+      "step": 1087
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00019823565087211212,
+      "loss": 0.9197,
+      "step": 1088
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00019823238566100667,
+      "loss": 0.9348,
+      "step": 1089
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00019822911745823646,
+      "loss": 1.0176,
+      "step": 1090
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00019822584626390106,
+      "loss": 1.0695,
+      "step": 1091
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00019822257207810008,
+      "loss": 0.9928,
+      "step": 1092
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00019821929490093326,
+      "loss": 1.0871,
+      "step": 1093
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00019821601473250038,
+      "loss": 1.1234,
+      "step": 1094
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00019821273157290134,
+      "loss": 0.8941,
+      "step": 1095
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00019820944542223615,
+      "loss": 1.0814,
+      "step": 1096
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0001982061562806049,
+      "loss": 0.9734,
+      "step": 1097
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00019820286414810772,
+      "loss": 1.0515,
+      "step": 1098
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00019819956902484488,
+      "loss": 1.0199,
+      "step": 1099
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00019819627091091676,
+      "loss": 0.9387,
+      "step": 1100
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00019819296980642375,
+      "loss": 1.0134,
+      "step": 1101
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00019818966571146646,
+      "loss": 0.9808,
+      "step": 1102
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00019818635862614548,
+      "loss": 1.0903,
+      "step": 1103
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00019818304855056152,
+      "loss": 0.8725,
+      "step": 1104
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0001981797354848154,
+      "loss": 0.9302,
+      "step": 1105
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00019817641942900799,
+      "loss": 1.0761,
+      "step": 1106
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0001981731003832403,
+      "loss": 1.0344,
+      "step": 1107
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00019816977834761343,
+      "loss": 0.9374,
+      "step": 1108
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00019816645332222855,
+      "loss": 0.9979,
+      "step": 1109
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00019816312530718686,
+      "loss": 0.9319,
+      "step": 1110
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00019815979430258982,
+      "loss": 1.0336,
+      "step": 1111
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0001981564603085388,
+      "loss": 1.0163,
+      "step": 1112
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00019815312332513533,
+      "loss": 0.9566,
+      "step": 1113
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0001981497833524811,
+      "loss": 0.9268,
+      "step": 1114
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00019814644039067777,
+      "loss": 0.8686,
+      "step": 1115
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0001981430944398272,
+      "loss": 0.9397,
+      "step": 1116
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00019813974550003126,
+      "loss": 0.95,
+      "step": 1117
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00019813639357139196,
+      "loss": 1.0146,
+      "step": 1118
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00019813303865401136,
+      "loss": 1.0269,
+      "step": 1119
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00019812968074799163,
+      "loss": 0.9968,
+      "step": 1120
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00019812631985343505,
+      "loss": 1.087,
+      "step": 1121
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00019812295597044396,
+      "loss": 1.0831,
+      "step": 1122
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00019811958909912085,
+      "loss": 1.0811,
+      "step": 1123
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00019811621923956823,
+      "loss": 1.019,
+      "step": 1124
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0001981128463918887,
+      "loss": 0.9932,
+      "step": 1125
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00019810947055618502,
+      "loss": 1.0653,
+      "step": 1126
+    },
+    {
+      "epoch": 0.25,
+      "eval_loss": 0.6458216905593872,
+      "eval_runtime": 322.1326,
+      "eval_samples_per_second": 56.309,
+      "eval_steps_per_second": 28.156,
+      "step": 1126
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00019810609173256,
+      "loss": 1.013,
+      "step": 1127
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00019810270992111652,
+      "loss": 1.0404,
+      "step": 1128
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00019809932512195762,
+      "loss": 1.1174,
+      "step": 1129
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0001980959373351863,
+      "loss": 1.0271,
+      "step": 1130
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0001980925465609058,
+      "loss": 0.9896,
+      "step": 1131
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00019808915279921938,
+      "loss": 0.9364,
+      "step": 1132
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00019808575605023037,
+      "loss": 0.9742,
+      "step": 1133
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00019808235631404224,
+      "loss": 0.9988,
+      "step": 1134
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00019807895359075852,
+      "loss": 0.8774,
+      "step": 1135
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00019807554788048285,
+      "loss": 1.009,
+      "step": 1136
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00019807213918331893,
+      "loss": 1.0295,
+      "step": 1137
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0001980687274993706,
+      "loss": 1.0025,
+      "step": 1138
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00019806531282874173,
+      "loss": 0.9094,
+      "step": 1139
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00019806189517153634,
+      "loss": 0.9804,
+      "step": 1140
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0001980584745278585,
+      "loss": 1.0385,
+      "step": 1141
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00019805505089781243,
+      "loss": 1.0444,
+      "step": 1142
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0001980516242815023,
+      "loss": 1.0167,
+      "step": 1143
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0001980481946790326,
+      "loss": 0.9822,
+      "step": 1144
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00019804476209050765,
+      "loss": 0.9827,
+      "step": 1145
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00019804132651603205,
+      "loss": 1.1233,
+      "step": 1146
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0001980378879557104,
+      "loss": 0.9585,
+      "step": 1147
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0001980344464096475,
+      "loss": 1.0499,
+      "step": 1148
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0001980310018779481,
+      "loss": 0.957,
+      "step": 1149
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00019802755436071706,
+      "loss": 0.9713,
+      "step": 1150
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00019802410385805948,
+      "loss": 1.0306,
+      "step": 1151
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0001980206503700804,
+      "loss": 0.852,
+      "step": 1152
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00019801719389688492,
+      "loss": 1.0416,
+      "step": 1153
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0001980137344385784,
+      "loss": 1.0022,
+      "step": 1154
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00019801027199526616,
+      "loss": 1.0526,
+      "step": 1155
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00019800680656705368,
+      "loss": 0.9306,
+      "step": 1156
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0001980033381540465,
+      "loss": 1.0885,
+      "step": 1157
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00019799986675635018,
+      "loss": 1.0264,
+      "step": 1158
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0001979963923740705,
+      "loss": 1.0387,
+      "step": 1159
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0001979929150073133,
+      "loss": 0.9465,
+      "step": 1160
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0001979894346561844,
+      "loss": 1.0147,
+      "step": 1161
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00019798595132078988,
+      "loss": 1.0763,
+      "step": 1162
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00019798246500123576,
+      "loss": 0.9063,
+      "step": 1163
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00019797897569762824,
+      "loss": 1.0094,
+      "step": 1164
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00019797548341007358,
+      "loss": 1.104,
+      "step": 1165
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00019797198813867816,
+      "loss": 1.1521,
+      "step": 1166
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0001979684898835484,
+      "loss": 0.9671,
+      "step": 1167
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00019796498864479086,
+      "loss": 0.9293,
+      "step": 1168
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00019796148442251217,
+      "loss": 1.079,
+      "step": 1169
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00019795797721681904,
+      "loss": 0.8382,
+      "step": 1170
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00019795446702781827,
+      "loss": 0.9261,
+      "step": 1171
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0001979509538556168,
+      "loss": 1.1614,
+      "step": 1172
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0001979474377003216,
+      "loss": 1.0681,
+      "step": 1173
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00019794391856203973,
+      "loss": 0.8877,
+      "step": 1174
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0001979403964408784,
+      "loss": 0.9985,
+      "step": 1175
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00019793687133694488,
+      "loss": 0.9788,
+      "step": 1176
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00019793334325034648,
+      "loss": 1.0495,
+      "step": 1177
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00019792981218119075,
+      "loss": 1.0016,
+      "step": 1178
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0001979262781295851,
+      "loss": 1.0966,
+      "step": 1179
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00019792274109563723,
+      "loss": 1.0493,
+      "step": 1180
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00019791920107945486,
+      "loss": 1.0164,
+      "step": 1181
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0001979156580811458,
+      "loss": 1.0747,
+      "step": 1182
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00019791211210081793,
+      "loss": 1.0193,
+      "step": 1183
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00019790856313857926,
+      "loss": 1.0397,
+      "step": 1184
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00019790501119453785,
+      "loss": 0.9759,
+      "step": 1185
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00019790145626880193,
+      "loss": 0.9062,
+      "step": 1186
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00019789789836147968,
+      "loss": 0.9514,
+      "step": 1187
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00019789433747267955,
+      "loss": 1.0537,
+      "step": 1188
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00019789077360250992,
+      "loss": 1.022,
+      "step": 1189
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00019788720675107934,
+      "loss": 1.0396,
+      "step": 1190
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00019788363691849648,
+      "loss": 1.1966,
+      "step": 1191
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00019788006410487,
+      "loss": 1.0839,
+      "step": 1192
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00019787648831030876,
+      "loss": 1.0297,
+      "step": 1193
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0001978729095349216,
+      "loss": 0.8893,
+      "step": 1194
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00019786932777881758,
+      "loss": 1.0607,
+      "step": 1195
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00019786574304210577,
+      "loss": 1.0131,
+      "step": 1196
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0001978621553248953,
+      "loss": 1.0101,
+      "step": 1197
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00019785856462729544,
+      "loss": 0.9938,
+      "step": 1198
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0001978549709494156,
+      "loss": 1.1419,
+      "step": 1199
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00019785137429136515,
+      "loss": 1.072,
+      "step": 1200
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00019784777465325367,
+      "loss": 1.1039,
+      "step": 1201
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0001978441720351908,
+      "loss": 0.9817,
+      "step": 1202
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00019784056643728623,
+      "loss": 1.0424,
+      "step": 1203
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0001978369578596498,
+      "loss": 0.9705,
+      "step": 1204
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00019783334630239137,
+      "loss": 0.9256,
+      "step": 1205
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00019782973176562092,
+      "loss": 1.0573,
+      "step": 1206
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0001978261142494486,
+      "loss": 1.0502,
+      "step": 1207
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0001978224937539845,
+      "loss": 0.9933,
+      "step": 1208
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00019781887027933893,
+      "loss": 0.9134,
+      "step": 1209
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00019781524382562224,
+      "loss": 1.018,
+      "step": 1210
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00019781161439294488,
+      "loss": 1.0565,
+      "step": 1211
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00019780798198141734,
+      "loss": 1.0935,
+      "step": 1212
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00019780434659115028,
+      "loss": 1.017,
+      "step": 1213
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00019780070822225443,
+      "loss": 0.9602,
+      "step": 1214
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0001977970668748406,
+      "loss": 1.0691,
+      "step": 1215
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00019779342254901963,
+      "loss": 1.0961,
+      "step": 1216
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00019778977524490253,
+      "loss": 1.0147,
+      "step": 1217
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00019778612496260041,
+      "loss": 0.8275,
+      "step": 1218
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00019778247170222442,
+      "loss": 1.0408,
+      "step": 1219
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00019777881546388583,
+      "loss": 1.0909,
+      "step": 1220
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00019777515624769598,
+      "loss": 0.9587,
+      "step": 1221
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0001977714940537663,
+      "loss": 0.9427,
+      "step": 1222
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00019776782888220835,
+      "loss": 1.0976,
+      "step": 1223
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00019776416073313374,
+      "loss": 1.0059,
+      "step": 1224
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00019776048960665416,
+      "loss": 1.0359,
+      "step": 1225
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00019775681550288145,
+      "loss": 0.9333,
+      "step": 1226
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0001977531384219275,
+      "loss": 0.9798,
+      "step": 1227
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00019774945836390425,
+      "loss": 0.9189,
+      "step": 1228
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00019774577532892385,
+      "loss": 1.0546,
+      "step": 1229
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0001977420893170984,
+      "loss": 0.9582,
+      "step": 1230
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00019773840032854018,
+      "loss": 1.0674,
+      "step": 1231
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00019773470836336157,
+      "loss": 1.0145,
+      "step": 1232
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00019773101342167497,
+      "loss": 0.9535,
+      "step": 1233
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00019772731550359292,
+      "loss": 0.9915,
+      "step": 1234
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00019772361460922803,
+      "loss": 1.1071,
+      "step": 1235
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00019771991073869305,
+      "loss": 1.1025,
+      "step": 1236
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00019771620389210073,
+      "loss": 1.0098,
+      "step": 1237
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.000197712494069564,
+      "loss": 0.9359,
+      "step": 1238
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00019770878127119582,
+      "loss": 1.0237,
+      "step": 1239
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0001977050654971093,
+      "loss": 0.9782,
+      "step": 1240
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0001977013467474175,
+      "loss": 0.9479,
+      "step": 1241
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0001976976250222338,
+      "loss": 1.0606,
+      "step": 1242
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0001976939003216715,
+      "loss": 0.9307,
+      "step": 1243
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00019769017264584403,
+      "loss": 1.0532,
+      "step": 1244
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00019768644199486487,
+      "loss": 1.1023,
+      "step": 1245
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00019768270836884773,
+      "loss": 0.9538,
+      "step": 1246
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00019767897176790626,
+      "loss": 1.112,
+      "step": 1247
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00019767523219215428,
+      "loss": 0.973,
+      "step": 1248
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00019767148964170567,
+      "loss": 1.0029,
+      "step": 1249
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00019766774411667437,
+      "loss": 0.9359,
+      "step": 1250
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0001976639956171745,
+      "loss": 1.0382,
+      "step": 1251
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00019766024414332023,
+      "loss": 1.0837,
+      "step": 1252
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00019765648969522577,
+      "loss": 0.9508,
+      "step": 1253
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00019765273227300548,
+      "loss": 0.9397,
+      "step": 1254
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00019764897187677378,
+      "loss": 0.9292,
+      "step": 1255
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00019764520850664523,
+      "loss": 0.9807,
+      "step": 1256
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00019764144216273438,
+      "loss": 0.9729,
+      "step": 1257
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00019763767284515602,
+      "loss": 1.0533,
+      "step": 1258
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00019763390055402486,
+      "loss": 1.0428,
+      "step": 1259
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00019763012528945586,
+      "loss": 0.96,
+      "step": 1260
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00019762634705156393,
+      "loss": 1.1171,
+      "step": 1261
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00019762256584046414,
+      "loss": 1.0264,
+      "step": 1262
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0001976187816562717,
+      "loss": 1.1071,
+      "step": 1263
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00019761499449910184,
+      "loss": 0.9712,
+      "step": 1264
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00019761120436906986,
+      "loss": 1.0293,
+      "step": 1265
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0001976074112662912,
+      "loss": 1.0252,
+      "step": 1266
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00019760361519088142,
+      "loss": 1.1033,
+      "step": 1267
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0001975998161429561,
+      "loss": 1.06,
+      "step": 1268
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0001975960141226309,
+      "loss": 1.0415,
+      "step": 1269
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0001975922091300217,
+      "loss": 0.9914,
+      "step": 1270
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00019758840116524434,
+      "loss": 1.0785,
+      "step": 1271
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00019758459022841475,
+      "loss": 0.9217,
+      "step": 1272
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00019758077631964902,
+      "loss": 0.959,
+      "step": 1273
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00019757695943906333,
+      "loss": 1.1044,
+      "step": 1274
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0001975731395867739,
+      "loss": 1.0005,
+      "step": 1275
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00019756931676289706,
+      "loss": 0.876,
+      "step": 1276
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00019756549096754925,
+      "loss": 1.0194,
+      "step": 1277
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00019756166220084696,
+      "loss": 1.1026,
+      "step": 1278
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00019755783046290678,
+      "loss": 0.8739,
+      "step": 1279
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00019755399575384548,
+      "loss": 1.1062,
+      "step": 1280
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0001975501580737798,
+      "loss": 0.9815,
+      "step": 1281
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0001975463174228266,
+      "loss": 1.0538,
+      "step": 1282
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00019754247380110286,
+      "loss": 0.9825,
+      "step": 1283
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00019753862720872564,
+      "loss": 1.0056,
+      "step": 1284
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0001975347776458121,
+      "loss": 1.0577,
+      "step": 1285
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00019753092511247945,
+      "loss": 0.9718,
+      "step": 1286
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00019752706960884505,
+      "loss": 1.1534,
+      "step": 1287
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0001975232111350263,
+      "loss": 1.0929,
+      "step": 1288
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0001975193496911407,
+      "loss": 0.9393,
+      "step": 1289
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00019751548527730588,
+      "loss": 0.9626,
+      "step": 1290
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00019751161789363953,
+      "loss": 0.9809,
+      "step": 1291
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0001975077475402594,
+      "loss": 1.0173,
+      "step": 1292
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00019750387421728339,
+      "loss": 0.9608,
+      "step": 1293
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00019749999792482943,
+      "loss": 0.9963,
+      "step": 1294
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00019749611866301558,
+      "loss": 1.0235,
+      "step": 1295
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00019749223643196001,
+      "loss": 0.8684,
+      "step": 1296
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00019748835123178097,
+      "loss": 1.0213,
+      "step": 1297
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00019748446306259673,
+      "loss": 0.9439,
+      "step": 1298
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0001974805719245257,
+      "loss": 0.9094,
+      "step": 1299
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00019747667781768645,
+      "loss": 1.1409,
+      "step": 1300
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00019747278074219753,
+      "loss": 0.9063,
+      "step": 1301
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00019746888069817762,
+      "loss": 0.9754,
+      "step": 1302
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00019746497768574546,
+      "loss": 1.0705,
+      "step": 1303
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00019746107170502003,
+      "loss": 1.0137,
+      "step": 1304
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00019745716275612022,
+      "loss": 1.14,
+      "step": 1305
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00019745325083916503,
+      "loss": 1.0804,
+      "step": 1306
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0001974493359542737,
+      "loss": 0.9778,
+      "step": 1307
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00019744541810156534,
+      "loss": 1.0215,
+      "step": 1308
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00019744149728115936,
+      "loss": 0.9585,
+      "step": 1309
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00019743757349317514,
+      "loss": 1.0852,
+      "step": 1310
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00019743364673773217,
+      "loss": 0.99,
+      "step": 1311
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00019742971701495004,
+      "loss": 0.9094,
+      "step": 1312
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00019742578432494845,
+      "loss": 1.1573,
+      "step": 1313
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00019742184866784713,
+      "loss": 0.9559,
+      "step": 1314
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.000197417910043766,
+      "loss": 0.8975,
+      "step": 1315
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00019741396845282498,
+      "loss": 0.932,
+      "step": 1316
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00019741002389514408,
+      "loss": 0.9483,
+      "step": 1317
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00019740607637084345,
+      "loss": 1.165,
+      "step": 1318
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00019740212588004337,
+      "loss": 1.0803,
+      "step": 1319
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00019739817242286404,
+      "loss": 0.8995,
+      "step": 1320
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.000197394215999426,
+      "loss": 0.9994,
+      "step": 1321
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0001973902566098496,
+      "loss": 1.0449,
+      "step": 1322
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00019738629425425553,
+      "loss": 0.9136,
+      "step": 1323
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0001973823289327644,
+      "loss": 0.885,
+      "step": 1324
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.000197378360645497,
+      "loss": 1.1525,
+      "step": 1325
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00019737438939257417,
+      "loss": 0.9678,
+      "step": 1326
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00019737041517411692,
+      "loss": 1.0017,
+      "step": 1327
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00019736643799024618,
+      "loss": 1.0206,
+      "step": 1328
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00019736245784108313,
+      "loss": 1.124,
+      "step": 1329
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00019735847472674903,
+      "loss": 0.9775,
+      "step": 1330
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0001973544886473651,
+      "loss": 1.0115,
+      "step": 1331
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00019735049960305278,
+      "loss": 1.0099,
+      "step": 1332
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00019734650759393354,
+      "loss": 1.135,
+      "step": 1333
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00019734251262012897,
+      "loss": 0.9436,
+      "step": 1334
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00019733851468176075,
+      "loss": 0.8519,
+      "step": 1335
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0001973345137789506,
+      "loss": 0.9857,
+      "step": 1336
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00019733050991182038,
+      "loss": 0.9064,
+      "step": 1337
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00019732650308049206,
+      "loss": 0.9914,
+      "step": 1338
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00019732249328508763,
+      "loss": 1.0434,
+      "step": 1339
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00019731848052572924,
+      "loss": 0.9497,
+      "step": 1340
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00019731446480253907,
+      "loss": 0.9168,
+      "step": 1341
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00019731044611563943,
+      "loss": 0.9966,
+      "step": 1342
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00019730642446515268,
+      "loss": 0.8687,
+      "step": 1343
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00019730239985120136,
+      "loss": 0.944,
+      "step": 1344
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00019729837227390798,
+      "loss": 1.0491,
+      "step": 1345
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00019729434173339526,
+      "loss": 1.1477,
+      "step": 1346
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00019729030822978588,
+      "loss": 1.0057,
+      "step": 1347
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00019728627176320273,
+      "loss": 1.0034,
+      "step": 1348
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00019728223233376875,
+      "loss": 0.8945,
+      "step": 1349
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0001972781899416069,
+      "loss": 1.0753,
+      "step": 1350
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00019727414458684038,
+      "loss": 1.1841,
+      "step": 1351
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00019727009626959227,
+      "loss": 0.8942,
+      "step": 1352
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00019726604498998597,
+      "loss": 0.9882,
+      "step": 1353
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00019726199074814483,
+      "loss": 1.024,
+      "step": 1354
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0001972579335441923,
+      "loss": 1.0994,
+      "step": 1355
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00019725387337825195,
+      "loss": 0.8449,
+      "step": 1356
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00019724981025044747,
+      "loss": 0.8734,
+      "step": 1357
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00019724574416090252,
+      "loss": 1.0882,
+      "step": 1358
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00019724167510974105,
+      "loss": 1.0129,
+      "step": 1359
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00019723760309708687,
+      "loss": 1.0387,
+      "step": 1360
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00019723352812306405,
+      "loss": 0.8313,
+      "step": 1361
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0001972294501877967,
+      "loss": 0.9917,
+      "step": 1362
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00019722536929140896,
+      "loss": 1.0526,
+      "step": 1363
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00019722128543402519,
+      "loss": 1.0659,
+      "step": 1364
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00019721719861576968,
+      "loss": 1.0207,
+      "step": 1365
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00019721310883676697,
+      "loss": 1.0478,
+      "step": 1366
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0001972090160971416,
+      "loss": 1.0374,
+      "step": 1367
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00019720492039701815,
+      "loss": 1.0496,
+      "step": 1368
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0001972008217365214,
+      "loss": 0.9543,
+      "step": 1369
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00019719672011577624,
+      "loss": 1.0109,
+      "step": 1370
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00019719261553490747,
+      "loss": 0.8756,
+      "step": 1371
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00019718850799404014,
+      "loss": 0.952,
+      "step": 1372
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00019718439749329939,
+      "loss": 0.9912,
+      "step": 1373
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0001971802840328103,
+      "loss": 1.0178,
+      "step": 1374
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00019717616761269827,
+      "loss": 0.9533,
+      "step": 1375
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00019717204823308857,
+      "loss": 0.9176,
+      "step": 1376
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0001971679258941067,
+      "loss": 0.9217,
+      "step": 1377
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0001971638005958782,
+      "loss": 1.1175,
+      "step": 1378
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0001971596723385287,
+      "loss": 0.8889,
+      "step": 1379
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00019715554112218395,
+      "loss": 1.0608,
+      "step": 1380
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00019715140694696973,
+      "loss": 1.0611,
+      "step": 1381
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00019714726981301194,
+      "loss": 1.0285,
+      "step": 1382
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00019714312972043664,
+      "loss": 0.9093,
+      "step": 1383
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00019713898666936985,
+      "loss": 0.999,
+      "step": 1384
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00019713484065993778,
+      "loss": 0.9887,
+      "step": 1385
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00019713069169226665,
+      "loss": 1.0372,
+      "step": 1386
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0001971265397664829,
+      "loss": 0.9563,
+      "step": 1387
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00019712238488271293,
+      "loss": 1.0251,
+      "step": 1388
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00019711822704108326,
+      "loss": 1.0673,
+      "step": 1389
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00019711406624172052,
+      "loss": 1.0352,
+      "step": 1390
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00019710990248475145,
+      "loss": 0.9455,
+      "step": 1391
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00019710573577030286,
+      "loss": 1.0673,
+      "step": 1392
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00019710156609850163,
+      "loss": 1.0922,
+      "step": 1393
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00019709739346947478,
+      "loss": 1.0051,
+      "step": 1394
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00019709321788334932,
+      "loss": 1.0885,
+      "step": 1395
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00019708903934025247,
+      "loss": 0.9765,
+      "step": 1396
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00019708485784031147,
+      "loss": 1.0529,
+      "step": 1397
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0001970806733836537,
+      "loss": 1.0696,
+      "step": 1398
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00019707648597040654,
+      "loss": 0.9379,
+      "step": 1399
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00019707229560069758,
+      "loss": 1.023,
+      "step": 1400
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00019706810227465438,
+      "loss": 0.9125,
+      "step": 1401
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00019706390599240464,
+      "loss": 1.0675,
+      "step": 1402
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00019705970675407627,
+      "loss": 0.967,
+      "step": 1403
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.000197055504559797,
+      "loss": 0.9768,
+      "step": 1404
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00019705129940969493,
+      "loss": 1.0636,
+      "step": 1405
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00019704709130389806,
+      "loss": 1.1519,
+      "step": 1406
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0001970428802425346,
+      "loss": 1.1784,
+      "step": 1407
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00019703866622573278,
+      "loss": 1.0678,
+      "step": 1408
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0001970344492536209,
+      "loss": 0.8472,
+      "step": 1409
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0001970302293263274,
+      "loss": 1.0692,
+      "step": 1410
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00019702600644398084,
+      "loss": 0.9448,
+      "step": 1411
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0001970217806067098,
+      "loss": 1.0172,
+      "step": 1412
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00019701755181464296,
+      "loss": 0.974,
+      "step": 1413
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00019701332006790914,
+      "loss": 0.9249,
+      "step": 1414
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00019700908536663718,
+      "loss": 1.1123,
+      "step": 1415
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00019700484771095612,
+      "loss": 0.9914,
+      "step": 1416
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00019700060710099493,
+      "loss": 0.9629,
+      "step": 1417
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0001969963635368828,
+      "loss": 1.0224,
+      "step": 1418
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00019699211701874897,
+      "loss": 0.9559,
+      "step": 1419
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00019698786754672275,
+      "loss": 1.0419,
+      "step": 1420
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0001969836151209336,
+      "loss": 0.9267,
+      "step": 1421
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00019697935974151095,
+      "loss": 0.9454,
+      "step": 1422
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0001969751014085845,
+      "loss": 0.8974,
+      "step": 1423
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00019697084012228384,
+      "loss": 0.9413,
+      "step": 1424
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00019696657588273882,
+      "loss": 0.8866,
+      "step": 1425
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00019696230869007927,
+      "loss": 0.963,
+      "step": 1426
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00019695803854443512,
+      "loss": 0.8956,
+      "step": 1427
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0001969537654459365,
+      "loss": 1.1471,
+      "step": 1428
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00019694948939471345,
+      "loss": 0.9075,
+      "step": 1429
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0001969452103908963,
+      "loss": 0.9452,
+      "step": 1430
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00019694092843461526,
+      "loss": 1.0637,
+      "step": 1431
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0001969366435260008,
+      "loss": 1.0232,
+      "step": 1432
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00019693235566518343,
+      "loss": 0.9828,
+      "step": 1433
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00019692806485229367,
+      "loss": 0.9481,
+      "step": 1434
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0001969237710874623,
+      "loss": 1.0264,
+      "step": 1435
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00019691947437081996,
+      "loss": 1.1061,
+      "step": 1436
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00019691517470249762,
+      "loss": 1.15,
+      "step": 1437
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00019691087208262615,
+      "loss": 0.9823,
+      "step": 1438
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00019690656651133662,
+      "loss": 1.004,
+      "step": 1439
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00019690225798876018,
+      "loss": 0.9444,
+      "step": 1440
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00019689794651502797,
+      "loss": 1.0062,
+      "step": 1441
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00019689363209027136,
+      "loss": 1.0116,
+      "step": 1442
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00019688931471462175,
+      "loss": 0.9236,
+      "step": 1443
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0001968849943882106,
+      "loss": 0.9827,
+      "step": 1444
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00019688067111116946,
+      "loss": 1.0205,
+      "step": 1445
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00019687634488363003,
+      "loss": 0.9782,
+      "step": 1446
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00019687201570572408,
+      "loss": 0.8711,
+      "step": 1447
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00019686768357758344,
+      "loss": 0.9695,
+      "step": 1448
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00019686334849933998,
+      "loss": 0.9145,
+      "step": 1449
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00019685901047112586,
+      "loss": 0.9815,
+      "step": 1450
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00019685466949307306,
+      "loss": 1.1415,
+      "step": 1451
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0001968503255653139,
+      "loss": 0.9481,
+      "step": 1452
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00019684597868798055,
+      "loss": 0.9759,
+      "step": 1453
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0001968416288612055,
+      "loss": 0.9475,
+      "step": 1454
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00019683727608512118,
+      "loss": 0.9635,
+      "step": 1455
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00019683292035986016,
+      "loss": 1.0146,
+      "step": 1456
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0001968285616855551,
+      "loss": 1.0728,
+      "step": 1457
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00019682420006233872,
+      "loss": 0.8355,
+      "step": 1458
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00019681983549034386,
+      "loss": 0.956,
+      "step": 1459
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00019681546796970347,
+      "loss": 0.9992,
+      "step": 1460
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00019681109750055054,
+      "loss": 0.942,
+      "step": 1461
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00019680672408301818,
+      "loss": 1.048,
+      "step": 1462
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00019680234771723957,
+      "loss": 0.9354,
+      "step": 1463
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00019679796840334798,
+      "loss": 0.9411,
+      "step": 1464
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00019679358614147684,
+      "loss": 0.927,
+      "step": 1465
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00019678920093175957,
+      "loss": 1.0458,
+      "step": 1466
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00019678481277432972,
+      "loss": 0.9623,
+      "step": 1467
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00019678042166932093,
+      "loss": 0.9233,
+      "step": 1468
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00019677602761686695,
+      "loss": 1.0088,
+      "step": 1469
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00019677163061710158,
+      "loss": 1.136,
+      "step": 1470
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00019676723067015874,
+      "loss": 1.0234,
+      "step": 1471
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00019676282777617243,
+      "loss": 0.9899,
+      "step": 1472
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00019675842193527676,
+      "loss": 0.9173,
+      "step": 1473
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00019675401314760585,
+      "loss": 1.0632,
+      "step": 1474
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00019674960141329403,
+      "loss": 0.8498,
+      "step": 1475
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00019674518673247566,
+      "loss": 0.8948,
+      "step": 1476
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00019674076910528512,
+      "loss": 1.1067,
+      "step": 1477
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00019673634853185703,
+      "loss": 0.9819,
+      "step": 1478
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00019673192501232597,
+      "loss": 1.0361,
+      "step": 1479
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00019672749854682668,
+      "loss": 1.0425,
+      "step": 1480
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00019672306913549395,
+      "loss": 1.0728,
+      "step": 1481
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00019671863677846272,
+      "loss": 1.0487,
+      "step": 1482
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00019671420147586793,
+      "loss": 1.0726,
+      "step": 1483
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00019670976322784466,
+      "loss": 0.9753,
+      "step": 1484
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00019670532203452807,
+      "loss": 1.1,
+      "step": 1485
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00019670087789605347,
+      "loss": 0.985,
+      "step": 1486
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00019669643081255618,
+      "loss": 0.9659,
+      "step": 1487
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0001966919807841716,
+      "loss": 0.9233,
+      "step": 1488
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0001966875278110353,
+      "loss": 1.0733,
+      "step": 1489
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00019668307189328287,
+      "loss": 1.0862,
+      "step": 1490
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00019667861303105,
+      "loss": 0.9761,
+      "step": 1491
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00019667415122447254,
+      "loss": 0.9947,
+      "step": 1492
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00019666968647368634,
+      "loss": 1.0064,
+      "step": 1493
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00019666521877882739,
+      "loss": 0.9614,
+      "step": 1494
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0001966607481400317,
+      "loss": 0.9493,
+      "step": 1495
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00019665627455743548,
+      "loss": 0.9113,
+      "step": 1496
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00019665179803117497,
+      "loss": 0.9826,
+      "step": 1497
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00019664731856138646,
+      "loss": 1.026,
+      "step": 1498
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0001966428361482064,
+      "loss": 1.1172,
+      "step": 1499
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00019663835079177132,
+      "loss": 1.1311,
+      "step": 1500
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0001966338624922178,
+      "loss": 0.9591,
+      "step": 1501
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00019662937124968256,
+      "loss": 0.9262,
+      "step": 1502
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0001966248770643023,
+      "loss": 0.9298,
+      "step": 1503
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00019662037993621397,
+      "loss": 1.0763,
+      "step": 1504
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00019661587986555452,
+      "loss": 0.9608,
+      "step": 1505
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00019661137685246099,
+      "loss": 0.9408,
+      "step": 1506
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0001966068708970705,
+      "loss": 1.0412,
+      "step": 1507
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00019660236199952027,
+      "loss": 0.9908,
+      "step": 1508
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00019659785015994764,
+      "loss": 1.0444,
+      "step": 1509
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00019659333537849004,
+      "loss": 0.9983,
+      "step": 1510
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00019658881765528493,
+      "loss": 1.1155,
+      "step": 1511
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00019658429699046995,
+      "loss": 0.9031,
+      "step": 1512
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0001965797733841827,
+      "loss": 0.992,
+      "step": 1513
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.000196575246836561,
+      "loss": 0.9759,
+      "step": 1514
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00019657071734774273,
+      "loss": 1.0204,
+      "step": 1515
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00019656618491786576,
+      "loss": 1.0843,
+      "step": 1516
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00019656164954706817,
+      "loss": 0.9311,
+      "step": 1517
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0001965571112354881,
+      "loss": 0.9147,
+      "step": 1518
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00019655256998326367,
+      "loss": 1.0539,
+      "step": 1519
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0001965480257905333,
+      "loss": 1.0225,
+      "step": 1520
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00019654347865743533,
+      "loss": 1.0792,
+      "step": 1521
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0001965389285841083,
+      "loss": 1.0285,
+      "step": 1522
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00019653437557069068,
+      "loss": 1.0676,
+      "step": 1523
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0001965298196173212,
+      "loss": 1.0892,
+      "step": 1524
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0001965252607241386,
+      "loss": 1.0275,
+      "step": 1525
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0001965206988912817,
+      "loss": 0.9334,
+      "step": 1526
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00019651613411888948,
+      "loss": 1.0616,
+      "step": 1527
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0001965115664071009,
+      "loss": 0.9615,
+      "step": 1528
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0001965069957560551,
+      "loss": 1.0959,
+      "step": 1529
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0001965024221658913,
+      "loss": 1.0427,
+      "step": 1530
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00019649784563674876,
+      "loss": 0.9417,
+      "step": 1531
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00019649326616876685,
+      "loss": 0.9956,
+      "step": 1532
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00019648868376208507,
+      "loss": 0.9876,
+      "step": 1533
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00019648409841684295,
+      "loss": 0.9277,
+      "step": 1534
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00019647951013318014,
+      "loss": 1.0295,
+      "step": 1535
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00019647491891123639,
+      "loss": 1.0309,
+      "step": 1536
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0001964703247511515,
+      "loss": 1.0077,
+      "step": 1537
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00019646572765306543,
+      "loss": 0.9986,
+      "step": 1538
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00019646112761711811,
+      "loss": 1.0166,
+      "step": 1539
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00019645652464344971,
+      "loss": 0.9826,
+      "step": 1540
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0001964519187322004,
+      "loss": 0.9504,
+      "step": 1541
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0001964473098835104,
+      "loss": 0.9352,
+      "step": 1542
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00019644269809752015,
+      "loss": 1.091,
+      "step": 1543
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00019643808337437005,
+      "loss": 0.9047,
+      "step": 1544
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00019643346571420064,
+      "loss": 1.0838,
+      "step": 1545
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00019642884511715254,
+      "loss": 1.0321,
+      "step": 1546
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00019642422158336654,
+      "loss": 0.9503,
+      "step": 1547
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0001964195951129834,
+      "loss": 0.9154,
+      "step": 1548
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00019641496570614396,
+      "loss": 1.0417,
+      "step": 1549
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00019641033336298934,
+      "loss": 1.0016,
+      "step": 1550
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0001964056980836605,
+      "loss": 1.1161,
+      "step": 1551
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00019640105986829868,
+      "loss": 0.9547,
+      "step": 1552
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0001963964187170451,
+      "loss": 0.9992,
+      "step": 1553
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00019639177463004113,
+      "loss": 0.9966,
+      "step": 1554
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0001963871276074282,
+      "loss": 0.9506,
+      "step": 1555
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00019638247764934777,
+      "loss": 0.9461,
+      "step": 1556
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00019637782475594157,
+      "loss": 1.0335,
+      "step": 1557
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00019637316892735126,
+      "loss": 1.0184,
+      "step": 1558
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00019636851016371858,
+      "loss": 0.8883,
+      "step": 1559
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00019636384846518547,
+      "loss": 0.9465,
+      "step": 1560
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0001963591838318939,
+      "loss": 1.0045,
+      "step": 1561
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0001963545162639859,
+      "loss": 1.1091,
+      "step": 1562
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00019634984576160361,
+      "loss": 0.9972,
+      "step": 1563
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00019634517232488933,
+      "loss": 0.9527,
+      "step": 1564
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00019634049595398533,
+      "loss": 0.9381,
+      "step": 1565
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00019633581664903408,
+      "loss": 1.163,
+      "step": 1566
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00019633113441017806,
+      "loss": 0.8677,
+      "step": 1567
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00019632644923755986,
+      "loss": 0.9472,
+      "step": 1568
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00019632176113132217,
+      "loss": 1.0003,
+      "step": 1569
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00019631707009160778,
+      "loss": 0.9905,
+      "step": 1570
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00019631237611855956,
+      "loss": 0.9636,
+      "step": 1571
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00019630767921232045,
+      "loss": 1.0619,
+      "step": 1572
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00019630297937303347,
+      "loss": 0.9904,
+      "step": 1573
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0001962982766008418,
+      "loss": 0.8911,
+      "step": 1574
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00019629357089588864,
+      "loss": 0.9798,
+      "step": 1575
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00019628886225831732,
+      "loss": 0.9763,
+      "step": 1576
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0001962841506882712,
+      "loss": 1.0315,
+      "step": 1577
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00019627943618589385,
+      "loss": 0.9111,
+      "step": 1578
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00019627471875132874,
+      "loss": 1.0051,
+      "step": 1579
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00019626999838471965,
+      "loss": 1.0039,
+      "step": 1580
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00019626527508621023,
+      "loss": 1.0194,
+      "step": 1581
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0001962605488559444,
+      "loss": 1.1598,
+      "step": 1582
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00019625581969406612,
+      "loss": 0.8454,
+      "step": 1583
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00019625108760071936,
+      "loss": 1.0626,
+      "step": 1584
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0001962463525760482,
+      "loss": 1.0466,
+      "step": 1585
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00019624161462019696,
+      "loss": 1.1413,
+      "step": 1586
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00019623687373330985,
+      "loss": 0.9904,
+      "step": 1587
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00019623212991553128,
+      "loss": 0.9809,
+      "step": 1588
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00019622738316700574,
+      "loss": 1.0385,
+      "step": 1589
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00019622263348787776,
+      "loss": 1.1696,
+      "step": 1590
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00019621788087829196,
+      "loss": 1.0724,
+      "step": 1591
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00019621312533839316,
+      "loss": 1.0216,
+      "step": 1592
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00019620836686832615,
+      "loss": 0.9404,
+      "step": 1593
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00019620360546823587,
+      "loss": 0.9555,
+      "step": 1594
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00019619884113826727,
+      "loss": 0.9024,
+      "step": 1595
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00019619407387856552,
+      "loss": 1.203,
+      "step": 1596
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00019618930368927578,
+      "loss": 0.8759,
+      "step": 1597
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0001961845305705433,
+      "loss": 1.063,
+      "step": 1598
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00019617975452251347,
+      "loss": 0.9876,
+      "step": 1599
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00019617497554533174,
+      "loss": 0.8945,
+      "step": 1600
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00019617019363914366,
+      "loss": 0.868,
+      "step": 1601
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00019616540880409484,
+      "loss": 0.9972,
+      "step": 1602
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00019616062104033105,
+      "loss": 1.0123,
+      "step": 1603
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00019615583034799804,
+      "loss": 0.9405,
+      "step": 1604
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00019615103672724176,
+      "loss": 0.9357,
+      "step": 1605
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00019614624017820815,
+      "loss": 1.0716,
+      "step": 1606
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00019614144070104333,
+      "loss": 0.9732,
+      "step": 1607
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00019613663829589348,
+      "loss": 1.002,
+      "step": 1608
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00019613183296290484,
+      "loss": 0.8887,
+      "step": 1609
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00019612702470222373,
+      "loss": 1.086,
+      "step": 1610
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00019612221351399662,
+      "loss": 0.9211,
+      "step": 1611
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00019611739939837,
+      "loss": 1.121,
+      "step": 1612
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0001961125823554905,
+      "loss": 1.1447,
+      "step": 1613
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00019610776238550485,
+      "loss": 0.9142,
+      "step": 1614
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0001961029394885598,
+      "loss": 0.8862,
+      "step": 1615
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00019609811366480225,
+      "loss": 0.9412,
+      "step": 1616
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0001960932849143792,
+      "loss": 1.0139,
+      "step": 1617
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00019608845323743765,
+      "loss": 0.9604,
+      "step": 1618
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00019608361863412478,
+      "loss": 0.9842,
+      "step": 1619
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00019607878110458783,
+      "loss": 0.9519,
+      "step": 1620
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00019607394064897416,
+      "loss": 0.9162,
+      "step": 1621
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0001960690972674311,
+      "loss": 1.101,
+      "step": 1622
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00019606425096010624,
+      "loss": 1.0219,
+      "step": 1623
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0001960594017271471,
+      "loss": 1.0355,
+      "step": 1624
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00019605454956870145,
+      "loss": 1.0514,
+      "step": 1625
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00019604969448491695,
+      "loss": 1.0456,
+      "step": 1626
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00019604483647594155,
+      "loss": 0.9415,
+      "step": 1627
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00019603997554192321,
+      "loss": 1.0404,
+      "step": 1628
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00019603511168300993,
+      "loss": 1.0897,
+      "step": 1629
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0001960302448993498,
+      "loss": 0.9752,
+      "step": 1630
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00019602537519109112,
+      "loss": 0.949,
+      "step": 1631
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00019602050255838215,
+      "loss": 1.013,
+      "step": 1632
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00019601562700137127,
+      "loss": 1.0755,
+      "step": 1633
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00019601074852020704,
+      "loss": 0.9366,
+      "step": 1634
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00019600586711503794,
+      "loss": 1.0683,
+      "step": 1635
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00019600098278601272,
+      "loss": 1.0545,
+      "step": 1636
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00019599609553328006,
+      "loss": 0.8883,
+      "step": 1637
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00019599120535698883,
+      "loss": 0.9779,
+      "step": 1638
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00019598631225728794,
+      "loss": 1.0288,
+      "step": 1639
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00019598141623432643,
+      "loss": 0.838,
+      "step": 1640
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00019597651728825346,
+      "loss": 1.0169,
+      "step": 1641
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00019597161541921813,
+      "loss": 1.1296,
+      "step": 1642
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0001959667106273698,
+      "loss": 1.0205,
+      "step": 1643
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00019596180291285779,
+      "loss": 1.0153,
+      "step": 1644
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001959568922758316,
+      "loss": 0.9956,
+      "step": 1645
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00019595197871644077,
+      "loss": 1.0087,
+      "step": 1646
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00019594706223483496,
+      "loss": 0.969,
+      "step": 1647
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001959421428311639,
+      "loss": 1.1736,
+      "step": 1648
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001959372205055774,
+      "loss": 1.0379,
+      "step": 1649
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00019593229525822532,
+      "loss": 0.8886,
+      "step": 1650
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00019592736708925774,
+      "loss": 0.9927,
+      "step": 1651
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001959224359988247,
+      "loss": 0.9186,
+      "step": 1652
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00019591750198707638,
+      "loss": 1.1012,
+      "step": 1653
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001959125650541631,
+      "loss": 0.9053,
+      "step": 1654
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001959076252002351,
+      "loss": 0.9142,
+      "step": 1655
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00019590268242544297,
+      "loss": 1.0965,
+      "step": 1656
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001958977367299371,
+      "loss": 1.0845,
+      "step": 1657
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00019589278811386822,
+      "loss": 1.1324,
+      "step": 1658
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00019588783657738696,
+      "loss": 1.0222,
+      "step": 1659
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00019588288212064417,
+      "loss": 1.0235,
+      "step": 1660
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00019587792474379072,
+      "loss": 0.9914,
+      "step": 1661
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001958729644469776,
+      "loss": 1.0821,
+      "step": 1662
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00019586800123035585,
+      "loss": 0.8003,
+      "step": 1663
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00019586303509407663,
+      "loss": 1.0425,
+      "step": 1664
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001958580660382912,
+      "loss": 1.0471,
+      "step": 1665
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001958530940631509,
+      "loss": 1.0131,
+      "step": 1666
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00019584811916880713,
+      "loss": 1.0109,
+      "step": 1667
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001958431413554114,
+      "loss": 1.1009,
+      "step": 1668
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001958381606231153,
+      "loss": 1.0177,
+      "step": 1669
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00019583317697207056,
+      "loss": 1.0699,
+      "step": 1670
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00019582819040242893,
+      "loss": 1.0272,
+      "step": 1671
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00019582320091434226,
+      "loss": 1.0114,
+      "step": 1672
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00019581820850796252,
+      "loss": 1.0215,
+      "step": 1673
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00019581321318344177,
+      "loss": 0.969,
+      "step": 1674
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001958082149409321,
+      "loss": 1.0669,
+      "step": 1675
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001958032137805858,
+      "loss": 0.9692,
+      "step": 1676
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001957982097025551,
+      "loss": 1.0718,
+      "step": 1677
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00019579320270699245,
+      "loss": 0.8677,
+      "step": 1678
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00019578819279405035,
+      "loss": 0.8024,
+      "step": 1679
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00019578317996388133,
+      "loss": 1.0551,
+      "step": 1680
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00019577816421663809,
+      "loss": 1.0642,
+      "step": 1681
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00019577314555247334,
+      "loss": 1.0526,
+      "step": 1682
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00019576812397154,
+      "loss": 0.9721,
+      "step": 1683
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00019576309947399095,
+      "loss": 0.963,
+      "step": 1684
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001957580720599792,
+      "loss": 1.1111,
+      "step": 1685
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001957530417296579,
+      "loss": 1.0044,
+      "step": 1686
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001957480084831802,
+      "loss": 1.0441,
+      "step": 1687
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00019574297232069943,
+      "loss": 0.9566,
+      "step": 1688
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00019573793324236896,
+      "loss": 0.9976,
+      "step": 1689
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00019573289124834222,
+      "loss": 0.8713,
+      "step": 1690
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00019572784633877278,
+      "loss": 0.8668,
+      "step": 1691
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00019572279851381435,
+      "loss": 0.9325,
+      "step": 1692
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00019571774777362054,
+      "loss": 0.9455,
+      "step": 1693
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00019571269411834528,
+      "loss": 0.9559,
+      "step": 1694
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00019570763754814244,
+      "loss": 0.9043,
+      "step": 1695
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00019570257806316596,
+      "loss": 0.9645,
+      "step": 1696
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00019569751566357002,
+      "loss": 1.0774,
+      "step": 1697
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00019569245034950874,
+      "loss": 0.9029,
+      "step": 1698
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00019568738212113638,
+      "loss": 1.048,
+      "step": 1699
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00019568231097860731,
+      "loss": 1.0559,
+      "step": 1700
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.000195677236922076,
+      "loss": 0.9535,
+      "step": 1701
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00019567215995169694,
+      "loss": 1.0015,
+      "step": 1702
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00019566708006762474,
+      "loss": 1.0181,
+      "step": 1703
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00019566199727001415,
+      "loss": 1.0748,
+      "step": 1704
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00019565691155901992,
+      "loss": 1.0213,
+      "step": 1705
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.000195651822934797,
+      "loss": 0.9629,
+      "step": 1706
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0001956467313975003,
+      "loss": 1.0426,
+      "step": 1707
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00019564163694728488,
+      "loss": 1.0839,
+      "step": 1708
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00019563653958430596,
+      "loss": 1.0829,
+      "step": 1709
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0001956314393087187,
+      "loss": 0.956,
+      "step": 1710
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00019562633612067848,
+      "loss": 0.9292,
+      "step": 1711
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0001956212300203407,
+      "loss": 1.0063,
+      "step": 1712
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0001956161210078609,
+      "loss": 1.0009,
+      "step": 1713
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0001956110090833946,
+      "loss": 1.0045,
+      "step": 1714
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00019560589424709758,
+      "loss": 1.0134,
+      "step": 1715
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00019560077649912548,
+      "loss": 1.1471,
+      "step": 1716
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0001955956558396343,
+      "loss": 0.9803,
+      "step": 1717
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00019559053226877993,
+      "loss": 0.9824,
+      "step": 1718
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0001955854057867184,
+      "loss": 0.8947,
+      "step": 1719
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00019558027639360583,
+      "loss": 1.0805,
+      "step": 1720
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00019557514408959849,
+      "loss": 1.0106,
+      "step": 1721
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0001955700088748526,
+      "loss": 1.0236,
+      "step": 1722
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0001955648707495246,
+      "loss": 0.9592,
+      "step": 1723
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.000195559729713771,
+      "loss": 0.9746,
+      "step": 1724
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00019555458576774832,
+      "loss": 1.0433,
+      "step": 1725
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00019554943891161327,
+      "loss": 1.0068,
+      "step": 1726
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00019554428914552254,
+      "loss": 0.8476,
+      "step": 1727
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00019553913646963298,
+      "loss": 0.9837,
+      "step": 1728
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00019553398088410155,
+      "loss": 0.9876,
+      "step": 1729
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0001955288223890852,
+      "loss": 1.0296,
+      "step": 1730
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0001955236609847411,
+      "loss": 0.968,
+      "step": 1731
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0001955184966712264,
+      "loss": 0.9475,
+      "step": 1732
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00019551332944869843,
+      "loss": 0.8768,
+      "step": 1733
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00019550815931731447,
+      "loss": 0.9493,
+      "step": 1734
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00019550298627723207,
+      "loss": 0.9682,
+      "step": 1735
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0001954978103286087,
+      "loss": 0.9827,
+      "step": 1736
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00019549263147160205,
+      "loss": 1.0409,
+      "step": 1737
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00019548744970636977,
+      "loss": 0.9936,
+      "step": 1738
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00019548226503306973,
+      "loss": 1.0536,
+      "step": 1739
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00019547707745185984,
+      "loss": 1.0944,
+      "step": 1740
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00019547188696289805,
+      "loss": 0.9744,
+      "step": 1741
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00019546669356634245,
+      "loss": 0.9646,
+      "step": 1742
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0001954614972623512,
+      "loss": 0.871,
+      "step": 1743
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00019545629805108253,
+      "loss": 1.0702,
+      "step": 1744
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00019545109593269484,
+      "loss": 1.1104,
+      "step": 1745
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00019544589090734652,
+      "loss": 0.9915,
+      "step": 1746
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0001954406829751961,
+      "loss": 1.0264,
+      "step": 1747
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00019543547213640217,
+      "loss": 1.0519,
+      "step": 1748
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00019543025839112348,
+      "loss": 1.0043,
+      "step": 1749
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00019542504173951874,
+      "loss": 0.8847,
+      "step": 1750
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00019541982218174685,
+      "loss": 0.9636,
+      "step": 1751
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0001954145997179668,
+      "loss": 1.091,
+      "step": 1752
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0001954093743483376,
+      "loss": 0.9883,
+      "step": 1753
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00019540414607301844,
+      "loss": 1.1087,
+      "step": 1754
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00019539891489216854,
+      "loss": 1.0549,
+      "step": 1755
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0001953936808059471,
+      "loss": 0.9007,
+      "step": 1756
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00019538844381451368,
+      "loss": 0.9976,
+      "step": 1757
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0001953832039180277,
+      "loss": 0.9706,
+      "step": 1758
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00019537796111664875,
+      "loss": 0.9058,
+      "step": 1759
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00019537271541053647,
+      "loss": 0.9094,
+      "step": 1760
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00019536746679985067,
+      "loss": 1.0567,
+      "step": 1761
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00019536221528475118,
+      "loss": 1.0348,
+      "step": 1762
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00019535696086539793,
+      "loss": 1.0565,
+      "step": 1763
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0001953517035419509,
+      "loss": 1.0954,
+      "step": 1764
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00019534644331457027,
+      "loss": 0.9448,
+      "step": 1765
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0001953411801834162,
+      "loss": 0.9476,
+      "step": 1766
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.000195335914148649,
+      "loss": 0.9789,
+      "step": 1767
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00019533064521042902,
+      "loss": 0.9989,
+      "step": 1768
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00019532537336891678,
+      "loss": 0.9726,
+      "step": 1769
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00019532009862427275,
+      "loss": 1.0339,
+      "step": 1770
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00019531482097665762,
+      "loss": 1.0733,
+      "step": 1771
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00019530954042623218,
+      "loss": 1.0103,
+      "step": 1772
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00019530425697315712,
+      "loss": 1.0745,
+      "step": 1773
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00019529897061759343,
+      "loss": 1.014,
+      "step": 1774
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00019529368135970211,
+      "loss": 0.9996,
+      "step": 1775
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00019528838919964422,
+      "loss": 0.9314,
+      "step": 1776
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00019528309413758096,
+      "loss": 1.0313,
+      "step": 1777
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00019527779617367353,
+      "loss": 0.9903,
+      "step": 1778
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00019527249530808334,
+      "loss": 0.9942,
+      "step": 1779
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0001952671915409718,
+      "loss": 1.0443,
+      "step": 1780
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00019526188487250043,
+      "loss": 1.0099,
+      "step": 1781
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00019525657530283086,
+      "loss": 0.9789,
+      "step": 1782
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00019525126283212482,
+      "loss": 1.0524,
+      "step": 1783
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00019524594746054403,
+      "loss": 1.0136,
+      "step": 1784
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00019524062918825044,
+      "loss": 0.8856,
+      "step": 1785
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00019523530801540594,
+      "loss": 0.9423,
+      "step": 1786
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0001952299839421727,
+      "loss": 0.9743,
+      "step": 1787
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00019522465696871275,
+      "loss": 0.9357,
+      "step": 1788
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0001952193270951884,
+      "loss": 0.8932,
+      "step": 1789
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00019521399432176193,
+      "loss": 1.1029,
+      "step": 1790
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00019520865864859578,
+      "loss": 0.9825,
+      "step": 1791
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00019520332007585237,
+      "loss": 0.9014,
+      "step": 1792
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00019519797860369443,
+      "loss": 1.0615,
+      "step": 1793
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00019519263423228449,
+      "loss": 0.9983,
+      "step": 1794
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00019518728696178538,
+      "loss": 0.9242,
+      "step": 1795
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00019518193679235997,
+      "loss": 1.0737,
+      "step": 1796
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00019517658372417116,
+      "loss": 1.0253,
+      "step": 1797
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00019517122775738197,
+      "loss": 1.0227,
+      "step": 1798
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00019516586889215556,
+      "loss": 0.9281,
+      "step": 1799
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00019516050712865512,
+      "loss": 0.8593,
+      "step": 1800
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0001951551424670439,
+      "loss": 1.1585,
+      "step": 1801
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00019514977490748536,
+      "loss": 1.0935,
+      "step": 1802
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00019514440445014288,
+      "loss": 0.9704,
+      "step": 1803
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00019513903109518008,
+      "loss": 1.1091,
+      "step": 1804
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00019513365484276056,
+      "loss": 0.9114,
+      "step": 1805
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0001951282756930481,
+      "loss": 0.8545,
+      "step": 1806
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00019512289364620653,
+      "loss": 1.0773,
+      "step": 1807
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0001951175087023997,
+      "loss": 0.9429,
+      "step": 1808
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00019511212086179163,
+      "loss": 1.0488,
+      "step": 1809
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00019510673012454645,
+      "loss": 1.0435,
+      "step": 1810
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00019510133649082824,
+      "loss": 0.8364,
+      "step": 1811
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00019509593996080136,
+      "loss": 0.9923,
+      "step": 1812
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0001950905405346301,
+      "loss": 1.0052,
+      "step": 1813
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00019508513821247898,
+      "loss": 1.015,
+      "step": 1814
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00019507973299451244,
+      "loss": 1.0449,
+      "step": 1815
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00019507432488089513,
+      "loss": 1.1019,
+      "step": 1816
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00019506891387179173,
+      "loss": 1.0237,
+      "step": 1817
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00019506349996736705,
+      "loss": 0.9909,
+      "step": 1818
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00019505808316778601,
+      "loss": 1.036,
+      "step": 1819
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00019505266347321355,
+      "loss": 0.8712,
+      "step": 1820
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00019504724088381466,
+      "loss": 0.9092,
+      "step": 1821
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0001950418153997546,
+      "loss": 1.0169,
+      "step": 1822
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0001950363870211985,
+      "loss": 1.1006,
+      "step": 1823
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00019503095574831177,
+      "loss": 1.0215,
+      "step": 1824
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00019502552158125975,
+      "loss": 1.0658,
+      "step": 1825
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00019502008452020796,
+      "loss": 1.1082,
+      "step": 1826
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00019501464456532204,
+      "loss": 1.0192,
+      "step": 1827
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00019500920171676756,
+      "loss": 1.0442,
+      "step": 1828
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00019500375597471034,
+      "loss": 1.0004,
+      "step": 1829
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00019499830733931623,
+      "loss": 0.9652,
+      "step": 1830
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00019499285581075118,
+      "loss": 0.9051,
+      "step": 1831
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0001949874013891812,
+      "loss": 1.0361,
+      "step": 1832
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0001949819440747724,
+      "loss": 1.1564,
+      "step": 1833
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00019497648386769098,
+      "loss": 0.9934,
+      "step": 1834
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00019497102076810322,
+      "loss": 0.9877,
+      "step": 1835
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00019496555477617556,
+      "loss": 1.0338,
+      "step": 1836
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00019496008589207437,
+      "loss": 1.0085,
+      "step": 1837
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00019495461411596628,
+      "loss": 1.0704,
+      "step": 1838
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00019494913944801792,
+      "loss": 1.0671,
+      "step": 1839
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.000194943661888396,
+      "loss": 0.9743,
+      "step": 1840
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00019493818143726733,
+      "loss": 0.9017,
+      "step": 1841
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00019493269809479888,
+      "loss": 1.1351,
+      "step": 1842
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00019492721186115756,
+      "loss": 1.0233,
+      "step": 1843
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0001949217227365105,
+      "loss": 1.0484,
+      "step": 1844
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00019491623072102486,
+      "loss": 0.9846,
+      "step": 1845
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00019491073581486793,
+      "loss": 1.0046,
+      "step": 1846
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00019490523801820705,
+      "loss": 1.0717,
+      "step": 1847
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00019489973733120958,
+      "loss": 1.0395,
+      "step": 1848
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00019489423375404315,
+      "loss": 0.9258,
+      "step": 1849
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0001948887272868753,
+      "loss": 0.9815,
+      "step": 1850
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00019488321792987377,
+      "loss": 0.8637,
+      "step": 1851
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0001948777056832063,
+      "loss": 1.0724,
+      "step": 1852
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0001948721905470408,
+      "loss": 1.0274,
+      "step": 1853
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00019486667252154524,
+      "loss": 0.9142,
+      "step": 1854
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00019486115160688764,
+      "loss": 0.959,
+      "step": 1855
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00019485562780323617,
+      "loss": 0.8675,
+      "step": 1856
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00019485010111075903,
+      "loss": 1.0416,
+      "step": 1857
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00019484457152962458,
+      "loss": 1.1091,
+      "step": 1858
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00019483903906000117,
+      "loss": 0.9196,
+      "step": 1859
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0001948335037020573,
+      "loss": 0.9064,
+      "step": 1860
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0001948279654559616,
+      "loss": 1.005,
+      "step": 1861
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00019482242432188268,
+      "loss": 1.0241,
+      "step": 1862
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00019481688029998932,
+      "loss": 0.9696,
+      "step": 1863
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00019481133339045037,
+      "loss": 0.9335,
+      "step": 1864
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00019480578359343476,
+      "loss": 1.0167,
+      "step": 1865
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00019480023090911146,
+      "loss": 1.0334,
+      "step": 1866
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00019479467533764963,
+      "loss": 0.8853,
+      "step": 1867
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00019478911687921847,
+      "loss": 1.0298,
+      "step": 1868
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00019478355553398722,
+      "loss": 1.0157,
+      "step": 1869
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00019477799130212527,
+      "loss": 0.9295,
+      "step": 1870
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0001947724241838021,
+      "loss": 0.9939,
+      "step": 1871
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00019476685417918726,
+      "loss": 1.0347,
+      "step": 1872
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00019476128128845032,
+      "loss": 0.8973,
+      "step": 1873
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00019475570551176106,
+      "loss": 1.144,
+      "step": 1874
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00019475012684928928,
+      "loss": 0.9658,
+      "step": 1875
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00019474454530120488,
+      "loss": 0.9404,
+      "step": 1876
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00019473896086767781,
+      "loss": 0.9696,
+      "step": 1877
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00019473337354887818,
+      "loss": 0.9067,
+      "step": 1878
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0001947277833449762,
+      "loss": 0.9724,
+      "step": 1879
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00019472219025614201,
+      "loss": 1.0052,
+      "step": 1880
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.000194716594282546,
+      "loss": 0.8851,
+      "step": 1881
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00019471099542435862,
+      "loss": 0.993,
+      "step": 1882
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00019470539368175035,
+      "loss": 1.0343,
+      "step": 1883
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00019469978905489185,
+      "loss": 0.9899,
+      "step": 1884
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0001946941815439537,
+      "loss": 0.897,
+      "step": 1885
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00019468857114910676,
+      "loss": 0.9725,
+      "step": 1886
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00019468295787052184,
+      "loss": 1.0217,
+      "step": 1887
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00019467734170836997,
+      "loss": 0.9931,
+      "step": 1888
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0001946717226628221,
+      "loss": 1.0113,
+      "step": 1889
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00019466610073404943,
+      "loss": 1.0013,
+      "step": 1890
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0001946604759222231,
+      "loss": 1.1052,
+      "step": 1891
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0001946548482275145,
+      "loss": 1.0741,
+      "step": 1892
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.000194649217650095,
+      "loss": 0.8772,
+      "step": 1893
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00019464358419013603,
+      "loss": 1.0945,
+      "step": 1894
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00019463794784780922,
+      "loss": 1.1438,
+      "step": 1895
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00019463230862328614,
+      "loss": 1.0315,
+      "step": 1896
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00019462666651673863,
+      "loss": 0.9256,
+      "step": 1897
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00019462102152833846,
+      "loss": 0.9463,
+      "step": 1898
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00019461537365825758,
+      "loss": 0.9555,
+      "step": 1899
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00019460972290666795,
+      "loss": 0.946,
+      "step": 1900
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00019460406927374169,
+      "loss": 1.0297,
+      "step": 1901
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.000194598412759651,
+      "loss": 0.9769,
+      "step": 1902
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00019459275336456813,
+      "loss": 0.9333,
+      "step": 1903
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00019458709108866545,
+      "loss": 1.0293,
+      "step": 1904
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00019458142593211536,
+      "loss": 1.0169,
+      "step": 1905
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00019457575789509046,
+      "loss": 0.9419,
+      "step": 1906
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00019457008697776332,
+      "loss": 0.7803,
+      "step": 1907
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00019456441318030664,
+      "loss": 1.0529,
+      "step": 1908
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00019455873650289326,
+      "loss": 1.0864,
+      "step": 1909
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00019455305694569605,
+      "loss": 1.1737,
+      "step": 1910
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00019454737450888798,
+      "loss": 1.0067,
+      "step": 1911
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0001945416891926421,
+      "loss": 1.0207,
+      "step": 1912
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00019453600099713153,
+      "loss": 0.8605,
+      "step": 1913
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00019453030992252956,
+      "loss": 0.9471,
+      "step": 1914
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00019452461596900945,
+      "loss": 0.9972,
+      "step": 1915
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00019451891913674466,
+      "loss": 1.0686,
+      "step": 1916
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00019451321942590868,
+      "loss": 1.0528,
+      "step": 1917
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00019450751683667506,
+      "loss": 0.8925,
+      "step": 1918
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0001945018113692175,
+      "loss": 1.064,
+      "step": 1919
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00019449610302370976,
+      "loss": 0.889,
+      "step": 1920
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0001944903918003257,
+      "loss": 1.0651,
+      "step": 1921
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00019448467769923923,
+      "loss": 1.1264,
+      "step": 1922
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00019447896072062438,
+      "loss": 1.0887,
+      "step": 1923
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00019447324086465523,
+      "loss": 1.0213,
+      "step": 1924
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00019446751813150605,
+      "loss": 0.9762,
+      "step": 1925
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00019446179252135106,
+      "loss": 0.9922,
+      "step": 1926
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00019445606403436467,
+      "loss": 0.9927,
+      "step": 1927
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00019445033267072133,
+      "loss": 0.931,
+      "step": 1928
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00019444459843059557,
+      "loss": 1.0093,
+      "step": 1929
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00019443886131416207,
+      "loss": 1.0516,
+      "step": 1930
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0001944331213215955,
+      "loss": 0.7737,
+      "step": 1931
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0001944273784530707,
+      "loss": 1.0385,
+      "step": 1932
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0001944216327087626,
+      "loss": 1.1465,
+      "step": 1933
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00019441588408884613,
+      "loss": 0.955,
+      "step": 1934
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00019441013259349638,
+      "loss": 1.0006,
+      "step": 1935
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00019440437822288855,
+      "loss": 1.0891,
+      "step": 1936
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00019439862097719784,
+      "loss": 1.1088,
+      "step": 1937
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0001943928608565996,
+      "loss": 0.9809,
+      "step": 1938
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00019438709786126927,
+      "loss": 0.9992,
+      "step": 1939
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00019438133199138236,
+      "loss": 0.9645,
+      "step": 1940
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00019437556324711444,
+      "loss": 1.1127,
+      "step": 1941
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00019436979162864123,
+      "loss": 1.0032,
+      "step": 1942
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0001943640171361385,
+      "loss": 0.9896,
+      "step": 1943
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00019435823976978207,
+      "loss": 0.9997,
+      "step": 1944
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00019435245952974798,
+      "loss": 0.9897,
+      "step": 1945
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00019434667641621216,
+      "loss": 1.1225,
+      "step": 1946
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0001943408904293508,
+      "loss": 0.9963,
+      "step": 1947
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0001943351015693401,
+      "loss": 0.8727,
+      "step": 1948
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0001943293098363564,
+      "loss": 1.0832,
+      "step": 1949
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.000194323515230576,
+      "loss": 0.9911,
+      "step": 1950
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00019431771775217543,
+      "loss": 0.955,
+      "step": 1951
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00019431191740133122,
+      "loss": 1.0014,
+      "step": 1952
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00019430611417822007,
+      "loss": 0.8227,
+      "step": 1953
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00019430030808301866,
+      "loss": 0.9481,
+      "step": 1954
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00019429449911590387,
+      "loss": 1.0541,
+      "step": 1955
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00019428868727705258,
+      "loss": 1.1773,
+      "step": 1956
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00019428287256664176,
+      "loss": 1.0008,
+      "step": 1957
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0001942770549848486,
+      "loss": 1.003,
+      "step": 1958
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00019427123453185014,
+      "loss": 1.0416,
+      "step": 1959
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00019426541120782373,
+      "loss": 0.9317,
+      "step": 1960
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0001942595850129467,
+      "loss": 0.9574,
+      "step": 1961
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00019425375594739648,
+      "loss": 0.9064,
+      "step": 1962
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0001942479240113506,
+      "loss": 1.062,
+      "step": 1963
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00019424208920498667,
+      "loss": 0.9436,
+      "step": 1964
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00019423625152848237,
+      "loss": 0.9972,
+      "step": 1965
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0001942304109820155,
+      "loss": 0.9068,
+      "step": 1966
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00019422456756576395,
+      "loss": 1.0602,
+      "step": 1967
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0001942187212799057,
+      "loss": 0.9551,
+      "step": 1968
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00019421287212461873,
+      "loss": 0.9158,
+      "step": 1969
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0001942070201000812,
+      "loss": 1.0254,
+      "step": 1970
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00019420116520647135,
+      "loss": 0.9472,
+      "step": 1971
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00019419530744396753,
+      "loss": 0.9337,
+      "step": 1972
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00019418944681274804,
+      "loss": 1.0212,
+      "step": 1973
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00019418358331299145,
+      "loss": 1.0588,
+      "step": 1974
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00019417771694487626,
+      "loss": 0.8391,
+      "step": 1975
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00019417184770858124,
+      "loss": 1.0849,
+      "step": 1976
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00019416597560428502,
+      "loss": 0.8121,
+      "step": 1977
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0001941601006321665,
+      "loss": 0.9066,
+      "step": 1978
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0001941542227924046,
+      "loss": 0.9758,
+      "step": 1979
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0001941483420851783,
+      "loss": 0.8931,
+      "step": 1980
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00019414245851066673,
+      "loss": 1.0424,
+      "step": 1981
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00019413657206904903,
+      "loss": 0.9463,
+      "step": 1982
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0001941306827605045,
+      "loss": 0.8909,
+      "step": 1983
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0001941247905852125,
+      "loss": 1.072,
+      "step": 1984
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0001941188955433525,
+      "loss": 0.9221,
+      "step": 1985
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00019411299763510398,
+      "loss": 1.0199,
+      "step": 1986
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0001941070968606466,
+      "loss": 0.8926,
+      "step": 1987
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00019410119322016005,
+      "loss": 1.0343,
+      "step": 1988
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00019409528671382413,
+      "loss": 0.9838,
+      "step": 1989
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00019408937734181873,
+      "loss": 1.0195,
+      "step": 1990
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0001940834651043238,
+      "loss": 0.9841,
+      "step": 1991
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00019407755000151942,
+      "loss": 1.0756,
+      "step": 1992
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0001940716320335857,
+      "loss": 1.0505,
+      "step": 1993
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00019406571120070295,
+      "loss": 1.0449,
+      "step": 1994
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00019405978750305138,
+      "loss": 1.0549,
+      "step": 1995
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00019405386094081145,
+      "loss": 0.8532,
+      "step": 1996
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00019404793151416366,
+      "loss": 0.9926,
+      "step": 1997
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00019404199922328863,
+      "loss": 1.1055,
+      "step": 1998
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00019403606406836693,
+      "loss": 0.9538,
+      "step": 1999
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0001940301260495794,
+      "loss": 0.9736,
+      "step": 2000
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00019402418516710682,
+      "loss": 0.9153,
+      "step": 2001
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0001940182414211302,
+      "loss": 1.1754,
+      "step": 2002
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00019401229481183046,
+      "loss": 1.1609,
+      "step": 2003
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00019400634533938874,
+      "loss": 1.0047,
+      "step": 2004
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00019400039300398626,
+      "loss": 0.9231,
+      "step": 2005
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0001939944378058043,
+      "loss": 0.9602,
+      "step": 2006
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00019398847974502417,
+      "loss": 1.107,
+      "step": 2007
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00019398251882182734,
+      "loss": 1.0019,
+      "step": 2008
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0001939765550363954,
+      "loss": 1.0037,
+      "step": 2009
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00019397058838890993,
+      "loss": 0.9579,
+      "step": 2010
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00019396461887955267,
+      "loss": 0.912,
+      "step": 2011
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00019395864650850538,
+      "loss": 0.986,
+      "step": 2012
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00019395267127595003,
+      "loss": 0.9479,
+      "step": 2013
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0001939466931820685,
+      "loss": 1.0991,
+      "step": 2014
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00019394071222704292,
+      "loss": 1.0301,
+      "step": 2015
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00019393472841105537,
+      "loss": 1.1119,
+      "step": 2016
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0001939287417342882,
+      "loss": 1.0107,
+      "step": 2017
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0001939227521969236,
+      "loss": 0.9891,
+      "step": 2018
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00019391675979914407,
+      "loss": 0.9367,
+      "step": 2019
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00019391076454113208,
+      "loss": 1.0491,
+      "step": 2020
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00019390476642307025,
+      "loss": 0.9784,
+      "step": 2021
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00019389876544514118,
+      "loss": 1.0165,
+      "step": 2022
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00019389276160752773,
+      "loss": 0.9441,
+      "step": 2023
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00019388675491041263,
+      "loss": 0.9094,
+      "step": 2024
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0001938807453539789,
+      "loss": 0.9322,
+      "step": 2025
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00019387473293840953,
+      "loss": 1.0377,
+      "step": 2026
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00019386871766388764,
+      "loss": 0.928,
+      "step": 2027
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00019386269953059642,
+      "loss": 0.9137,
+      "step": 2028
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00019385667853871914,
+      "loss": 0.9228,
+      "step": 2029
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00019385065468843917,
+      "loss": 1.011,
+      "step": 2030
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00019384462797993998,
+      "loss": 0.989,
+      "step": 2031
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00019383859841340512,
+      "loss": 1.0259,
+      "step": 2032
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0001938325659890182,
+      "loss": 0.8039,
+      "step": 2033
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00019382653070696294,
+      "loss": 1.0337,
+      "step": 2034
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0001938204925674232,
+      "loss": 0.9593,
+      "step": 2035
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00019381445157058274,
+      "loss": 1.0367,
+      "step": 2036
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00019380840771662566,
+      "loss": 0.9563,
+      "step": 2037
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.000193802361005736,
+      "loss": 0.8675,
+      "step": 2038
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00019379631143809787,
+      "loss": 0.8572,
+      "step": 2039
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00019379025901389554,
+      "loss": 0.9291,
+      "step": 2040
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00019378420373331333,
+      "loss": 1.0037,
+      "step": 2041
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0001937781455965357,
+      "loss": 0.9724,
+      "step": 2042
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00019377208460374706,
+      "loss": 0.9284,
+      "step": 2043
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00019376602075513207,
+      "loss": 0.8702,
+      "step": 2044
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00019375995405087536,
+      "loss": 0.9158,
+      "step": 2045
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00019375388449116172,
+      "loss": 1.0224,
+      "step": 2046
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00019374781207617597,
+      "loss": 1.0363,
+      "step": 2047
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0001937417368061031,
+      "loss": 1.1012,
+      "step": 2048
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0001937356586811281,
+      "loss": 1.0711,
+      "step": 2049
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00019372957770143602,
+      "loss": 0.9952,
+      "step": 2050
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00019372349386721216,
+      "loss": 0.9262,
+      "step": 2051
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0001937174071786417,
+      "loss": 1.0356,
+      "step": 2052
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00019371131763591012,
+      "loss": 1.1124,
+      "step": 2053
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0001937052252392028,
+      "loss": 0.9899,
+      "step": 2054
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0001936991299887053,
+      "loss": 0.9177,
+      "step": 2055
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00019369303188460326,
+      "loss": 0.9319,
+      "step": 2056
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0001936869309270824,
+      "loss": 0.9709,
+      "step": 2057
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00019368082711632853,
+      "loss": 1.0485,
+      "step": 2058
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00019367472045252751,
+      "loss": 1.0685,
+      "step": 2059
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00019366861093586532,
+      "loss": 0.9058,
+      "step": 2060
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00019366249856652809,
+      "loss": 0.9453,
+      "step": 2061
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00019365638334470188,
+      "loss": 0.9094,
+      "step": 2062
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.000193650265270573,
+      "loss": 1.0097,
+      "step": 2063
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00019364414434432773,
+      "loss": 0.9863,
+      "step": 2064
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0001936380205661525,
+      "loss": 1.046,
+      "step": 2065
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00019363189393623384,
+      "loss": 0.9647,
+      "step": 2066
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0001936257644547583,
+      "loss": 0.9153,
+      "step": 2067
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00019361963212191253,
+      "loss": 0.9531,
+      "step": 2068
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00019361349693788334,
+      "loss": 1.0815,
+      "step": 2069
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00019360735890285757,
+      "loss": 0.91,
+      "step": 2070
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00019360121801702214,
+      "loss": 0.9042,
+      "step": 2071
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00019359507428056406,
+      "loss": 1.0263,
+      "step": 2072
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00019358892769367046,
+      "loss": 1.0439,
+      "step": 2073
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0001935827782565285,
+      "loss": 0.9894,
+      "step": 2074
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00019357662596932552,
+      "loss": 1.0729,
+      "step": 2075
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00019357047083224882,
+      "loss": 1.0581,
+      "step": 2076
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00019356431284548592,
+      "loss": 1.0122,
+      "step": 2077
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00019355815200922428,
+      "loss": 1.1223,
+      "step": 2078
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0001935519883236516,
+      "loss": 1.0252,
+      "step": 2079
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0001935458217889556,
+      "loss": 0.9183,
+      "step": 2080
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00019353965240532402,
+      "loss": 0.9625,
+      "step": 2081
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0001935334801729448,
+      "loss": 0.9331,
+      "step": 2082
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00019352730509200586,
+      "loss": 0.9585,
+      "step": 2083
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00019352112716269532,
+      "loss": 1.0105,
+      "step": 2084
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00019351494638520132,
+      "loss": 0.9867,
+      "step": 2085
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00019350876275971209,
+      "loss": 1.0158,
+      "step": 2086
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0001935025762864159,
+      "loss": 0.987,
+      "step": 2087
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00019349638696550123,
+      "loss": 0.9,
+      "step": 2088
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00019349019479715656,
+      "loss": 0.8773,
+      "step": 2089
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00019348399978157043,
+      "loss": 0.9757,
+      "step": 2090
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00019347780191893157,
+      "loss": 1.0264,
+      "step": 2091
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0001934716012094287,
+      "loss": 0.9823,
+      "step": 2092
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00019346539765325064,
+      "loss": 0.8942,
+      "step": 2093
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0001934591912505864,
+      "loss": 0.9558,
+      "step": 2094
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00019345298200162492,
+      "loss": 0.8608,
+      "step": 2095
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00019344676990655532,
+      "loss": 1.0808,
+      "step": 2096
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0001934405549655668,
+      "loss": 0.9549,
+      "step": 2097
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0001934343371788486,
+      "loss": 1.0059,
+      "step": 2098
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00019342811654659013,
+      "loss": 1.0148,
+      "step": 2099
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00019342189306898087,
+      "loss": 0.9668,
+      "step": 2100
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00019341566674621028,
+      "loss": 1.049,
+      "step": 2101
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.000193409437578468,
+      "loss": 1.0229,
+      "step": 2102
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00019340320556594375,
+      "loss": 0.988,
+      "step": 2103
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00019339697070882731,
+      "loss": 0.9942,
+      "step": 2104
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00019339073300730862,
+      "loss": 0.9565,
+      "step": 2105
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00019338449246157757,
+      "loss": 0.9543,
+      "step": 2106
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00019337824907182427,
+      "loss": 0.9882,
+      "step": 2107
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00019337200283823884,
+      "loss": 1.0428,
+      "step": 2108
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00019336575376101154,
+      "loss": 1.0354,
+      "step": 2109
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00019335950184033263,
+      "loss": 0.8947,
+      "step": 2110
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00019335324707639253,
+      "loss": 0.9184,
+      "step": 2111
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00019334698946938178,
+      "loss": 0.9195,
+      "step": 2112
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0001933407290194909,
+      "loss": 1.0132,
+      "step": 2113
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00019333446572691054,
+      "loss": 1.052,
+      "step": 2114
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0001933281995918315,
+      "loss": 1.0483,
+      "step": 2115
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00019332193061444462,
+      "loss": 0.8911,
+      "step": 2116
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00019331565879494075,
+      "loss": 0.9345,
+      "step": 2117
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00019330938413351096,
+      "loss": 0.9798,
+      "step": 2118
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0001933031066303463,
+      "loss": 1.007,
+      "step": 2119
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.000193296826285638,
+      "loss": 0.9291,
+      "step": 2120
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00019329054309957731,
+      "loss": 0.877,
+      "step": 2121
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00019328425707235556,
+      "loss": 0.9171,
+      "step": 2122
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00019327796820416425,
+      "loss": 0.9137,
+      "step": 2123
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0001932716764951948,
+      "loss": 1.0237,
+      "step": 2124
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00019326538194563894,
+      "loss": 1.1492,
+      "step": 2125
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0001932590845556883,
+      "loss": 1.0233,
+      "step": 2126
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00019325278432553468,
+      "loss": 0.9939,
+      "step": 2127
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00019324648125537,
+      "loss": 0.9682,
+      "step": 2128
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00019324017534538615,
+      "loss": 0.928,
+      "step": 2129
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0001932338665957752,
+      "loss": 1.0144,
+      "step": 2130
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0001932275550067293,
+      "loss": 1.061,
+      "step": 2131
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00019322124057844067,
+      "loss": 0.9299,
+      "step": 2132
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00019321492331110158,
+      "loss": 1.1439,
+      "step": 2133
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00019320860320490448,
+      "loss": 0.9307,
+      "step": 2134
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0001932022802600418,
+      "loss": 0.9673,
+      "step": 2135
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0001931959544767061,
+      "loss": 1.0421,
+      "step": 2136
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00019318962585509008,
+      "loss": 1.0606,
+      "step": 2137
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00019318329439538644,
+      "loss": 1.0527,
+      "step": 2138
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.000193176960097788,
+      "loss": 1.0037,
+      "step": 2139
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00019317062296248772,
+      "loss": 1.0834,
+      "step": 2140
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00019316428298967855,
+      "loss": 1.0249,
+      "step": 2141
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00019315794017955358,
+      "loss": 1.0143,
+      "step": 2142
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.000193151594532306,
+      "loss": 0.9762,
+      "step": 2143
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00019314524604812904,
+      "loss": 0.9669,
+      "step": 2144
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00019313889472721604,
+      "loss": 0.9612,
+      "step": 2145
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0001931325405697605,
+      "loss": 0.8986,
+      "step": 2146
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00019312618357595582,
+      "loss": 1.0602,
+      "step": 2147
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00019311982374599568,
+      "loss": 1.0251,
+      "step": 2148
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00019311346108007377,
+      "loss": 1.06,
+      "step": 2149
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00019310709557838384,
+      "loss": 0.8027,
+      "step": 2150
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0001931007272411198,
+      "loss": 0.9615,
+      "step": 2151
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00019309435606847548,
+      "loss": 0.9369,
+      "step": 2152
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00019308798206064503,
+      "loss": 1.0805,
+      "step": 2153
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0001930816052178225,
+      "loss": 0.9617,
+      "step": 2154
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00019307522554020217,
+      "loss": 1.0116,
+      "step": 2155
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00019306884302797825,
+      "loss": 0.9777,
+      "step": 2156
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0001930624576813452,
+      "loss": 1.0857,
+      "step": 2157
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0001930560695004974,
+      "loss": 0.8586,
+      "step": 2158
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00019304967848562948,
+      "loss": 0.8639,
+      "step": 2159
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.000193043284636936,
+      "loss": 1.0095,
+      "step": 2160
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00019303688795461178,
+      "loss": 0.8814,
+      "step": 2161
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00019303048843885155,
+      "loss": 1.0404,
+      "step": 2162
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00019302408608985023,
+      "loss": 0.9212,
+      "step": 2163
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00019301768090780282,
+      "loss": 1.047,
+      "step": 2164
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00019301127289290439,
+      "loss": 0.9821,
+      "step": 2165
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0001930048620453501,
+      "loss": 0.9833,
+      "step": 2166
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00019299844836533515,
+      "loss": 0.8858,
+      "step": 2167
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00019299203185305487,
+      "loss": 0.9887,
+      "step": 2168
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00019298561250870476,
+      "loss": 1.0067,
+      "step": 2169
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00019297919033248021,
+      "loss": 1.0101,
+      "step": 2170
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0001929727653245769,
+      "loss": 1.0802,
+      "step": 2171
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00019296633748519044,
+      "loss": 0.8422,
+      "step": 2172
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0001929599068145166,
+      "loss": 1.0433,
+      "step": 2173
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00019295347331275126,
+      "loss": 0.911,
+      "step": 2174
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0001929470369800903,
+      "loss": 1.0173,
+      "step": 2175
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00019294059781672983,
+      "loss": 0.9517,
+      "step": 2176
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00019293415582286586,
+      "loss": 1.0215,
+      "step": 2177
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0001929277109986946,
+      "loss": 0.9704,
+      "step": 2178
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00019292126334441234,
+      "loss": 1.0863,
+      "step": 2179
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00019291481286021545,
+      "loss": 0.9483,
+      "step": 2180
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00019290835954630038,
+      "loss": 0.9278,
+      "step": 2181
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00019290190340286365,
+      "loss": 0.9514,
+      "step": 2182
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00019289544443010188,
+      "loss": 0.996,
+      "step": 2183
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00019288898262821182,
+      "loss": 1.0558,
+      "step": 2184
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00019288251799739022,
+      "loss": 1.0634,
+      "step": 2185
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00019287605053783396,
+      "loss": 1.0716,
+      "step": 2186
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00019286958024974002,
+      "loss": 0.9985,
+      "step": 2187
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00019286310713330544,
+      "loss": 1.0714,
+      "step": 2188
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00019285663118872738,
+      "loss": 0.9219,
+      "step": 2189
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00019285015241620308,
+      "loss": 0.9778,
+      "step": 2190
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0001928436708159298,
+      "loss": 0.8832,
+      "step": 2191
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00019283718638810497,
+      "loss": 1.0275,
+      "step": 2192
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00019283069913292607,
+      "loss": 0.9442,
+      "step": 2193
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00019282420905059064,
+      "loss": 0.9151,
+      "step": 2194
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00019281771614129638,
+      "loss": 1.0606,
+      "step": 2195
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00019281122040524102,
+      "loss": 1.0199,
+      "step": 2196
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00019280472184262238,
+      "loss": 0.899,
+      "step": 2197
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00019279822045363833,
+      "loss": 0.9877,
+      "step": 2198
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00019279171623848694,
+      "loss": 1.0479,
+      "step": 2199
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00019278520919736628,
+      "loss": 0.7998,
+      "step": 2200
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0001927786993304745,
+      "loss": 0.9011,
+      "step": 2201
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00019277218663800983,
+      "loss": 0.9434,
+      "step": 2202
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0001927656711201707,
+      "loss": 0.917,
+      "step": 2203
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00019275915277715548,
+      "loss": 0.8733,
+      "step": 2204
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0001927526316091627,
+      "loss": 0.9503,
+      "step": 2205
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00019274610761639096,
+      "loss": 0.9859,
+      "step": 2206
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00019273958079903892,
+      "loss": 0.9703,
+      "step": 2207
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0001927330511573054,
+      "loss": 0.8689,
+      "step": 2208
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00019272651869138924,
+      "loss": 1.0171,
+      "step": 2209
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00019271998340148941,
+      "loss": 1.0023,
+      "step": 2210
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00019271344528780485,
+      "loss": 0.928,
+      "step": 2211
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00019270690435053481,
+      "loss": 0.943,
+      "step": 2212
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00019270036058987843,
+      "loss": 1.0894,
+      "step": 2213
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.000192693814006035,
+      "loss": 0.9944,
+      "step": 2214
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00019268726459920386,
+      "loss": 1.0579,
+      "step": 2215
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00019268071236958455,
+      "loss": 1.0415,
+      "step": 2216
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0001926741573173765,
+      "loss": 1.0605,
+      "step": 2217
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00019266759944277953,
+      "loss": 1.0679,
+      "step": 2218
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0001926610387459932,
+      "loss": 1.2277,
+      "step": 2219
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00019265447522721736,
+      "loss": 0.9654,
+      "step": 2220
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00019264790888665192,
+      "loss": 0.9098,
+      "step": 2221
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00019264133972449684,
+      "loss": 1.0352,
+      "step": 2222
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0001926347677409522,
+      "loss": 0.9677,
+      "step": 2223
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00019262819293621814,
+      "loss": 1.004,
+      "step": 2224
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00019262161531049488,
+      "loss": 1.0285,
+      "step": 2225
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00019261503486398275,
+      "loss": 0.8849,
+      "step": 2226
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0001926084515968822,
+      "loss": 0.8765,
+      "step": 2227
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00019260186550939366,
+      "loss": 0.9182,
+      "step": 2228
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00019259527660171774,
+      "loss": 0.8799,
+      "step": 2229
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0001925886848740551,
+      "loss": 0.9911,
+      "step": 2230
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0001925820903266065,
+      "loss": 1.0164,
+      "step": 2231
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00019257549295957277,
+      "loss": 0.9943,
+      "step": 2232
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00019256889277315483,
+      "loss": 0.9732,
+      "step": 2233
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0001925622897675537,
+      "loss": 1.1329,
+      "step": 2234
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00019255568394297044,
+      "loss": 0.9437,
+      "step": 2235
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0001925490752996063,
+      "loss": 1.0044,
+      "step": 2236
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00019254246383766245,
+      "loss": 1.0494,
+      "step": 2237
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0001925358495573403,
+      "loss": 1.1255,
+      "step": 2238
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00019252923245884128,
+      "loss": 0.9667,
+      "step": 2239
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00019252261254236694,
+      "loss": 0.894,
+      "step": 2240
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00019251598980811886,
+      "loss": 1.0424,
+      "step": 2241
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00019250936425629873,
+      "loss": 0.8505,
+      "step": 2242
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00019250273588710832,
+      "loss": 0.9659,
+      "step": 2243
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00019249610470074954,
+      "loss": 0.9633,
+      "step": 2244
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00019248947069742432,
+      "loss": 0.9338,
+      "step": 2245
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00019248283387733467,
+      "loss": 1.0264,
+      "step": 2246
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00019247619424068277,
+      "loss": 0.9937,
+      "step": 2247
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0001924695517876708,
+      "loss": 0.9373,
+      "step": 2248
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00019246290651850104,
+      "loss": 1.0967,
+      "step": 2249
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0001924562584333759,
+      "loss": 0.9418,
+      "step": 2250
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00019244960753249783,
+      "loss": 0.8806,
+      "step": 2251
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00019244295381606942,
+      "loss": 1.0279,
+      "step": 2252
+    },
+    {
+      "epoch": 0.5,
+      "eval_loss": 0.6187470555305481,
+      "eval_runtime": 322.9541,
+      "eval_samples_per_second": 56.166,
+      "eval_steps_per_second": 28.084,
+      "step": 2252
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00019243629728429325,
+      "loss": 1.1856,
+      "step": 2253
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00019242963793737207,
+      "loss": 1.0412,
+      "step": 2254
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0001924229757755087,
+      "loss": 0.9558,
+      "step": 2255
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.000192416310798906,
+      "loss": 0.9817,
+      "step": 2256
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00019240964300776703,
+      "loss": 0.9407,
+      "step": 2257
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0001924029724022948,
+      "loss": 0.9469,
+      "step": 2258
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00019239629898269245,
+      "loss": 1.0785,
+      "step": 2259
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00019238962274916322,
+      "loss": 1.0259,
+      "step": 2260
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0001923829437019105,
+      "loss": 0.9049,
+      "step": 2261
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00019237626184113763,
+      "loss": 0.9227,
+      "step": 2262
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00019236957716704814,
+      "loss": 0.8747,
+      "step": 2263
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0001923628896798456,
+      "loss": 0.9309,
+      "step": 2264
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00019235619937973367,
+      "loss": 0.8997,
+      "step": 2265
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0001923495062669161,
+      "loss": 1.0323,
+      "step": 2266
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00019234281034159674,
+      "loss": 0.9668,
+      "step": 2267
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00019233611160397954,
+      "loss": 0.981,
+      "step": 2268
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0001923294100542685,
+      "loss": 1.1254,
+      "step": 2269
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00019232270569266767,
+      "loss": 0.8491,
+      "step": 2270
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00019231599851938125,
+      "loss": 0.8864,
+      "step": 2271
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00019230928853461354,
+      "loss": 0.9139,
+      "step": 2272
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00019230257573856887,
+      "loss": 0.8401,
+      "step": 2273
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00019229586013145167,
+      "loss": 1.0594,
+      "step": 2274
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00019228914171346648,
+      "loss": 0.9667,
+      "step": 2275
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0001922824204848179,
+      "loss": 0.982,
+      "step": 2276
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00019227569644571062,
+      "loss": 0.9874,
+      "step": 2277
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00019226896959634944,
+      "loss": 0.8824,
+      "step": 2278
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0001922622399369392,
+      "loss": 1.0304,
+      "step": 2279
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00019225550746768485,
+      "loss": 1.1114,
+      "step": 2280
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00019224877218879146,
+      "loss": 0.9353,
+      "step": 2281
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00019224203410046412,
+      "loss": 0.9839,
+      "step": 2282
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00019223529320290807,
+      "loss": 0.9491,
+      "step": 2283
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0001922285494963286,
+      "loss": 0.9283,
+      "step": 2284
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00019222180298093106,
+      "loss": 1.161,
+      "step": 2285
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00019221505365692095,
+      "loss": 0.9717,
+      "step": 2286
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00019220830152450374,
+      "loss": 1.0024,
+      "step": 2287
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0001922015465838852,
+      "loss": 0.9105,
+      "step": 2288
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00019219478883527094,
+      "loss": 0.9798,
+      "step": 2289
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0001921880282788668,
+      "loss": 1.0659,
+      "step": 2290
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0001921812649148787,
+      "loss": 0.9892,
+      "step": 2291
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00019217449874351256,
+      "loss": 1.0361,
+      "step": 2292
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00019216772976497453,
+      "loss": 1.0947,
+      "step": 2293
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00019216095797947066,
+      "loss": 1.0533,
+      "step": 2294
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00019215418338720727,
+      "loss": 1.0689,
+      "step": 2295
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00019214740598839063,
+      "loss": 1.0003,
+      "step": 2296
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00019214062578322717,
+      "loss": 1.0324,
+      "step": 2297
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00019213384277192336,
+      "loss": 1.0301,
+      "step": 2298
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0001921270569546858,
+      "loss": 1.0329,
+      "step": 2299
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00019212026833172112,
+      "loss": 1.0277,
+      "step": 2300
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00019211347690323608,
+      "loss": 1.0478,
+      "step": 2301
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00019210668266943755,
+      "loss": 1.0367,
+      "step": 2302
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00019209988563053242,
+      "loss": 0.9704,
+      "step": 2303
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00019209308578672763,
+      "loss": 0.9897,
+      "step": 2304
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0001920862831382304,
+      "loss": 0.9141,
+      "step": 2305
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00019207947768524778,
+      "loss": 0.9729,
+      "step": 2306
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0001920726694279871,
+      "loss": 1.0569,
+      "step": 2307
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0001920658583666557,
+      "loss": 0.9646,
+      "step": 2308
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.000192059044501461,
+      "loss": 0.9604,
+      "step": 2309
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0001920522278326105,
+      "loss": 0.9737,
+      "step": 2310
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00019204540836031183,
+      "loss": 1.0229,
+      "step": 2311
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00019203858608477265,
+      "loss": 1.092,
+      "step": 2312
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0001920317610062008,
+      "loss": 0.9216,
+      "step": 2313
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00019202493312480402,
+      "loss": 0.9925,
+      "step": 2314
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00019201810244079035,
+      "loss": 0.9922,
+      "step": 2315
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00019201126895436777,
+      "loss": 1.0126,
+      "step": 2316
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00019200443266574442,
+      "loss": 0.9144,
+      "step": 2317
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0001919975935751285,
+      "loss": 0.9459,
+      "step": 2318
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00019199075168272825,
+      "loss": 0.9634,
+      "step": 2319
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00019198390698875207,
+      "loss": 0.9664,
+      "step": 2320
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00019197705949340846,
+      "loss": 0.9753,
+      "step": 2321
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00019197020919690586,
+      "loss": 0.9129,
+      "step": 2322
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.000191963356099453,
+      "loss": 1.0068,
+      "step": 2323
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00019195650020125852,
+      "loss": 1.0667,
+      "step": 2324
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00019194964150253126,
+      "loss": 1.0631,
+      "step": 2325
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00019194278000348006,
+      "loss": 0.8465,
+      "step": 2326
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0001919359157043139,
+      "loss": 0.8713,
+      "step": 2327
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00019192904860524187,
+      "loss": 0.9928,
+      "step": 2328
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00019192217870647306,
+      "loss": 0.9935,
+      "step": 2329
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00019191530600821673,
+      "loss": 1.0142,
+      "step": 2330
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00019190843051068214,
+      "loss": 0.8805,
+      "step": 2331
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00019190155221407874,
+      "loss": 0.9415,
+      "step": 2332
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00019189467111861594,
+      "loss": 1.0389,
+      "step": 2333
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00019188778722450337,
+      "loss": 0.9585,
+      "step": 2334
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00019188090053195067,
+      "loss": 1.0597,
+      "step": 2335
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00019187401104116752,
+      "loss": 0.8868,
+      "step": 2336
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0001918671187523638,
+      "loss": 0.988,
+      "step": 2337
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00019186022366574936,
+      "loss": 1.034,
+      "step": 2338
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00019185332578153426,
+      "loss": 0.9703,
+      "step": 2339
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0001918464250999285,
+      "loss": 0.9331,
+      "step": 2340
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0001918395216211423,
+      "loss": 0.9424,
+      "step": 2341
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00019183261534538585,
+      "loss": 0.9836,
+      "step": 2342
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00019182570627286952,
+      "loss": 0.8522,
+      "step": 2343
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00019181879440380374,
+      "loss": 1.0359,
+      "step": 2344
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00019181187973839895,
+      "loss": 0.9822,
+      "step": 2345
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00019180496227686578,
+      "loss": 1.0865,
+      "step": 2346
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00019179804201941493,
+      "loss": 1.0087,
+      "step": 2347
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0001917911189662571,
+      "loss": 0.9934,
+      "step": 2348
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00019178419311760312,
+      "loss": 1.0393,
+      "step": 2349
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.000191777264473664,
+      "loss": 0.9826,
+      "step": 2350
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00019177033303465065,
+      "loss": 1.0749,
+      "step": 2351
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00019176339880077424,
+      "loss": 1.0953,
+      "step": 2352
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00019175646177224591,
+      "loss": 1.1179,
+      "step": 2353
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00019174952194927697,
+      "loss": 0.9687,
+      "step": 2354
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00019174257933207876,
+      "loss": 1.0982,
+      "step": 2355
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00019173563392086267,
+      "loss": 0.961,
+      "step": 2356
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00019172868571584026,
+      "loss": 0.9624,
+      "step": 2357
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00019172173471722315,
+      "loss": 1.0337,
+      "step": 2358
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.000191714780925223,
+      "loss": 0.996,
+      "step": 2359
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00019170782434005162,
+      "loss": 0.9655,
+      "step": 2360
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00019170086496192082,
+      "loss": 1.1095,
+      "step": 2361
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00019169390279104266,
+      "loss": 1.0724,
+      "step": 2362
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00019168693782762903,
+      "loss": 0.8794,
+      "step": 2363
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00019167997007189214,
+      "loss": 0.9891,
+      "step": 2364
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00019167299952404417,
+      "loss": 1.0665,
+      "step": 2365
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00019166602618429738,
+      "loss": 0.9388,
+      "step": 2366
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0001916590500528642,
+      "loss": 1.0873,
+      "step": 2367
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00019165207112995704,
+      "loss": 0.8597,
+      "step": 2368
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00019164508941578847,
+      "loss": 0.9765,
+      "step": 2369
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0001916381049105711,
+      "loss": 0.9995,
+      "step": 2370
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00019163111761451768,
+      "loss": 1.0162,
+      "step": 2371
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00019162412752784093,
+      "loss": 0.9444,
+      "step": 2372
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00019161713465075378,
+      "loss": 0.9427,
+      "step": 2373
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00019161013898346925,
+      "loss": 0.8471,
+      "step": 2374
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0001916031405262003,
+      "loss": 0.8535,
+      "step": 2375
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0001915961392791601,
+      "loss": 1.0368,
+      "step": 2376
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00019158913524256193,
+      "loss": 0.9767,
+      "step": 2377
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.000191582128416619,
+      "loss": 1.1243,
+      "step": 2378
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00019157511880154478,
+      "loss": 0.8693,
+      "step": 2379
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00019156810639755275,
+      "loss": 0.9954,
+      "step": 2380
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0001915610912048564,
+      "loss": 1.0391,
+      "step": 2381
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0001915540732236694,
+      "loss": 1.0324,
+      "step": 2382
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00019154705245420556,
+      "loss": 1.063,
+      "step": 2383
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0001915400288966786,
+      "loss": 1.0576,
+      "step": 2384
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00019153300255130246,
+      "loss": 0.8391,
+      "step": 2385
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00019152597341829118,
+      "loss": 1.1134,
+      "step": 2386
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00019151894149785874,
+      "loss": 0.9382,
+      "step": 2387
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00019151190679021933,
+      "loss": 0.9598,
+      "step": 2388
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00019150486929558723,
+      "loss": 1.0382,
+      "step": 2389
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00019149782901417672,
+      "loss": 1.0586,
+      "step": 2390
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00019149078594620221,
+      "loss": 1.0192,
+      "step": 2391
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00019148374009187827,
+      "loss": 0.9347,
+      "step": 2392
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00019147669145141936,
+      "loss": 1.0639,
+      "step": 2393
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00019146964002504026,
+      "loss": 1.0083,
+      "step": 2394
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00019146258581295567,
+      "loss": 0.8173,
+      "step": 2395
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0001914555288153804,
+      "loss": 1.0788,
+      "step": 2396
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0001914484690325294,
+      "loss": 0.9956,
+      "step": 2397
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00019144140646461773,
+      "loss": 0.9278,
+      "step": 2398
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0001914343411118604,
+      "loss": 1.0002,
+      "step": 2399
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0001914272729744726,
+      "loss": 1.0009,
+      "step": 2400
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0001914202020526696,
+      "loss": 0.9471,
+      "step": 2401
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00019141312834666677,
+      "loss": 1.0379,
+      "step": 2402
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0001914060518566795,
+      "loss": 0.961,
+      "step": 2403
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0001913989725829233,
+      "loss": 0.9484,
+      "step": 2404
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0001913918905256138,
+      "loss": 1.075,
+      "step": 2405
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0001913848056849667,
+      "loss": 1.0313,
+      "step": 2406
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0001913777180611977,
+      "loss": 0.9957,
+      "step": 2407
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0001913706276545227,
+      "loss": 0.9555,
+      "step": 2408
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00019136353446515765,
+      "loss": 0.9427,
+      "step": 2409
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00019135643849331856,
+      "loss": 0.799,
+      "step": 2410
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00019134933973922154,
+      "loss": 1.0327,
+      "step": 2411
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00019134223820308277,
+      "loss": 0.9157,
+      "step": 2412
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0001913351338851185,
+      "loss": 1.0805,
+      "step": 2413
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00019132802678554516,
+      "loss": 0.9577,
+      "step": 2414
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00019132091690457915,
+      "loss": 1.065,
+      "step": 2415
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00019131380424243703,
+      "loss": 1.0058,
+      "step": 2416
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0001913066887993354,
+      "loss": 1.0564,
+      "step": 2417
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00019129957057549093,
+      "loss": 1.0272,
+      "step": 2418
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00019129244957112043,
+      "loss": 1.0461,
+      "step": 2419
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0001912853257864408,
+      "loss": 0.9603,
+      "step": 2420
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00019127819922166897,
+      "loss": 1.0085,
+      "step": 2421
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00019127106987702194,
+      "loss": 0.9983,
+      "step": 2422
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00019126393775271692,
+      "loss": 0.9365,
+      "step": 2423
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00019125680284897106,
+      "loss": 1.0009,
+      "step": 2424
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00019124966516600162,
+      "loss": 1.054,
+      "step": 2425
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00019124252470402607,
+      "loss": 1.0047,
+      "step": 2426
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00019123538146326182,
+      "loss": 0.9532,
+      "step": 2427
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00019122823544392642,
+      "loss": 1.0994,
+      "step": 2428
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00019122108664623748,
+      "loss": 1.0057,
+      "step": 2429
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00019121393507041278,
+      "loss": 1.0893,
+      "step": 2430
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00019120678071667008,
+      "loss": 0.8817,
+      "step": 2431
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00019119962358522723,
+      "loss": 0.8205,
+      "step": 2432
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00019119246367630224,
+      "loss": 0.9115,
+      "step": 2433
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00019118530099011317,
+      "loss": 0.9718,
+      "step": 2434
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00019117813552687818,
+      "loss": 0.9416,
+      "step": 2435
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00019117096728681542,
+      "loss": 1.031,
+      "step": 2436
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0001911637962701433,
+      "loss": 0.9389,
+      "step": 2437
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00019115662247708013,
+      "loss": 1.0206,
+      "step": 2438
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00019114944590784442,
+      "loss": 0.9119,
+      "step": 2439
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0001911422665626547,
+      "loss": 1.0173,
+      "step": 2440
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00019113508444172967,
+      "loss": 0.8972,
+      "step": 2441
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00019112789954528803,
+      "loss": 0.9298,
+      "step": 2442
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0001911207118735486,
+      "loss": 0.9852,
+      "step": 2443
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00019111352142673027,
+      "loss": 1.0158,
+      "step": 2444
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00019110632820505205,
+      "loss": 1.0953,
+      "step": 2445
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.000191099132208733,
+      "loss": 1.0777,
+      "step": 2446
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00019109193343799224,
+      "loss": 0.957,
+      "step": 2447
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0001910847318930491,
+      "loss": 1.0025,
+      "step": 2448
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0001910775275741228,
+      "loss": 1.0267,
+      "step": 2449
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00019107032048143276,
+      "loss": 1.102,
+      "step": 2450
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00019106311061519856,
+      "loss": 1.0434,
+      "step": 2451
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00019105589797563965,
+      "loss": 0.9669,
+      "step": 2452
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00019104868256297582,
+      "loss": 0.9829,
+      "step": 2453
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0001910414643774267,
+      "loss": 0.9927,
+      "step": 2454
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00019103424341921222,
+      "loss": 1.1225,
+      "step": 2455
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00019102701968855222,
+      "loss": 1.014,
+      "step": 2456
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0001910197931856667,
+      "loss": 0.9329,
+      "step": 2457
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00019101256391077583,
+      "loss": 0.9361,
+      "step": 2458
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00019100533186409968,
+      "loss": 0.8985,
+      "step": 2459
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00019099809704585852,
+      "loss": 0.898,
+      "step": 2460
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00019099085945627274,
+      "loss": 1.0093,
+      "step": 2461
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0001909836190955627,
+      "loss": 1.0061,
+      "step": 2462
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00019097637596394892,
+      "loss": 0.9952,
+      "step": 2463
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00019096913006165202,
+      "loss": 1.0558,
+      "step": 2464
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00019096188138889263,
+      "loss": 0.9673,
+      "step": 2465
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00019095462994589155,
+      "loss": 0.9914,
+      "step": 2466
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0001909473757328696,
+      "loss": 1.001,
+      "step": 2467
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0001909401187500477,
+      "loss": 1.072,
+      "step": 2468
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00019093285899764685,
+      "loss": 1.0592,
+      "step": 2469
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0001909255964758882,
+      "loss": 0.9439,
+      "step": 2470
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00019091833118499289,
+      "loss": 0.9639,
+      "step": 2471
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00019091106312518218,
+      "loss": 0.9976,
+      "step": 2472
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00019090379229667743,
+      "loss": 0.9765,
+      "step": 2473
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00019089651869970008,
+      "loss": 0.8628,
+      "step": 2474
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00019088924233447163,
+      "loss": 0.9427,
+      "step": 2475
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0001908819632012137,
+      "loss": 0.9681,
+      "step": 2476
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00019087468130014797,
+      "loss": 0.9741,
+      "step": 2477
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00019086739663149617,
+      "loss": 0.9247,
+      "step": 2478
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00019086010919548024,
+      "loss": 1.0419,
+      "step": 2479
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00019085281899232205,
+      "loss": 1.0188,
+      "step": 2480
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00019084552602224362,
+      "loss": 1.0179,
+      "step": 2481
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0001908382302854671,
+      "loss": 1.0476,
+      "step": 2482
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00019083093178221466,
+      "loss": 0.9964,
+      "step": 2483
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00019082363051270857,
+      "loss": 1.0127,
+      "step": 2484
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0001908163264771712,
+      "loss": 1.0375,
+      "step": 2485
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.000190809019675825,
+      "loss": 1.022,
+      "step": 2486
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00019080171010889247,
+      "loss": 0.9985,
+      "step": 2487
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00019079439777659624,
+      "loss": 0.9163,
+      "step": 2488
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00019078708267915903,
+      "loss": 1.0189,
+      "step": 2489
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0001907797648168036,
+      "loss": 0.886,
+      "step": 2490
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0001907724441897528,
+      "loss": 0.9139,
+      "step": 2491
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00019076512079822962,
+      "loss": 0.9795,
+      "step": 2492
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00019075779464245704,
+      "loss": 1.0582,
+      "step": 2493
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00019075046572265822,
+      "loss": 1.0496,
+      "step": 2494
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00019074313403905633,
+      "loss": 0.9422,
+      "step": 2495
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00019073579959187468,
+      "loss": 1.0076,
+      "step": 2496
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00019072846238133665,
+      "loss": 0.9756,
+      "step": 2497
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0001907211224076657,
+      "loss": 0.9231,
+      "step": 2498
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0001907137796710853,
+      "loss": 0.9842,
+      "step": 2499
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00019070643417181914,
+      "loss": 1.0645,
+      "step": 2500
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0001906990859100909,
+      "loss": 0.9065,
+      "step": 2501
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00019069173488612441,
+      "loss": 1.0204,
+      "step": 2502
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0001906843811001435,
+      "loss": 0.9289,
+      "step": 2503
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00019067702455237214,
+      "loss": 0.9034,
+      "step": 2504
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00019066966524303438,
+      "loss": 1.0818,
+      "step": 2505
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00019066230317235433,
+      "loss": 1.0259,
+      "step": 2506
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00019065493834055622,
+      "loss": 1.0072,
+      "step": 2507
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00019064757074786437,
+      "loss": 1.0875,
+      "step": 2508
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0001906402003945031,
+      "loss": 0.9242,
+      "step": 2509
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00019063282728069692,
+      "loss": 0.9116,
+      "step": 2510
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00019062545140667036,
+      "loss": 1.1267,
+      "step": 2511
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00019061807277264805,
+      "loss": 0.9767,
+      "step": 2512
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0001906106913788547,
+      "loss": 0.9969,
+      "step": 2513
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00019060330722551515,
+      "loss": 1.0102,
+      "step": 2514
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00019059592031285427,
+      "loss": 0.9217,
+      "step": 2515
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00019058853064109695,
+      "loss": 0.9869,
+      "step": 2516
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0001905811382104684,
+      "loss": 0.943,
+      "step": 2517
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00019057374302119357,
+      "loss": 1.1038,
+      "step": 2518
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00019056634507349783,
+      "loss": 1.01,
+      "step": 2519
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00019055894436760643,
+      "loss": 1.0496,
+      "step": 2520
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00019055154090374474,
+      "loss": 0.8667,
+      "step": 2521
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00019054413468213826,
+      "loss": 0.8922,
+      "step": 2522
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0001905367257030125,
+      "loss": 0.9668,
+      "step": 2523
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00019052931396659317,
+      "loss": 0.9635,
+      "step": 2524
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00019052189947310594,
+      "loss": 0.9059,
+      "step": 2525
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00019051448222277663,
+      "loss": 1.0271,
+      "step": 2526
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00019050706221583117,
+      "loss": 1.0366,
+      "step": 2527
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00019049963945249546,
+      "loss": 0.9152,
+      "step": 2528
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00019049221393299565,
+      "loss": 1.0026,
+      "step": 2529
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00019048478565755782,
+      "loss": 1.019,
+      "step": 2530
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00019047735462640825,
+      "loss": 0.9784,
+      "step": 2531
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00019046992083977316,
+      "loss": 0.8982,
+      "step": 2532
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00019046248429787906,
+      "loss": 1.0533,
+      "step": 2533
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00019045504500095234,
+      "loss": 0.9733,
+      "step": 2534
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0001904476029492196,
+      "loss": 0.8271,
+      "step": 2535
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00019044015814290751,
+      "loss": 0.9662,
+      "step": 2536
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00019043271058224277,
+      "loss": 1.1361,
+      "step": 2537
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0001904252602674522,
+      "loss": 0.8412,
+      "step": 2538
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0001904178071987627,
+      "loss": 0.8089,
+      "step": 2539
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00019041035137640127,
+      "loss": 0.9908,
+      "step": 2540
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00019040289280059493,
+      "loss": 0.9844,
+      "step": 2541
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0001903954314715709,
+      "loss": 0.9681,
+      "step": 2542
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00019038796738955632,
+      "loss": 0.9267,
+      "step": 2543
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00019038050055477862,
+      "loss": 1.0669,
+      "step": 2544
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00019037303096746514,
+      "loss": 1.0614,
+      "step": 2545
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00019036555862784335,
+      "loss": 0.9289,
+      "step": 2546
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00019035808353614085,
+      "loss": 1.0589,
+      "step": 2547
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00019035060569258533,
+      "loss": 0.8234,
+      "step": 2548
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00019034312509740443,
+      "loss": 0.8866,
+      "step": 2549
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00019033564175082606,
+      "loss": 0.9899,
+      "step": 2550
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00019032815565307807,
+      "loss": 0.8007,
+      "step": 2551
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00019032066680438846,
+      "loss": 0.9737,
+      "step": 2552
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00019031317520498532,
+      "loss": 0.9163,
+      "step": 2553
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00019030568085509684,
+      "loss": 0.8406,
+      "step": 2554
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00019029818375495118,
+      "loss": 0.9625,
+      "step": 2555
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00019029068390477672,
+      "loss": 1.0849,
+      "step": 2556
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00019028318130480184,
+      "loss": 1.0115,
+      "step": 2557
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00019027567595525507,
+      "loss": 0.9449,
+      "step": 2558
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00019026816785636493,
+      "loss": 0.9702,
+      "step": 2559
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0001902606570083601,
+      "loss": 0.9516,
+      "step": 2560
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00019025314341146934,
+      "loss": 0.9801,
+      "step": 2561
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00019024562706592148,
+      "loss": 0.811,
+      "step": 2562
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00019023810797194542,
+      "loss": 0.9662,
+      "step": 2563
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00019023058612977013,
+      "loss": 0.9409,
+      "step": 2564
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0001902230615396247,
+      "loss": 0.9479,
+      "step": 2565
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00019021553420173833,
+      "loss": 0.9965,
+      "step": 2566
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00019020800411634022,
+      "loss": 1.0099,
+      "step": 2567
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0001902004712836597,
+      "loss": 0.9577,
+      "step": 2568
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0001901929357039262,
+      "loss": 1.0607,
+      "step": 2569
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0001901853973773692,
+      "loss": 0.9744,
+      "step": 2570
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00019017785630421835,
+      "loss": 1.0394,
+      "step": 2571
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00019017031248470318,
+      "loss": 1.0152,
+      "step": 2572
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00019016276591905354,
+      "loss": 1.0175,
+      "step": 2573
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00019015521660749923,
+      "loss": 0.9512,
+      "step": 2574
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00019014766455027017,
+      "loss": 0.9977,
+      "step": 2575
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00019014010974759634,
+      "loss": 0.919,
+      "step": 2576
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00019013255219970784,
+      "loss": 0.9942,
+      "step": 2577
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0001901249919068348,
+      "loss": 1.0108,
+      "step": 2578
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00019011742886920754,
+      "loss": 0.9731,
+      "step": 2579
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00019010986308705632,
+      "loss": 0.8974,
+      "step": 2580
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0001901022945606116,
+      "loss": 0.9797,
+      "step": 2581
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00019009472329010386,
+      "loss": 1.0166,
+      "step": 2582
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0001900871492757637,
+      "loss": 1.0621,
+      "step": 2583
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00019007957251782176,
+      "loss": 0.9087,
+      "step": 2584
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0001900719930165088,
+      "loss": 0.9499,
+      "step": 2585
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00019006441077205568,
+      "loss": 1.0229,
+      "step": 2586
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00019005682578469325,
+      "loss": 1.0341,
+      "step": 2587
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00019004923805465257,
+      "loss": 1.0762,
+      "step": 2588
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00019004164758216474,
+      "loss": 0.9772,
+      "step": 2589
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00019003405436746087,
+      "loss": 0.9206,
+      "step": 2590
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00019002645841077225,
+      "loss": 1.0224,
+      "step": 2591
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0001900188597123302,
+      "loss": 0.8794,
+      "step": 2592
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0001900112582723661,
+      "loss": 0.9865,
+      "step": 2593
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00019000365409111157,
+      "loss": 0.9653,
+      "step": 2594
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00018999604716879803,
+      "loss": 1.0074,
+      "step": 2595
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00018998843750565728,
+      "loss": 0.9965,
+      "step": 2596
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00018998082510192105,
+      "loss": 1.0759,
+      "step": 2597
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00018997320995782114,
+      "loss": 1.0551,
+      "step": 2598
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00018996559207358943,
+      "loss": 0.9376,
+      "step": 2599
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00018995797144945805,
+      "loss": 1.0226,
+      "step": 2600
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00018995034808565894,
+      "loss": 0.9578,
+      "step": 2601
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0001899427219824244,
+      "loss": 1.0126,
+      "step": 2602
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0001899350931399866,
+      "loss": 0.9469,
+      "step": 2603
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0001899274615585779,
+      "loss": 0.9809,
+      "step": 2604
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00018991982723843068,
+      "loss": 1.0192,
+      "step": 2605
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00018991219017977756,
+      "loss": 1.0003,
+      "step": 2606
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.000189904550382851,
+      "loss": 0.866,
+      "step": 2607
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00018989690784788373,
+      "loss": 0.9788,
+      "step": 2608
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0001898892625751085,
+      "loss": 1.119,
+      "step": 2609
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00018988161456475809,
+      "loss": 0.9508,
+      "step": 2610
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0001898739638170655,
+      "loss": 0.9641,
+      "step": 2611
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00018986631033226373,
+      "loss": 1.034,
+      "step": 2612
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0001898586541105858,
+      "loss": 1.1043,
+      "step": 2613
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0001898509951522649,
+      "loss": 0.8689,
+      "step": 2614
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00018984333345753432,
+      "loss": 1.0814,
+      "step": 2615
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0001898356690266274,
+      "loss": 0.9343,
+      "step": 2616
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0001898280018597775,
+      "loss": 0.99,
+      "step": 2617
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0001898203319572182,
+      "loss": 0.979,
+      "step": 2618
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00018981265931918305,
+      "loss": 0.9675,
+      "step": 2619
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0001898049839459057,
+      "loss": 1.0068,
+      "step": 2620
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0001897973058376199,
+      "loss": 1.055,
+      "step": 2621
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00018978962499455952,
+      "loss": 0.9725,
+      "step": 2622
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0001897819414169585,
+      "loss": 0.9691,
+      "step": 2623
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00018977425510505075,
+      "loss": 0.993,
+      "step": 2624
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00018976656605907048,
+      "loss": 1.075,
+      "step": 2625
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00018975887427925177,
+      "loss": 1.0335,
+      "step": 2626
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0001897511797658289,
+      "loss": 0.9429,
+      "step": 2627
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0001897434825190362,
+      "loss": 0.9312,
+      "step": 2628
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0001897357825391081,
+      "loss": 0.9613,
+      "step": 2629
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0001897280798262791,
+      "loss": 0.9233,
+      "step": 2630
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00018972037438078377,
+      "loss": 1.0922,
+      "step": 2631
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00018971266620285683,
+      "loss": 1.0399,
+      "step": 2632
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00018970495529273292,
+      "loss": 1.0168,
+      "step": 2633
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.000189697241650647,
+      "loss": 0.8943,
+      "step": 2634
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00018968952527683396,
+      "loss": 0.9167,
+      "step": 2635
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0001896818061715288,
+      "loss": 0.9567,
+      "step": 2636
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00018967408433496653,
+      "loss": 1.0159,
+      "step": 2637
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00018966635976738237,
+      "loss": 1.0702,
+      "step": 2638
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00018965863246901162,
+      "loss": 0.8841,
+      "step": 2639
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00018965090244008956,
+      "loss": 1.108,
+      "step": 2640
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00018964316968085162,
+      "loss": 1.0654,
+      "step": 2641
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00018963543419153327,
+      "loss": 0.7772,
+      "step": 2642
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00018962769597237014,
+      "loss": 1.0071,
+      "step": 2643
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00018961995502359788,
+      "loss": 0.9285,
+      "step": 2644
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00018961221134545226,
+      "loss": 0.8565,
+      "step": 2645
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0001896044649381691,
+      "loss": 0.9842,
+      "step": 2646
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00018959671580198427,
+      "loss": 0.9986,
+      "step": 2647
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00018958896393713386,
+      "loss": 0.9367,
+      "step": 2648
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00018958120934385388,
+      "loss": 0.8552,
+      "step": 2649
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00018957345202238053,
+      "loss": 0.9628,
+      "step": 2650
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00018956569197295004,
+      "loss": 0.9595,
+      "step": 2651
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00018955792919579876,
+      "loss": 1.0462,
+      "step": 2652
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0001895501636911631,
+      "loss": 0.8916,
+      "step": 2653
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00018954239545927954,
+      "loss": 0.9747,
+      "step": 2654
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0001895346245003847,
+      "loss": 1.1022,
+      "step": 2655
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0001895268508147152,
+      "loss": 0.8942,
+      "step": 2656
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0001895190744025078,
+      "loss": 0.9098,
+      "step": 2657
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00018951129526399935,
+      "loss": 0.9911,
+      "step": 2658
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00018950351339942678,
+      "loss": 1.1645,
+      "step": 2659
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00018949572880902702,
+      "loss": 0.9208,
+      "step": 2660
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0001894879414930372,
+      "loss": 0.8985,
+      "step": 2661
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00018948015145169446,
+      "loss": 0.9542,
+      "step": 2662
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00018947235868523607,
+      "loss": 0.9815,
+      "step": 2663
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00018946456319389934,
+      "loss": 0.9694,
+      "step": 2664
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0001894567649779217,
+      "loss": 1.0149,
+      "step": 2665
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00018944896403754063,
+      "loss": 1.0543,
+      "step": 2666
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00018944116037299368,
+      "loss": 1.0116,
+      "step": 2667
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00018943335398451856,
+      "loss": 1.0393,
+      "step": 2668
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00018942554487235298,
+      "loss": 0.8918,
+      "step": 2669
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00018941773303673477,
+      "loss": 1.0242,
+      "step": 2670
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0001894099184779019,
+      "loss": 1.0228,
+      "step": 2671
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00018940210119609225,
+      "loss": 0.9935,
+      "step": 2672
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.000189394281191544,
+      "loss": 1.0107,
+      "step": 2673
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00018938645846449523,
+      "loss": 0.9237,
+      "step": 2674
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00018937863301518422,
+      "loss": 1.011,
+      "step": 2675
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0001893708048438493,
+      "loss": 1.0117,
+      "step": 2676
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00018936297395072884,
+      "loss": 0.9418,
+      "step": 2677
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00018935514033606137,
+      "loss": 0.9509,
+      "step": 2678
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00018934730400008544,
+      "loss": 0.9946,
+      "step": 2679
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0001893394649430397,
+      "loss": 0.8825,
+      "step": 2680
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00018933162316516297,
+      "loss": 1.0351,
+      "step": 2681
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00018932377866669393,
+      "loss": 0.9701,
+      "step": 2682
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00018931593144787154,
+      "loss": 0.8896,
+      "step": 2683
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00018930808150893484,
+      "loss": 0.9099,
+      "step": 2684
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00018930022885012287,
+      "loss": 1.0401,
+      "step": 2685
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00018929237347167476,
+      "loss": 0.9797,
+      "step": 2686
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00018928451537382975,
+      "loss": 1.016,
+      "step": 2687
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0001892766545568272,
+      "loss": 0.9317,
+      "step": 2688
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00018926879102090644,
+      "loss": 1.0037,
+      "step": 2689
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.000189260924766307,
+      "loss": 1.0048,
+      "step": 2690
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00018925305579326843,
+      "loss": 0.9099,
+      "step": 2691
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00018924518410203044,
+      "loss": 1.0058,
+      "step": 2692
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00018923730969283266,
+      "loss": 1.0469,
+      "step": 2693
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00018922943256591496,
+      "loss": 0.9725,
+      "step": 2694
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00018922155272151726,
+      "loss": 0.8171,
+      "step": 2695
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0001892136701598795,
+      "loss": 0.9545,
+      "step": 2696
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00018920578488124178,
+      "loss": 1.0656,
+      "step": 2697
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00018919789688584418,
+      "loss": 0.9875,
+      "step": 2698
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00018919000617392698,
+      "loss": 1.1312,
+      "step": 2699
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0001891821127457305,
+      "loss": 0.9652,
+      "step": 2700
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0001891742166014951,
+      "loss": 0.9664,
+      "step": 2701
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0001891663177414613,
+      "loss": 1.0002,
+      "step": 2702
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00018915841616586962,
+      "loss": 1.0625,
+      "step": 2703
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0001891505118749607,
+      "loss": 0.8867,
+      "step": 2704
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0001891426048689753,
+      "loss": 0.9655,
+      "step": 2705
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00018913469514815422,
+      "loss": 0.9609,
+      "step": 2706
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0001891267827127383,
+      "loss": 0.8887,
+      "step": 2707
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00018911886756296857,
+      "loss": 1.0813,
+      "step": 2708
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00018911094969908607,
+      "loss": 1.0378,
+      "step": 2709
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00018910302912133196,
+      "loss": 0.9596,
+      "step": 2710
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00018909510582994742,
+      "loss": 1.0556,
+      "step": 2711
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00018908717982517374,
+      "loss": 1.112,
+      "step": 2712
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00018907925110725238,
+      "loss": 0.8725,
+      "step": 2713
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00018907131967642473,
+      "loss": 1.021,
+      "step": 2714
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0001890633855329324,
+      "loss": 0.8008,
+      "step": 2715
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.000189055448677017,
+      "loss": 0.9296,
+      "step": 2716
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00018904750910892022,
+      "loss": 1.0362,
+      "step": 2717
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00018903956682888394,
+      "loss": 0.8448,
+      "step": 2718
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00018903162183714993,
+      "loss": 1.0534,
+      "step": 2719
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00018902367413396024,
+      "loss": 0.9572,
+      "step": 2720
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0001890157237195569,
+      "loss": 0.9016,
+      "step": 2721
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.000189007770594182,
+      "loss": 0.8768,
+      "step": 2722
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00018899981475807782,
+      "loss": 0.9924,
+      "step": 2723
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00018899185621148662,
+      "loss": 0.9555,
+      "step": 2724
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00018898389495465074,
+      "loss": 0.9988,
+      "step": 2725
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0001889759309878127,
+      "loss": 0.8864,
+      "step": 2726
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.000188967964311215,
+      "loss": 0.9694,
+      "step": 2727
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0001889599949251003,
+      "loss": 1.0984,
+      "step": 2728
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00018895202282971128,
+      "loss": 0.9291,
+      "step": 2729
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00018894404802529072,
+      "loss": 0.9392,
+      "step": 2730
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00018893607051208156,
+      "loss": 0.9264,
+      "step": 2731
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00018892809029032667,
+      "loss": 1.1114,
+      "step": 2732
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00018892010736026914,
+      "loss": 0.986,
+      "step": 2733
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00018891212172215206,
+      "loss": 1.006,
+      "step": 2734
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00018890413337621867,
+      "loss": 0.9929,
+      "step": 2735
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0001888961423227122,
+      "loss": 1.0351,
+      "step": 2736
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00018888814856187606,
+      "loss": 0.983,
+      "step": 2737
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00018888015209395372,
+      "loss": 1.0114,
+      "step": 2738
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00018887215291918863,
+      "loss": 1.026,
+      "step": 2739
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00018886415103782449,
+      "loss": 1.0171,
+      "step": 2740
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00018885614645010494,
+      "loss": 1.1592,
+      "step": 2741
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0001888481391562738,
+      "loss": 0.9578,
+      "step": 2742
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0001888401291565749,
+      "loss": 0.9917,
+      "step": 2743
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00018883211645125223,
+      "loss": 0.8561,
+      "step": 2744
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00018882410104054974,
+      "loss": 1.0113,
+      "step": 2745
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0001888160829247116,
+      "loss": 0.9387,
+      "step": 2746
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.000188808062103982,
+      "loss": 0.9551,
+      "step": 2747
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0001888000385786052,
+      "loss": 0.9669,
+      "step": 2748
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00018879201234882557,
+      "loss": 0.9023,
+      "step": 2749
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0001887839834148875,
+      "loss": 0.9232,
+      "step": 2750
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00018877595177703554,
+      "loss": 0.908,
+      "step": 2751
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0001887679174355143,
+      "loss": 1.0193,
+      "step": 2752
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00018875988039056848,
+      "loss": 0.8206,
+      "step": 2753
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00018875184064244283,
+      "loss": 1.2381,
+      "step": 2754
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00018874379819138222,
+      "loss": 0.8815,
+      "step": 2755
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00018873575303763156,
+      "loss": 0.9627,
+      "step": 2756
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00018872770518143584,
+      "loss": 1.0334,
+      "step": 2757
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00018871965462304022,
+      "loss": 1.0136,
+      "step": 2758
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0001887116013626898,
+      "loss": 0.9631,
+      "step": 2759
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00018870354540062996,
+      "loss": 1.0226,
+      "step": 2760
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00018869548673710592,
+      "loss": 0.9357,
+      "step": 2761
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0001886874253723632,
+      "loss": 0.9072,
+      "step": 2762
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00018867936130664723,
+      "loss": 1.0399,
+      "step": 2763
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00018867129454020365,
+      "loss": 1.0671,
+      "step": 2764
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00018866322507327815,
+      "loss": 1.11,
+      "step": 2765
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00018865515290611643,
+      "loss": 0.9469,
+      "step": 2766
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00018864707803896433,
+      "loss": 0.9817,
+      "step": 2767
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00018863900047206783,
+      "loss": 1.0617,
+      "step": 2768
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0001886309202056729,
+      "loss": 1.0579,
+      "step": 2769
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0001886228372400256,
+      "loss": 0.9277,
+      "step": 2770
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00018861475157537215,
+      "loss": 1.1224,
+      "step": 2771
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00018860666321195873,
+      "loss": 1.0267,
+      "step": 2772
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00018859857215003172,
+      "loss": 0.9033,
+      "step": 2773
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00018859047838983752,
+      "loss": 0.8849,
+      "step": 2774
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00018858238193162263,
+      "loss": 0.8589,
+      "step": 2775
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00018857428277563364,
+      "loss": 1.0767,
+      "step": 2776
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00018856618092211714,
+      "loss": 1.0424,
+      "step": 2777
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00018855807637132,
+      "loss": 1.0249,
+      "step": 2778
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0001885499691234889,
+      "loss": 0.9475,
+      "step": 2779
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00018854185917887084,
+      "loss": 1.0818,
+      "step": 2780
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00018853374653771278,
+      "loss": 1.0002,
+      "step": 2781
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0001885256312002618,
+      "loss": 0.986,
+      "step": 2782
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00018851751316676505,
+      "loss": 1.0356,
+      "step": 2783
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00018850939243746976,
+      "loss": 1.1801,
+      "step": 2784
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00018850126901262323,
+      "loss": 1.0511,
+      "step": 2785
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00018849314289247292,
+      "loss": 1.0368,
+      "step": 2786
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00018848501407726622,
+      "loss": 1.0153,
+      "step": 2787
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00018847688256725073,
+      "loss": 0.8257,
+      "step": 2788
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00018846874836267414,
+      "loss": 0.9443,
+      "step": 2789
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00018846061146378411,
+      "loss": 0.8177,
+      "step": 2790
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0001884524718708285,
+      "loss": 0.9639,
+      "step": 2791
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00018844432958405518,
+      "loss": 0.9095,
+      "step": 2792
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00018843618460371211,
+      "loss": 0.9628,
+      "step": 2793
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00018842803693004735,
+      "loss": 0.986,
+      "step": 2794
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00018841988656330907,
+      "loss": 0.9891,
+      "step": 2795
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00018841173350374546,
+      "loss": 0.9831,
+      "step": 2796
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00018840357775160485,
+      "loss": 0.9591,
+      "step": 2797
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00018839541930713556,
+      "loss": 0.8924,
+      "step": 2798
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0001883872581705861,
+      "loss": 1.0116,
+      "step": 2799
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00018837909434220502,
+      "loss": 0.9285,
+      "step": 2800
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00018837092782224095,
+      "loss": 0.9329,
+      "step": 2801
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0001883627586109426,
+      "loss": 0.7499,
+      "step": 2802
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00018835458670855872,
+      "loss": 0.9982,
+      "step": 2803
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00018834641211533826,
+      "loss": 0.9776,
+      "step": 2804
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00018833823483153011,
+      "loss": 1.0871,
+      "step": 2805
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00018833005485738335,
+      "loss": 0.9005,
+      "step": 2806
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0001883218721931471,
+      "loss": 0.962,
+      "step": 2807
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00018831368683907052,
+      "loss": 0.9526,
+      "step": 2808
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00018830549879540292,
+      "loss": 0.883,
+      "step": 2809
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0001882973080623937,
+      "loss": 0.975,
+      "step": 2810
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00018828911464029228,
+      "loss": 1.0392,
+      "step": 2811
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00018828091852934818,
+      "loss": 0.9814,
+      "step": 2812
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00018827271972981103,
+      "loss": 1.0303,
+      "step": 2813
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00018826451824193051,
+      "loss": 1.0212,
+      "step": 2814
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00018825631406595644,
+      "loss": 0.9908,
+      "step": 2815
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00018824810720213862,
+      "loss": 0.9537,
+      "step": 2816
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00018823989765072698,
+      "loss": 0.9336,
+      "step": 2817
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00018823168541197159,
+      "loss": 0.9335,
+      "step": 2818
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00018822347048612253,
+      "loss": 0.9722,
+      "step": 2819
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00018821525287343,
+      "loss": 1.1006,
+      "step": 2820
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0001882070325741443,
+      "loss": 0.8459,
+      "step": 2821
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00018819880958851568,
+      "loss": 1.032,
+      "step": 2822
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00018819058391679466,
+      "loss": 1.05,
+      "step": 2823
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0001881823555592317,
+      "loss": 1.0307,
+      "step": 2824
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00018817412451607743,
+      "loss": 0.8432,
+      "step": 2825
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00018816589078758252,
+      "loss": 0.8641,
+      "step": 2826
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00018815765437399772,
+      "loss": 1.1001,
+      "step": 2827
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00018814941527557383,
+      "loss": 0.9957,
+      "step": 2828
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00018814117349256183,
+      "loss": 1.1903,
+      "step": 2829
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00018813292902521272,
+      "loss": 1.0366,
+      "step": 2830
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0001881246818737776,
+      "loss": 0.9229,
+      "step": 2831
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00018811643203850756,
+      "loss": 0.9674,
+      "step": 2832
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0001881081795196539,
+      "loss": 1.0241,
+      "step": 2833
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00018809992431746795,
+      "loss": 1.066,
+      "step": 2834
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0001880916664322011,
+      "loss": 0.9577,
+      "step": 2835
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00018808340586410488,
+      "loss": 0.9285,
+      "step": 2836
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00018807514261343083,
+      "loss": 0.9619,
+      "step": 2837
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00018806687668043067,
+      "loss": 1.0512,
+      "step": 2838
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00018805860806535605,
+      "loss": 0.9926,
+      "step": 2839
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00018805033676845884,
+      "loss": 0.9326,
+      "step": 2840
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00018804206278999095,
+      "loss": 0.9118,
+      "step": 2841
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00018803378613020436,
+      "loss": 1.069,
+      "step": 2842
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0001880255067893511,
+      "loss": 0.8093,
+      "step": 2843
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00018801722476768332,
+      "loss": 0.9856,
+      "step": 2844
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0001880089400654533,
+      "loss": 1.1013,
+      "step": 2845
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00018800065268291332,
+      "loss": 0.8929,
+      "step": 2846
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00018799236262031577,
+      "loss": 1.1038,
+      "step": 2847
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00018798406987791312,
+      "loss": 0.8575,
+      "step": 2848
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00018797577445595792,
+      "loss": 0.8608,
+      "step": 2849
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00018796747635470282,
+      "loss": 0.8414,
+      "step": 2850
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00018795917557440055,
+      "loss": 1.174,
+      "step": 2851
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0001879508721153039,
+      "loss": 1.0122,
+      "step": 2852
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00018794256597766572,
+      "loss": 0.9329,
+      "step": 2853
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.000187934257161739,
+      "loss": 1.0227,
+      "step": 2854
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00018792594566777683,
+      "loss": 0.9625,
+      "step": 2855
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0001879176314960323,
+      "loss": 0.9195,
+      "step": 2856
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00018790931464675856,
+      "loss": 0.9383,
+      "step": 2857
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.000187900995120209,
+      "loss": 0.8838,
+      "step": 2858
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0001878926729166369,
+      "loss": 0.9828,
+      "step": 2859
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00018788434803629575,
+      "loss": 0.9581,
+      "step": 2860
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0001878760204794391,
+      "loss": 0.9438,
+      "step": 2861
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00018786769024632056,
+      "loss": 0.9805,
+      "step": 2862
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0001878593573371938,
+      "loss": 0.9647,
+      "step": 2863
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00018785102175231263,
+      "loss": 0.9442,
+      "step": 2864
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0001878426834919309,
+      "loss": 0.9177,
+      "step": 2865
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00018783434255630256,
+      "loss": 1.0569,
+      "step": 2866
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00018782599894568162,
+      "loss": 1.1094,
+      "step": 2867
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0001878176526603222,
+      "loss": 0.9511,
+      "step": 2868
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00018780930370047844,
+      "loss": 0.961,
+      "step": 2869
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00018780095206640467,
+      "loss": 0.9882,
+      "step": 2870
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00018779259775835518,
+      "loss": 0.8547,
+      "step": 2871
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00018778424077658446,
+      "loss": 1.0485,
+      "step": 2872
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00018777588112134694,
+      "loss": 0.9144,
+      "step": 2873
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00018776751879289732,
+      "loss": 0.9614,
+      "step": 2874
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0001877591537914902,
+      "loss": 1.1598,
+      "step": 2875
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00018775078611738035,
+      "loss": 0.988,
+      "step": 2876
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0001877424157708226,
+      "loss": 1.0419,
+      "step": 2877
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0001877340427520719,
+      "loss": 1.0072,
+      "step": 2878
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0001877256670613832,
+      "loss": 0.8651,
+      "step": 2879
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00018771728869901167,
+      "loss": 1.0449,
+      "step": 2880
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00018770890766521236,
+      "loss": 0.9097,
+      "step": 2881
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0001877005239602406,
+      "loss": 0.978,
+      "step": 2882
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0001876921375843517,
+      "loss": 0.9564,
+      "step": 2883
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.000187683748537801,
+      "loss": 0.9835,
+      "step": 2884
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00018767535682084408,
+      "loss": 0.98,
+      "step": 2885
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00018766696243373648,
+      "loss": 0.9613,
+      "step": 2886
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00018765856537673382,
+      "loss": 0.8857,
+      "step": 2887
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00018765016565009182,
+      "loss": 1.0746,
+      "step": 2888
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00018764176325406634,
+      "loss": 1.0777,
+      "step": 2889
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00018763335818891327,
+      "loss": 0.9293,
+      "step": 2890
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0001876249504548886,
+      "loss": 1.0939,
+      "step": 2891
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00018761654005224835,
+      "loss": 1.0166,
+      "step": 2892
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00018760812698124867,
+      "loss": 1.032,
+      "step": 2893
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0001875997112421458,
+      "loss": 1.0241,
+      "step": 2894
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00018759129283519598,
+      "loss": 1.0202,
+      "step": 2895
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00018758287176065565,
+      "loss": 0.9149,
+      "step": 2896
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00018757444801878128,
+      "loss": 0.9707,
+      "step": 2897
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0001875660216098294,
+      "loss": 1.0363,
+      "step": 2898
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0001875575925340566,
+      "loss": 1.0094,
+      "step": 2899
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00018754916079171964,
+      "loss": 1.0242,
+      "step": 2900
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00018754072638307525,
+      "loss": 1.0827,
+      "step": 2901
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00018753228930838036,
+      "loss": 1.0685,
+      "step": 2902
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0001875238495678919,
+      "loss": 0.9755,
+      "step": 2903
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0001875154071618669,
+      "loss": 1.0303,
+      "step": 2904
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00018750696209056245,
+      "loss": 1.0027,
+      "step": 2905
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00018749851435423582,
+      "loss": 0.9263,
+      "step": 2906
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0001874900639531442,
+      "loss": 0.8596,
+      "step": 2907
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00018748161088754498,
+      "loss": 1.0702,
+      "step": 2908
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00018747315515769557,
+      "loss": 0.8703,
+      "step": 2909
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00018746469676385354,
+      "loss": 0.9811,
+      "step": 2910
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00018745623570627645,
+      "loss": 0.9062,
+      "step": 2911
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00018744777198522204,
+      "loss": 1.0529,
+      "step": 2912
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00018743930560094797,
+      "loss": 0.9912,
+      "step": 2913
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0001874308365537122,
+      "loss": 0.9394,
+      "step": 2914
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00018742236484377257,
+      "loss": 1.029,
+      "step": 2915
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0001874138904713871,
+      "loss": 0.9632,
+      "step": 2916
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00018740541343681392,
+      "loss": 1.052,
+      "step": 2917
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00018739693374031114,
+      "loss": 0.9536,
+      "step": 2918
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00018738845138213702,
+      "loss": 0.8343,
+      "step": 2919
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00018737996636254995,
+      "loss": 0.8484,
+      "step": 2920
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00018737147868180828,
+      "loss": 0.8796,
+      "step": 2921
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00018736298834017053,
+      "loss": 0.9295,
+      "step": 2922
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00018735449533789528,
+      "loss": 1.0938,
+      "step": 2923
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0001873459996752411,
+      "loss": 0.9673,
+      "step": 2924
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00018733750135246682,
+      "loss": 0.9206,
+      "step": 2925
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00018732900036983128,
+      "loss": 0.8784,
+      "step": 2926
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00018732049672759326,
+      "loss": 0.9461,
+      "step": 2927
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00018731199042601183,
+      "loss": 0.9875,
+      "step": 2928
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00018730348146534605,
+      "loss": 1.0181,
+      "step": 2929
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.000187294969845855,
+      "loss": 0.9425,
+      "step": 2930
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00018728645556779795,
+      "loss": 1.0379,
+      "step": 2931
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0001872779386314342,
+      "loss": 1.1348,
+      "step": 2932
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0001872694190370231,
+      "loss": 1.1125,
+      "step": 2933
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00018726089678482413,
+      "loss": 0.8314,
+      "step": 2934
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00018725237187509683,
+      "loss": 1.1156,
+      "step": 2935
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00018724384430810083,
+      "loss": 0.9777,
+      "step": 2936
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0001872353140840959,
+      "loss": 0.9661,
+      "step": 2937
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0001872267812033417,
+      "loss": 1.0461,
+      "step": 2938
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0001872182456660982,
+      "loss": 1.0821,
+      "step": 2939
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00018720970747262533,
+      "loss": 0.971,
+      "step": 2940
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00018720116662318306,
+      "loss": 1.0001,
+      "step": 2941
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0001871926231180316,
+      "loss": 1.0186,
+      "step": 2942
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00018718407695743108,
+      "loss": 1.0104,
+      "step": 2943
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00018717552814164176,
+      "loss": 0.983,
+      "step": 2944
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00018716697667092404,
+      "loss": 0.9716,
+      "step": 2945
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0001871584225455383,
+      "loss": 1.0032,
+      "step": 2946
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00018714986576574515,
+      "loss": 1.0893,
+      "step": 2947
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0001871413063318051,
+      "loss": 0.9114,
+      "step": 2948
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00018713274424397885,
+      "loss": 1.1302,
+      "step": 2949
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00018712417950252715,
+      "loss": 1.0874,
+      "step": 2950
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00018711561210771086,
+      "loss": 1.0258,
+      "step": 2951
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0001871070420597909,
+      "loss": 0.9079,
+      "step": 2952
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00018709846935902822,
+      "loss": 1.0061,
+      "step": 2953
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.000187089894005684,
+      "loss": 1.0028,
+      "step": 2954
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0001870813160000193,
+      "loss": 0.9332,
+      "step": 2955
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00018707273534229542,
+      "loss": 0.9611,
+      "step": 2956
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00018706415203277366,
+      "loss": 0.9754,
+      "step": 2957
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00018705556607171543,
+      "loss": 0.9033,
+      "step": 2958
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00018704697745938224,
+      "loss": 1.0006,
+      "step": 2959
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00018703838619603565,
+      "loss": 0.9979,
+      "step": 2960
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00018702979228193726,
+      "loss": 1.0145,
+      "step": 2961
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00018702119571734887,
+      "loss": 0.9253,
+      "step": 2962
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00018701259650253223,
+      "loss": 0.9937,
+      "step": 2963
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00018700399463774924,
+      "loss": 0.9338,
+      "step": 2964
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00018699539012326187,
+      "loss": 0.8953,
+      "step": 2965
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00018698678295933218,
+      "loss": 1.0217,
+      "step": 2966
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0001869781731462223,
+      "loss": 1.0427,
+      "step": 2967
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00018696956068419446,
+      "loss": 0.9629,
+      "step": 2968
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0001869609455735109,
+      "loss": 0.9244,
+      "step": 2969
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.000186952327814434,
+      "loss": 0.9063,
+      "step": 2970
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00018694370740722632,
+      "loss": 1.0583,
+      "step": 2971
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00018693508435215026,
+      "loss": 0.978,
+      "step": 2972
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0001869264586494685,
+      "loss": 1.0215,
+      "step": 2973
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00018691783029944371,
+      "loss": 0.9848,
+      "step": 2974
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0001869091993023387,
+      "loss": 0.8606,
+      "step": 2975
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00018690056565841628,
+      "loss": 0.927,
+      "step": 2976
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00018689192936793947,
+      "loss": 0.9188,
+      "step": 2977
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00018688329043117116,
+      "loss": 1.0169,
+      "step": 2978
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00018687464884837457,
+      "loss": 0.9943,
+      "step": 2979
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0001868660046198128,
+      "loss": 0.924,
+      "step": 2980
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00018685735774574917,
+      "loss": 1.0427,
+      "step": 2981
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00018684870822644695,
+      "loss": 1.1846,
+      "step": 2982
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00018684005606216963,
+      "loss": 0.9554,
+      "step": 2983
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00018683140125318068,
+      "loss": 1.0807,
+      "step": 2984
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00018682274379974367,
+      "loss": 0.9654,
+      "step": 2985
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0001868140837021223,
+      "loss": 1.0683,
+      "step": 2986
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00018680542096058023,
+      "loss": 0.976,
+      "step": 2987
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00018679675557538143,
+      "loss": 0.8819,
+      "step": 2988
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00018678808754678966,
+      "loss": 0.9958,
+      "step": 2989
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00018677941687506898,
+      "loss": 0.9417,
+      "step": 2990
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00018677074356048342,
+      "loss": 0.8873,
+      "step": 2991
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00018676206760329716,
+      "loss": 0.9636,
+      "step": 2992
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0001867533890037744,
+      "loss": 1.0728,
+      "step": 2993
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00018674470776217948,
+      "loss": 1.0625,
+      "step": 2994
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0001867360238787767,
+      "loss": 0.9043,
+      "step": 2995
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00018672733735383066,
+      "loss": 0.9007,
+      "step": 2996
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0001867186481876058,
+      "loss": 1.0385,
+      "step": 2997
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0001867099563803668,
+      "loss": 1.0614,
+      "step": 2998
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00018670126193237835,
+      "loss": 0.9757,
+      "step": 2999
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00018669256484390523,
+      "loss": 0.9254,
+      "step": 3000
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00018668386511521234,
+      "loss": 0.8526,
+      "step": 3001
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0001866751627465646,
+      "loss": 1.0738,
+      "step": 3002
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00018666645773822706,
+      "loss": 0.9617,
+      "step": 3003
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0001866577500904648,
+      "loss": 0.9865,
+      "step": 3004
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00018664903980354308,
+      "loss": 0.8618,
+      "step": 3005
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0001866403268777271,
+      "loss": 0.9779,
+      "step": 3006
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00018663161131328223,
+      "loss": 0.9834,
+      "step": 3007
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00018662289311047395,
+      "loss": 1.1623,
+      "step": 3008
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0001866141722695677,
+      "loss": 0.9772,
+      "step": 3009
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00018660544879082913,
+      "loss": 1.0087,
+      "step": 3010
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00018659672267452385,
+      "loss": 1.1028,
+      "step": 3011
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00018658799392091768,
+      "loss": 0.9022,
+      "step": 3012
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00018657926253027644,
+      "loss": 0.9854,
+      "step": 3013
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00018657052850286603,
+      "loss": 0.9599,
+      "step": 3014
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00018656179183895244,
+      "loss": 0.9611,
+      "step": 3015
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00018655305253880173,
+      "loss": 0.838,
+      "step": 3016
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00018654431060268013,
+      "loss": 1.0147,
+      "step": 3017
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00018653556603085377,
+      "loss": 0.9666,
+      "step": 3018
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00018652681882358902,
+      "loss": 0.9545,
+      "step": 3019
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0001865180689811523,
+      "loss": 0.9936,
+      "step": 3020
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00018650931650381004,
+      "loss": 1.1216,
+      "step": 3021
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00018650056139182884,
+      "loss": 0.92,
+      "step": 3022
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00018649180364547528,
+      "loss": 0.9706,
+      "step": 3023
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00018648304326501615,
+      "loss": 0.8704,
+      "step": 3024
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00018647428025071817,
+      "loss": 0.8762,
+      "step": 3025
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00018646551460284826,
+      "loss": 0.8774,
+      "step": 3026
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0001864567463216734,
+      "loss": 0.9014,
+      "step": 3027
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00018644797540746057,
+      "loss": 1.0085,
+      "step": 3028
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00018643920186047695,
+      "loss": 0.8545,
+      "step": 3029
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0001864304256809897,
+      "loss": 0.961,
+      "step": 3030
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00018642164686926606,
+      "loss": 1.0072,
+      "step": 3031
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00018641286542557347,
+      "loss": 1.0437,
+      "step": 3032
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00018640408135017932,
+      "loss": 0.8153,
+      "step": 3033
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00018639529464335115,
+      "loss": 1.0075,
+      "step": 3034
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00018638650530535653,
+      "loss": 1.0133,
+      "step": 3035
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0001863777133364632,
+      "loss": 1.001,
+      "step": 3036
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00018636891873693884,
+      "loss": 0.9529,
+      "step": 3037
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00018636012150705133,
+      "loss": 0.9817,
+      "step": 3038
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00018635132164706862,
+      "loss": 0.8795,
+      "step": 3039
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00018634251915725862,
+      "loss": 1.0489,
+      "step": 3040
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00018633371403788948,
+      "loss": 0.9559,
+      "step": 3041
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00018632490628922937,
+      "loss": 1.1209,
+      "step": 3042
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0001863160959115465,
+      "loss": 0.9223,
+      "step": 3043
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0001863072829051092,
+      "loss": 1.015,
+      "step": 3044
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00018629846727018584,
+      "loss": 1.0835,
+      "step": 3045
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00018628964900704493,
+      "loss": 0.972,
+      "step": 3046
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00018628082811595505,
+      "loss": 0.9202,
+      "step": 3047
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0001862720045971848,
+      "loss": 0.8944,
+      "step": 3048
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0001862631784510029,
+      "loss": 0.9366,
+      "step": 3049
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00018625434967767815,
+      "loss": 0.9216,
+      "step": 3050
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0001862455182774795,
+      "loss": 0.8417,
+      "step": 3051
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00018623668425067578,
+      "loss": 1.1885,
+      "step": 3052
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00018622784759753618,
+      "loss": 0.9859,
+      "step": 3053
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00018621900831832972,
+      "loss": 1.0304,
+      "step": 3054
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0001862101664133256,
+      "loss": 0.9296,
+      "step": 3055
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00018620132188279317,
+      "loss": 1.0985,
+      "step": 3056
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0001861924747270017,
+      "loss": 0.8983,
+      "step": 3057
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00018618362494622067,
+      "loss": 0.8865,
+      "step": 3058
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00018617477254071965,
+      "loss": 0.9917,
+      "step": 3059
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00018616591751076817,
+      "loss": 0.906,
+      "step": 3060
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00018615705985663592,
+      "loss": 0.9764,
+      "step": 3061
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00018614819957859271,
+      "loss": 1.09,
+      "step": 3062
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00018613933667690832,
+      "loss": 1.1331,
+      "step": 3063
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00018613047115185268,
+      "loss": 0.8764,
+      "step": 3064
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0001861216030036958,
+      "loss": 0.9217,
+      "step": 3065
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00018611273223270779,
+      "loss": 1.03,
+      "step": 3066
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00018610385883915877,
+      "loss": 0.8718,
+      "step": 3067
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00018609498282331897,
+      "loss": 0.9704,
+      "step": 3068
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00018608610418545873,
+      "loss": 0.9687,
+      "step": 3069
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00018607722292584845,
+      "loss": 0.8967,
+      "step": 3070
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0001860683390447586,
+      "loss": 1.022,
+      "step": 3071
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0001860594525424598,
+      "loss": 0.8978,
+      "step": 3072
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00018605056341922257,
+      "loss": 0.9411,
+      "step": 3073
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00018604167167531773,
+      "loss": 1.0432,
+      "step": 3074
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.000186032777311016,
+      "loss": 0.8866,
+      "step": 3075
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0001860238803265883,
+      "loss": 0.8723,
+      "step": 3076
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00018601498072230558,
+      "loss": 0.9497,
+      "step": 3077
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00018600607849843893,
+      "loss": 0.8617,
+      "step": 3078
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00018599717365525934,
+      "loss": 0.9416,
+      "step": 3079
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00018598826619303815,
+      "loss": 0.9255,
+      "step": 3080
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00018597935611204652,
+      "loss": 0.9524,
+      "step": 3081
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0001859704434125559,
+      "loss": 0.9164,
+      "step": 3082
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00018596152809483764,
+      "loss": 1.0281,
+      "step": 3083
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00018595261015916332,
+      "loss": 0.8547,
+      "step": 3084
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00018594368960580453,
+      "loss": 0.955,
+      "step": 3085
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0001859347664350329,
+      "loss": 1.0195,
+      "step": 3086
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00018592584064712023,
+      "loss": 0.9934,
+      "step": 3087
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0001859169122423383,
+      "loss": 0.9812,
+      "step": 3088
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0001859079812209591,
+      "loss": 0.9534,
+      "step": 3089
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00018589904758325461,
+      "loss": 0.9487,
+      "step": 3090
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00018589011132949686,
+      "loss": 0.9736,
+      "step": 3091
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.000185881172459958,
+      "loss": 0.8925,
+      "step": 3092
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0001858722309749103,
+      "loss": 0.9104,
+      "step": 3093
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00018586328687462608,
+      "loss": 1.0915,
+      "step": 3094
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00018585434015937772,
+      "loss": 0.9695,
+      "step": 3095
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00018584539082943763,
+      "loss": 0.9549,
+      "step": 3096
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00018583643888507844,
+      "loss": 0.9674,
+      "step": 3097
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00018582748432657278,
+      "loss": 0.9539,
+      "step": 3098
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0001858185271541933,
+      "loss": 1.0522,
+      "step": 3099
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00018580956736821285,
+      "loss": 1.1333,
+      "step": 3100
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00018580060496890428,
+      "loss": 0.9384,
+      "step": 3101
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00018579163995654052,
+      "loss": 0.8628,
+      "step": 3102
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00018578267233139464,
+      "loss": 0.9562,
+      "step": 3103
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0001857737020937397,
+      "loss": 0.9503,
+      "step": 3104
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00018576472924384895,
+      "loss": 0.9671,
+      "step": 3105
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0001857557537819956,
+      "loss": 1.0269,
+      "step": 3106
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.000185746775708453,
+      "loss": 0.9781,
+      "step": 3107
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00018573779502349463,
+      "loss": 0.9533,
+      "step": 3108
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00018572881172739393,
+      "loss": 0.7373,
+      "step": 3109
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00018571982582042456,
+      "loss": 0.9792,
+      "step": 3110
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0001857108373028601,
+      "loss": 0.8765,
+      "step": 3111
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00018570184617497436,
+      "loss": 0.8512,
+      "step": 3112
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00018569285243704114,
+      "loss": 1.0222,
+      "step": 3113
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00018568385608933437,
+      "loss": 1.0529,
+      "step": 3114
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.000185674857132128,
+      "loss": 0.9839,
+      "step": 3115
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0001856658555656961,
+      "loss": 0.9704,
+      "step": 3116
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00018565685139031282,
+      "loss": 1.0503,
+      "step": 3117
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00018564784460625236,
+      "loss": 0.9451,
+      "step": 3118
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00018563883521378907,
+      "loss": 1.0024,
+      "step": 3119
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0001856298232131973,
+      "loss": 0.9269,
+      "step": 3120
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0001856208086047515,
+      "loss": 0.9674,
+      "step": 3121
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00018561179138872622,
+      "loss": 1.0417,
+      "step": 3122
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00018560277156539606,
+      "loss": 0.867,
+      "step": 3123
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00018559374913503578,
+      "loss": 1.0283,
+      "step": 3124
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00018558472409792008,
+      "loss": 0.944,
+      "step": 3125
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00018557569645432387,
+      "loss": 0.8935,
+      "step": 3126
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00018556666620452207,
+      "loss": 0.9705,
+      "step": 3127
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0001855576333487897,
+      "loss": 0.8764,
+      "step": 3128
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00018554859788740187,
+      "loss": 0.8866,
+      "step": 3129
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00018553955982063372,
+      "loss": 0.9759,
+      "step": 3130
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0001855305191487605,
+      "loss": 1.004,
+      "step": 3131
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0001855214758720576,
+      "loss": 0.9752,
+      "step": 3132
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00018551242999080035,
+      "loss": 0.916,
+      "step": 3133
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00018550338150526434,
+      "loss": 0.9526,
+      "step": 3134
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00018549433041572505,
+      "loss": 1.0936,
+      "step": 3135
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0001854852767224582,
+      "loss": 0.9771,
+      "step": 3136
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00018547622042573947,
+      "loss": 1.1076,
+      "step": 3137
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0001854671615258447,
+      "loss": 0.9686,
+      "step": 3138
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00018545810002304976,
+      "loss": 1.0679,
+      "step": 3139
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00018544903591763065,
+      "loss": 0.9314,
+      "step": 3140
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0001854399692098634,
+      "loss": 0.9457,
+      "step": 3141
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0001854308999000241,
+      "loss": 1.0127,
+      "step": 3142
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.000185421827988389,
+      "loss": 1.0716,
+      "step": 3143
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00018541275347523436,
+      "loss": 0.8045,
+      "step": 3144
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00018540367636083656,
+      "loss": 1.0305,
+      "step": 3145
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00018539459664547204,
+      "loss": 0.8621,
+      "step": 3146
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00018538551432941734,
+      "loss": 0.9259,
+      "step": 3147
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.000185376429412949,
+      "loss": 1.0005,
+      "step": 3148
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00018536734189634376,
+      "loss": 0.8689,
+      "step": 3149
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00018535825177987838,
+      "loss": 0.9874,
+      "step": 3150
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00018534915906382967,
+      "loss": 0.937,
+      "step": 3151
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00018534006374847455,
+      "loss": 0.9217,
+      "step": 3152
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00018533096583409005,
+      "loss": 0.8964,
+      "step": 3153
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00018532186532095322,
+      "loss": 0.9722,
+      "step": 3154
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00018531276220934122,
+      "loss": 0.8432,
+      "step": 3155
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00018530365649953128,
+      "loss": 0.8951,
+      "step": 3156
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0001852945481918007,
+      "loss": 0.913,
+      "step": 3157
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00018528543728642694,
+      "loss": 0.9997,
+      "step": 3158
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00018527632378368737,
+      "loss": 0.9352,
+      "step": 3159
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00018526720768385967,
+      "loss": 0.9168,
+      "step": 3160
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00018525808898722134,
+      "loss": 1.0242,
+      "step": 3161
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0001852489676940502,
+      "loss": 1.1251,
+      "step": 3162
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0001852398438046239,
+      "loss": 0.9986,
+      "step": 3163
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00018523071731922045,
+      "loss": 1.1238,
+      "step": 3164
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00018522158823811776,
+      "loss": 0.9119,
+      "step": 3165
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00018521245656159382,
+      "loss": 0.9249,
+      "step": 3166
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00018520332228992673,
+      "loss": 0.9851,
+      "step": 3167
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0001851941854233947,
+      "loss": 1.042,
+      "step": 3168
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00018518504596227601,
+      "loss": 0.9133,
+      "step": 3169
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.000185175903906849,
+      "loss": 0.9305,
+      "step": 3170
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00018516675925739202,
+      "loss": 1.0342,
+      "step": 3171
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00018515761201418364,
+      "loss": 0.9089,
+      "step": 3172
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00018514846217750243,
+      "loss": 0.8941,
+      "step": 3173
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00018513930974762703,
+      "loss": 0.9822,
+      "step": 3174
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00018513015472483623,
+      "loss": 0.8111,
+      "step": 3175
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00018512099710940873,
+      "loss": 1.6494,
+      "step": 3176
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00018511183690162356,
+      "loss": 0.8781,
+      "step": 3177
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00018510267410175958,
+      "loss": 0.9974,
+      "step": 3178
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0001850935087100959,
+      "loss": 0.9887,
+      "step": 3179
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00018508434072691168,
+      "loss": 0.9619,
+      "step": 3180
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00018507517015248608,
+      "loss": 0.9519,
+      "step": 3181
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0001850659969870984,
+      "loss": 0.9892,
+      "step": 3182
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00018505682123102802,
+      "loss": 0.9876,
+      "step": 3183
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00018504764288455438,
+      "loss": 0.9673,
+      "step": 3184
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.000185038461947957,
+      "loss": 0.9155,
+      "step": 3185
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00018502927842151552,
+      "loss": 0.9289,
+      "step": 3186
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00018502009230550955,
+      "loss": 0.972,
+      "step": 3187
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00018501090360021895,
+      "loss": 0.8786,
+      "step": 3188
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00018500171230592349,
+      "loss": 0.9952,
+      "step": 3189
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0001849925184229031,
+      "loss": 0.9752,
+      "step": 3190
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00018498332195143781,
+      "loss": 0.8393,
+      "step": 3191
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0001849741228918077,
+      "loss": 0.9951,
+      "step": 3192
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00018496492124429287,
+      "loss": 1.0805,
+      "step": 3193
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00018495571700917363,
+      "loss": 0.9127,
+      "step": 3194
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00018494651018673023,
+      "loss": 0.9935,
+      "step": 3195
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0001849373007772431,
+      "loss": 1.0497,
+      "step": 3196
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0001849280887809927,
+      "loss": 0.9538,
+      "step": 3197
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00018491887419825958,
+      "loss": 1.0308,
+      "step": 3198
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0001849096570293244,
+      "loss": 1.0034,
+      "step": 3199
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00018490043727446782,
+      "loss": 1.0667,
+      "step": 3200
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00018489121493397065,
+      "loss": 0.9283,
+      "step": 3201
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00018488199000811376,
+      "loss": 0.9581,
+      "step": 3202
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0001848727624971781,
+      "loss": 1.2201,
+      "step": 3203
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0001848635324014447,
+      "loss": 1.0359,
+      "step": 3204
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00018485429972119462,
+      "loss": 1.0542,
+      "step": 3205
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0001848450644567091,
+      "loss": 1.0171,
+      "step": 3206
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00018483582660826936,
+      "loss": 0.9998,
+      "step": 3207
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0001848265861761567,
+      "loss": 1.1871,
+      "step": 3208
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00018481734316065263,
+      "loss": 1.0547,
+      "step": 3209
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0001848080975620386,
+      "loss": 1.0069,
+      "step": 3210
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00018479884938059617,
+      "loss": 0.9845,
+      "step": 3211
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.000184789598616607,
+      "loss": 0.8941,
+      "step": 3212
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00018478034527035287,
+      "loss": 0.96,
+      "step": 3213
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00018477108934211552,
+      "loss": 0.9297,
+      "step": 3214
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00018476183083217688,
+      "loss": 1.0623,
+      "step": 3215
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00018475256974081886,
+      "loss": 1.066,
+      "step": 3216
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00018474330606832362,
+      "loss": 1.0525,
+      "step": 3217
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0001847340398149732,
+      "loss": 0.9354,
+      "step": 3218
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0001847247709810498,
+      "loss": 1.1632,
+      "step": 3219
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00018471549956683574,
+      "loss": 0.9631,
+      "step": 3220
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00018470622557261335,
+      "loss": 0.8371,
+      "step": 3221
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0001846969489986651,
+      "loss": 0.9746,
+      "step": 3222
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0001846876698452735,
+      "loss": 0.8794,
+      "step": 3223
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0001846783881127211,
+      "loss": 0.9075,
+      "step": 3224
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00018466910380129066,
+      "loss": 0.99,
+      "step": 3225
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00018465981691126484,
+      "loss": 0.9853,
+      "step": 3226
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00018465052744292654,
+      "loss": 0.8311,
+      "step": 3227
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00018464123539655865,
+      "loss": 0.9839,
+      "step": 3228
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00018463194077244415,
+      "loss": 1.0167,
+      "step": 3229
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00018462264357086615,
+      "loss": 0.9835,
+      "step": 3230
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0001846133437921077,
+      "loss": 1.0103,
+      "step": 3231
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0001846040414364521,
+      "loss": 1.0635,
+      "step": 3232
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00018459473650418268,
+      "loss": 1.0594,
+      "step": 3233
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00018458542899558276,
+      "loss": 0.9074,
+      "step": 3234
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0001845761189109358,
+      "loss": 0.9589,
+      "step": 3235
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00018456680625052538,
+      "loss": 0.9827,
+      "step": 3236
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00018455749101463508,
+      "loss": 1.0443,
+      "step": 3237
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00018454817320354863,
+      "loss": 0.9487,
+      "step": 3238
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00018453885281754977,
+      "loss": 0.9138,
+      "step": 3239
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00018452952985692237,
+      "loss": 1.0598,
+      "step": 3240
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00018452020432195037,
+      "loss": 1.0865,
+      "step": 3241
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00018451087621291774,
+      "loss": 0.8558,
+      "step": 3242
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00018450154553010858,
+      "loss": 0.9862,
+      "step": 3243
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00018449221227380714,
+      "loss": 0.9819,
+      "step": 3244
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00018448287644429754,
+      "loss": 0.8498,
+      "step": 3245
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00018447353804186415,
+      "loss": 1.0307,
+      "step": 3246
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0001844641970667914,
+      "loss": 0.9978,
+      "step": 3247
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0001844548535193637,
+      "loss": 0.9029,
+      "step": 3248
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0001844455073998657,
+      "loss": 0.9746,
+      "step": 3249
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00018443615870858198,
+      "loss": 0.9893,
+      "step": 3250
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00018442680744579722,
+      "loss": 1.0224,
+      "step": 3251
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00018441745361179632,
+      "loss": 0.9188,
+      "step": 3252
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00018440809720686406,
+      "loss": 0.7969,
+      "step": 3253
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0001843987382312854,
+      "loss": 0.9287,
+      "step": 3254
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0001843893766853454,
+      "loss": 0.9754,
+      "step": 3255
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00018438001256932914,
+      "loss": 0.935,
+      "step": 3256
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00018437064588352185,
+      "loss": 1.0683,
+      "step": 3257
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0001843612766282087,
+      "loss": 0.8732,
+      "step": 3258
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0001843519048036751,
+      "loss": 0.9745,
+      "step": 3259
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00018434253041020647,
+      "loss": 0.8603,
+      "step": 3260
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0001843331534480883,
+      "loss": 0.8326,
+      "step": 3261
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00018432377391760612,
+      "loss": 0.8998,
+      "step": 3262
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00018431439181904562,
+      "loss": 1.0757,
+      "step": 3263
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00018430500715269253,
+      "loss": 0.945,
+      "step": 3264
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0001842956199188327,
+      "loss": 0.9206,
+      "step": 3265
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00018428623011775192,
+      "loss": 0.9517,
+      "step": 3266
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00018427683774973626,
+      "loss": 1.0542,
+      "step": 3267
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0001842674428150717,
+      "loss": 0.9049,
+      "step": 3268
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00018425804531404437,
+      "loss": 1.0552,
+      "step": 3269
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0001842486452469405,
+      "loss": 0.9127,
+      "step": 3270
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00018423924261404633,
+      "loss": 0.8513,
+      "step": 3271
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00018422983741564825,
+      "loss": 1.0504,
+      "step": 3272
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00018422042965203267,
+      "loss": 0.9816,
+      "step": 3273
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00018421101932348613,
+      "loss": 1.021,
+      "step": 3274
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00018420160643029522,
+      "loss": 0.9404,
+      "step": 3275
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0001841921909727466,
+      "loss": 0.975,
+      "step": 3276
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00018418277295112698,
+      "loss": 0.8937,
+      "step": 3277
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00018417335236572323,
+      "loss": 0.9632,
+      "step": 3278
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00018416392921682224,
+      "loss": 1.0139,
+      "step": 3279
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00018415450350471103,
+      "loss": 0.9735,
+      "step": 3280
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00018414507522967658,
+      "loss": 1.0015,
+      "step": 3281
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0001841356443920061,
+      "loss": 0.9374,
+      "step": 3282
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00018412621099198675,
+      "loss": 0.8918,
+      "step": 3283
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00018411677502990588,
+      "loss": 0.8518,
+      "step": 3284
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00018410733650605084,
+      "loss": 0.9676,
+      "step": 3285
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00018409789542070902,
+      "loss": 0.9639,
+      "step": 3286
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00018408845177416807,
+      "loss": 1.0319,
+      "step": 3287
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0001840790055667155,
+      "loss": 0.8457,
+      "step": 3288
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00018406955679863902,
+      "loss": 0.86,
+      "step": 3289
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00018406010547022638,
+      "loss": 0.8967,
+      "step": 3290
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00018405065158176543,
+      "loss": 0.8928,
+      "step": 3291
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0001840411951335441,
+      "loss": 1.0091,
+      "step": 3292
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00018403173612585043,
+      "loss": 0.853,
+      "step": 3293
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0001840222745589724,
+      "loss": 0.8769,
+      "step": 3294
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0001840128104331982,
+      "loss": 1.0464,
+      "step": 3295
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00018400334374881608,
+      "loss": 0.957,
+      "step": 3296
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0001839938745061143,
+      "loss": 0.7981,
+      "step": 3297
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0001839844027053813,
+      "loss": 1.0585,
+      "step": 3298
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00018397492834690553,
+      "loss": 0.968,
+      "step": 3299
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00018396545143097546,
+      "loss": 1.062,
+      "step": 3300
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00018395597195787985,
+      "loss": 1.0226,
+      "step": 3301
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0001839464899279073,
+      "loss": 0.9793,
+      "step": 3302
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0001839370053413466,
+      "loss": 1.0762,
+      "step": 3303
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00018392751819848658,
+      "loss": 0.9709,
+      "step": 3304
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00018391802849961622,
+      "loss": 0.983,
+      "step": 3305
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00018390853624502447,
+      "loss": 0.9942,
+      "step": 3306
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00018389904143500049,
+      "loss": 0.9867,
+      "step": 3307
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0001838895440698334,
+      "loss": 1.0906,
+      "step": 3308
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0001838800441498124,
+      "loss": 1.0805,
+      "step": 3309
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0001838705416752269,
+      "loss": 0.9748,
+      "step": 3310
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00018386103664636626,
+      "loss": 1.0005,
+      "step": 3311
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00018385152906351992,
+      "loss": 0.9412,
+      "step": 3312
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00018384201892697748,
+      "loss": 0.9746,
+      "step": 3313
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00018383250623702857,
+      "loss": 1.0051,
+      "step": 3314
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00018382299099396288,
+      "loss": 0.9629,
+      "step": 3315
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00018381347319807017,
+      "loss": 0.9609,
+      "step": 3316
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00018380395284964035,
+      "loss": 0.984,
+      "step": 3317
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00018379442994896335,
+      "loss": 0.9875,
+      "step": 3318
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00018378490449632917,
+      "loss": 1.0113,
+      "step": 3319
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00018377537649202794,
+      "loss": 1.1338,
+      "step": 3320
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00018376584593634982,
+      "loss": 1.0425,
+      "step": 3321
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00018375631282958505,
+      "loss": 1.0318,
+      "step": 3322
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.000183746777172024,
+      "loss": 1.0972,
+      "step": 3323
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00018373723896395697,
+      "loss": 1.093,
+      "step": 3324
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0001837276982056746,
+      "loss": 1.0247,
+      "step": 3325
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00018371815489746736,
+      "loss": 0.8205,
+      "step": 3326
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00018370860903962593,
+      "loss": 0.8085,
+      "step": 3327
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00018369906063244097,
+      "loss": 0.9749,
+      "step": 3328
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00018368950967620332,
+      "loss": 1.0765,
+      "step": 3329
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0001836799561712039,
+      "loss": 1.0438,
+      "step": 3330
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00018367040011773352,
+      "loss": 0.8861,
+      "step": 3331
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0001836608415160834,
+      "loss": 1.09,
+      "step": 3332
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0001836512803665445,
+      "loss": 1.0937,
+      "step": 3333
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00018364171666940805,
+      "loss": 1.0251,
+      "step": 3334
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0001836321504249653,
+      "loss": 0.9774,
+      "step": 3335
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00018362258163350762,
+      "loss": 0.9331,
+      "step": 3336
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0001836130102953264,
+      "loss": 1.0194,
+      "step": 3337
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00018360343641071316,
+      "loss": 0.9297,
+      "step": 3338
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00018359385997995947,
+      "loss": 0.834,
+      "step": 3339
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00018358428100335693,
+      "loss": 0.9742,
+      "step": 3340
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00018357469948119734,
+      "loss": 1.0019,
+      "step": 3341
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00018356511541377246,
+      "loss": 0.924,
+      "step": 3342
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00018355552880137418,
+      "loss": 1.03,
+      "step": 3343
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00018354593964429447,
+      "loss": 1.0865,
+      "step": 3344
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00018353634794282536,
+      "loss": 1.0527,
+      "step": 3345
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00018352675369725895,
+      "loss": 1.0261,
+      "step": 3346
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00018351715690788747,
+      "loss": 0.9537,
+      "step": 3347
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00018350755757500314,
+      "loss": 0.9632,
+      "step": 3348
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00018349795569889838,
+      "loss": 1.0271,
+      "step": 3349
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00018348835127986554,
+      "loss": 1.0093,
+      "step": 3350
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00018347874431819716,
+      "loss": 0.9981,
+      "step": 3351
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00018346913481418584,
+      "loss": 0.9764,
+      "step": 3352
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00018345952276812417,
+      "loss": 0.9488,
+      "step": 3353
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00018344990818030494,
+      "loss": 0.9942,
+      "step": 3354
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00018344029105102095,
+      "loss": 1.0037,
+      "step": 3355
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0001834306713805651,
+      "loss": 0.8987,
+      "step": 3356
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0001834210491692303,
+      "loss": 1.0266,
+      "step": 3357
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00018341142441730964,
+      "loss": 1.0351,
+      "step": 3358
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00018340179712509626,
+      "loss": 1.0296,
+      "step": 3359
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00018339216729288334,
+      "loss": 0.8576,
+      "step": 3360
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00018338253492096416,
+      "loss": 0.9629,
+      "step": 3361
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00018337290000963207,
+      "loss": 0.9055,
+      "step": 3362
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00018336326255918048,
+      "loss": 0.8976,
+      "step": 3363
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00018335362256990292,
+      "loss": 0.9898,
+      "step": 3364
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00018334398004209295,
+      "loss": 0.9555,
+      "step": 3365
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0001833343349760443,
+      "loss": 0.9187,
+      "step": 3366
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00018332468737205067,
+      "loss": 1.0737,
+      "step": 3367
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00018331503723040583,
+      "loss": 1.0777,
+      "step": 3368
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00018330538455140375,
+      "loss": 0.9789,
+      "step": 3369
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00018329572933533834,
+      "loss": 0.9163,
+      "step": 3370
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0001832860715825037,
+      "loss": 0.9027,
+      "step": 3371
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00018327641129319396,
+      "loss": 0.9101,
+      "step": 3372
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00018326674846770325,
+      "loss": 1.1397,
+      "step": 3373
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00018325708310632592,
+      "loss": 0.9931,
+      "step": 3374
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0001832474152093563,
+      "loss": 0.8859,
+      "step": 3375
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00018323774477708887,
+      "loss": 1.0686,
+      "step": 3376
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00018322807180981806,
+      "loss": 0.9081,
+      "step": 3377
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00018321839630783854,
+      "loss": 0.8918,
+      "step": 3378
+    },
+    {
+      "epoch": 0.75,
+      "eval_loss": 0.6041999459266663,
+      "eval_runtime": 322.3612,
+      "eval_samples_per_second": 56.269,
+      "eval_steps_per_second": 28.136,
+      "step": 3378
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00018320871827144495,
+      "loss": 0.9068,
+      "step": 3379
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.000183199037700932,
+      "loss": 0.912,
+      "step": 3380
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00018318935459659457,
+      "loss": 1.0322,
+      "step": 3381
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00018317966895872753,
+      "loss": 1.0102,
+      "step": 3382
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00018316998078762585,
+      "loss": 0.9198,
+      "step": 3383
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00018316029008358454,
+      "loss": 0.9931,
+      "step": 3384
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00018315059684689881,
+      "loss": 1.0367,
+      "step": 3385
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00018314090107786387,
+      "loss": 0.8943,
+      "step": 3386
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00018313120277677492,
+      "loss": 1.073,
+      "step": 3387
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0001831215019439274,
+      "loss": 1.0005,
+      "step": 3388
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00018311179857961673,
+      "loss": 0.9407,
+      "step": 3389
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00018310209268413837,
+      "loss": 0.8238,
+      "step": 3390
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.000183092384257788,
+      "loss": 0.9657,
+      "step": 3391
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00018308267330086124,
+      "loss": 1.064,
+      "step": 3392
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00018307295981365385,
+      "loss": 0.8524,
+      "step": 3393
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00018306324379646163,
+      "loss": 1.0694,
+      "step": 3394
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0001830535252495805,
+      "loss": 0.9469,
+      "step": 3395
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00018304380417330645,
+      "loss": 1.0101,
+      "step": 3396
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00018303408056793554,
+      "loss": 0.8617,
+      "step": 3397
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00018302435443376386,
+      "loss": 0.9432,
+      "step": 3398
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00018301462577108765,
+      "loss": 0.9931,
+      "step": 3399
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0001830048945802032,
+      "loss": 1.1204,
+      "step": 3400
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00018299516086140686,
+      "loss": 0.9282,
+      "step": 3401
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00018298542461499509,
+      "loss": 0.9589,
+      "step": 3402
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00018297568584126433,
+      "loss": 0.9292,
+      "step": 3403
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0001829659445405113,
+      "loss": 0.9519,
+      "step": 3404
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00018295620071303258,
+      "loss": 0.9114,
+      "step": 3405
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00018294645435912497,
+      "loss": 1.0064,
+      "step": 3406
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00018293670547908522,
+      "loss": 0.9706,
+      "step": 3407
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00018292695407321033,
+      "loss": 0.9635,
+      "step": 3408
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00018291720014179723,
+      "loss": 1.0364,
+      "step": 3409
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00018290744368514294,
+      "loss": 0.9166,
+      "step": 3410
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00018289768470354466,
+      "loss": 1.0661,
+      "step": 3411
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00018288792319729954,
+      "loss": 1.0569,
+      "step": 3412
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0001828781591667049,
+      "loss": 0.9662,
+      "step": 3413
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00018286839261205814,
+      "loss": 0.967,
+      "step": 3414
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00018285862353365658,
+      "loss": 1.0278,
+      "step": 3415
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0001828488519317979,
+      "loss": 1.1957,
+      "step": 3416
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00018283907780677956,
+      "loss": 0.8776,
+      "step": 3417
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0001828293011588993,
+      "loss": 1.0013,
+      "step": 3418
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0001828195219884548,
+      "loss": 1.0163,
+      "step": 3419
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.000182809740295744,
+      "loss": 0.9119,
+      "step": 3420
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0001827999560810647,
+      "loss": 0.9259,
+      "step": 3421
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0001827901693447149,
+      "loss": 1.0132,
+      "step": 3422
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00018278038008699268,
+      "loss": 1.0237,
+      "step": 3423
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00018277058830819614,
+      "loss": 0.8932,
+      "step": 3424
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00018276079400862354,
+      "loss": 0.9767,
+      "step": 3425
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00018275099718857309,
+      "loss": 1.0698,
+      "step": 3426
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00018274119784834322,
+      "loss": 0.9283,
+      "step": 3427
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00018273139598823233,
+      "loss": 0.9128,
+      "step": 3428
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00018272159160853896,
+      "loss": 0.9056,
+      "step": 3429
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00018271178470956169,
+      "loss": 1.0638,
+      "step": 3430
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00018270197529159915,
+      "loss": 0.9286,
+      "step": 3431
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00018269216335495015,
+      "loss": 0.868,
+      "step": 3432
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0001826823488999135,
+      "loss": 0.8834,
+      "step": 3433
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00018267253192678806,
+      "loss": 0.9848,
+      "step": 3434
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00018266271243587285,
+      "loss": 0.9403,
+      "step": 3435
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00018265289042746687,
+      "loss": 0.9984,
+      "step": 3436
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00018264306590186932,
+      "loss": 1.0413,
+      "step": 3437
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00018263323885937935,
+      "loss": 1.0002,
+      "step": 3438
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00018262340930029627,
+      "loss": 0.9511,
+      "step": 3439
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0001826135772249194,
+      "loss": 0.8693,
+      "step": 3440
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00018260374263354826,
+      "loss": 0.9748,
+      "step": 3441
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00018259390552648224,
+      "loss": 0.9934,
+      "step": 3442
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00018258406590402102,
+      "loss": 0.9239,
+      "step": 3443
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00018257422376646425,
+      "loss": 0.962,
+      "step": 3444
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00018256437911411165,
+      "loss": 0.9402,
+      "step": 3445
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00018255453194726307,
+      "loss": 1.0421,
+      "step": 3446
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00018254468226621838,
+      "loss": 1.0537,
+      "step": 3447
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00018253483007127754,
+      "loss": 0.9875,
+      "step": 3448
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00018252497536274062,
+      "loss": 1.1031,
+      "step": 3449
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00018251511814090777,
+      "loss": 1.009,
+      "step": 3450
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00018250525840607913,
+      "loss": 0.9683,
+      "step": 3451
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.000182495396158555,
+      "loss": 1.1105,
+      "step": 3452
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0001824855313986358,
+      "loss": 1.0652,
+      "step": 3453
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00018247566412662186,
+      "loss": 1.0165,
+      "step": 3454
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00018246579434281376,
+      "loss": 0.8782,
+      "step": 3455
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00018245592204751203,
+      "loss": 1.0315,
+      "step": 3456
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00018244604724101737,
+      "loss": 0.832,
+      "step": 3457
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00018243616992363052,
+      "loss": 0.9764,
+      "step": 3458
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00018242629009565227,
+      "loss": 0.9327,
+      "step": 3459
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00018241640775738352,
+      "loss": 0.9203,
+      "step": 3460
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0001824065229091252,
+      "loss": 0.9751,
+      "step": 3461
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00018239663555117842,
+      "loss": 0.964,
+      "step": 3462
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00018238674568384427,
+      "loss": 0.9605,
+      "step": 3463
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00018237685330742395,
+      "loss": 1.1392,
+      "step": 3464
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00018236695842221873,
+      "loss": 1.0049,
+      "step": 3465
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00018235706102852993,
+      "loss": 1.0553,
+      "step": 3466
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.000182347161126659,
+      "loss": 0.996,
+      "step": 3467
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00018233725871690743,
+      "loss": 1.0419,
+      "step": 3468
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00018232735379957683,
+      "loss": 0.8903,
+      "step": 3469
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00018231744637496882,
+      "loss": 1.0283,
+      "step": 3470
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00018230753644338514,
+      "loss": 1.0249,
+      "step": 3471
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0001822976240051276,
+      "loss": 0.8141,
+      "step": 3472
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00018228770906049809,
+      "loss": 1.0569,
+      "step": 3473
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00018227779160979855,
+      "loss": 0.8443,
+      "step": 3474
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00018226787165333104,
+      "loss": 1.052,
+      "step": 3475
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00018225794919139761,
+      "loss": 0.89,
+      "step": 3476
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00018224802422430055,
+      "loss": 1.0733,
+      "step": 3477
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00018223809675234204,
+      "loss": 0.9495,
+      "step": 3478
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00018222816677582445,
+      "loss": 1.1676,
+      "step": 3479
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00018221823429505026,
+      "loss": 0.991,
+      "step": 3480
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00018220829931032184,
+      "loss": 0.9211,
+      "step": 3481
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00018219836182194183,
+      "loss": 0.9358,
+      "step": 3482
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00018218842183021286,
+      "loss": 0.9295,
+      "step": 3483
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0001821784793354377,
+      "loss": 0.8412,
+      "step": 3484
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0001821685343379191,
+      "loss": 0.9214,
+      "step": 3485
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00018215858683795993,
+      "loss": 0.9055,
+      "step": 3486
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00018214863683586313,
+      "loss": 1.0417,
+      "step": 3487
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00018213868433193176,
+      "loss": 0.9754,
+      "step": 3488
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00018212872932646894,
+      "loss": 0.9855,
+      "step": 3489
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00018211877181977784,
+      "loss": 0.907,
+      "step": 3490
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00018210881181216165,
+      "loss": 0.8236,
+      "step": 3491
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00018209884930392377,
+      "loss": 0.866,
+      "step": 3492
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0001820888842953676,
+      "loss": 1.0306,
+      "step": 3493
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00018207891678679664,
+      "loss": 1.0092,
+      "step": 3494
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00018206894677851436,
+      "loss": 1.007,
+      "step": 3495
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00018205897427082447,
+      "loss": 0.9447,
+      "step": 3496
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0001820489992640307,
+      "loss": 0.973,
+      "step": 3497
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00018203902175843682,
+      "loss": 0.9987,
+      "step": 3498
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00018202904175434665,
+      "loss": 1.0319,
+      "step": 3499
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0001820190592520642,
+      "loss": 0.8523,
+      "step": 3500
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00018200907425189344,
+      "loss": 1.0844,
+      "step": 3501
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00018199908675413848,
+      "loss": 0.9109,
+      "step": 3502
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00018198909675910346,
+      "loss": 0.9259,
+      "step": 3503
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00018197910426709268,
+      "loss": 1.0141,
+      "step": 3504
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0001819691092784104,
+      "loss": 0.9525,
+      "step": 3505
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0001819591117933611,
+      "loss": 0.9109,
+      "step": 3506
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00018194911181224917,
+      "loss": 0.9446,
+      "step": 3507
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00018193910933537918,
+      "loss": 0.8252,
+      "step": 3508
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00018192910436305577,
+      "loss": 0.9583,
+      "step": 3509
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00018191909689558363,
+      "loss": 0.9944,
+      "step": 3510
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0001819090869332676,
+      "loss": 1.0338,
+      "step": 3511
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0001818990744764124,
+      "loss": 0.9611,
+      "step": 3512
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00018188905952532309,
+      "loss": 0.9712,
+      "step": 3513
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00018187904208030463,
+      "loss": 0.9959,
+      "step": 3514
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00018186902214166204,
+      "loss": 1.0547,
+      "step": 3515
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00018185899970970054,
+      "loss": 0.9675,
+      "step": 3516
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0001818489747847254,
+      "loss": 0.9208,
+      "step": 3517
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00018183894736704183,
+      "loss": 0.9516,
+      "step": 3518
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00018182891745695528,
+      "loss": 0.8643,
+      "step": 3519
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0001818188850547712,
+      "loss": 0.9263,
+      "step": 3520
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0001818088501607951,
+      "loss": 1.081,
+      "step": 3521
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00018179881277533268,
+      "loss": 0.935,
+      "step": 3522
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00018178877289868952,
+      "loss": 0.9374,
+      "step": 3523
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00018177873053117142,
+      "loss": 0.9763,
+      "step": 3524
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00018176868567308424,
+      "loss": 1.0686,
+      "step": 3525
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00018175863832473388,
+      "loss": 0.872,
+      "step": 3526
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00018174858848642636,
+      "loss": 0.9879,
+      "step": 3527
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0001817385361584677,
+      "loss": 0.9455,
+      "step": 3528
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00018172848134116406,
+      "loss": 0.9198,
+      "step": 3529
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00018171842403482167,
+      "loss": 0.9712,
+      "step": 3530
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00018170836423974683,
+      "loss": 1.0764,
+      "step": 3531
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00018169830195624592,
+      "loss": 0.8457,
+      "step": 3532
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00018168823718462532,
+      "loss": 0.9827,
+      "step": 3533
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00018167816992519162,
+      "loss": 0.8398,
+      "step": 3534
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00018166810017825142,
+      "loss": 1.0648,
+      "step": 3535
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00018165802794411134,
+      "loss": 0.9839,
+      "step": 3536
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0001816479532230782,
+      "loss": 1.0256,
+      "step": 3537
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00018163787601545875,
+      "loss": 1.0138,
+      "step": 3538
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00018162779632155998,
+      "loss": 0.9665,
+      "step": 3539
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00018161771414168878,
+      "loss": 0.9688,
+      "step": 3540
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00018160762947615222,
+      "loss": 0.9693,
+      "step": 3541
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00018159754232525749,
+      "loss": 0.8987,
+      "step": 3542
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00018158745268931172,
+      "loss": 0.9845,
+      "step": 3543
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00018157736056862223,
+      "loss": 0.9793,
+      "step": 3544
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00018156726596349634,
+      "loss": 0.8193,
+      "step": 3545
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00018155716887424155,
+      "loss": 1.077,
+      "step": 3546
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00018154706930116529,
+      "loss": 1.1485,
+      "step": 3547
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0001815369672445752,
+      "loss": 1.1227,
+      "step": 3548
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00018152686270477887,
+      "loss": 0.8763,
+      "step": 3549
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0001815167556820841,
+      "loss": 0.9977,
+      "step": 3550
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00018150664617679868,
+      "loss": 0.947,
+      "step": 3551
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0001814965341892305,
+      "loss": 0.9373,
+      "step": 3552
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0001814864197196875,
+      "loss": 1.1264,
+      "step": 3553
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0001814763027684777,
+      "loss": 0.9221,
+      "step": 3554
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00018146618333590928,
+      "loss": 0.9984,
+      "step": 3555
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00018145606142229038,
+      "loss": 0.8354,
+      "step": 3556
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00018144593702792926,
+      "loss": 1.0977,
+      "step": 3557
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00018143581015313425,
+      "loss": 0.9923,
+      "step": 3558
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0001814256807982138,
+      "loss": 0.916,
+      "step": 3559
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00018141554896347636,
+      "loss": 1.0893,
+      "step": 3560
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00018140541464923052,
+      "loss": 1.0371,
+      "step": 3561
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0001813952778557849,
+      "loss": 0.9493,
+      "step": 3562
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00018138513858344826,
+      "loss": 1.0417,
+      "step": 3563
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00018137499683252935,
+      "loss": 0.97,
+      "step": 3564
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00018136485260333702,
+      "loss": 0.9163,
+      "step": 3565
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00018135470589618027,
+      "loss": 0.9747,
+      "step": 3566
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00018134455671136806,
+      "loss": 0.9508,
+      "step": 3567
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00018133440504920954,
+      "loss": 0.9878,
+      "step": 3568
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00018132425091001384,
+      "loss": 0.9817,
+      "step": 3569
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00018131409429409021,
+      "loss": 0.9499,
+      "step": 3570
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00018130393520174797,
+      "loss": 0.9748,
+      "step": 3571
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00018129377363329652,
+      "loss": 0.9601,
+      "step": 3572
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00018128360958904533,
+      "loss": 1.0032,
+      "step": 3573
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00018127344306930393,
+      "loss": 1.0672,
+      "step": 3574
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00018126327407438198,
+      "loss": 0.883,
+      "step": 3575
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0001812531026045891,
+      "loss": 0.9691,
+      "step": 3576
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0001812429286602352,
+      "loss": 0.8727,
+      "step": 3577
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00018123275224163,
+      "loss": 1.0601,
+      "step": 3578
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00018122257334908341,
+      "loss": 0.9926,
+      "step": 3579
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00018121239198290554,
+      "loss": 1.0321,
+      "step": 3580
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0001812022081434064,
+      "loss": 0.8955,
+      "step": 3581
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0001811920218308961,
+      "loss": 0.9989,
+      "step": 3582
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00018118183304568494,
+      "loss": 0.9299,
+      "step": 3583
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0001811716417880832,
+      "loss": 0.9004,
+      "step": 3584
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0001811614480584012,
+      "loss": 1.0376,
+      "step": 3585
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00018115125185694947,
+      "loss": 1.062,
+      "step": 3586
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00018114105318403848,
+      "loss": 0.9698,
+      "step": 3587
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00018113085203997882,
+      "loss": 0.9465,
+      "step": 3588
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0001811206484250812,
+      "loss": 0.9555,
+      "step": 3589
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0001811104423396564,
+      "loss": 0.9527,
+      "step": 3590
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00018110023378401516,
+      "loss": 1.1098,
+      "step": 3591
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00018109002275846844,
+      "loss": 0.9095,
+      "step": 3592
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00018107980926332723,
+      "loss": 1.123,
+      "step": 3593
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00018106959329890253,
+      "loss": 0.9078,
+      "step": 3594
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00018105937486550551,
+      "loss": 1.0554,
+      "step": 3595
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00018104915396344735,
+      "loss": 0.9053,
+      "step": 3596
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00018103893059303932,
+      "loss": 1.0208,
+      "step": 3597
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00018102870475459278,
+      "loss": 1.1568,
+      "step": 3598
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0001810184764484192,
+      "loss": 0.9423,
+      "step": 3599
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00018100824567483,
+      "loss": 0.9478,
+      "step": 3600
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00018099801243413687,
+      "loss": 1.0352,
+      "step": 3601
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00018098777672665136,
+      "loss": 0.9727,
+      "step": 3602
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00018097753855268522,
+      "loss": 1.0355,
+      "step": 3603
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0001809672979125503,
+      "loss": 0.9563,
+      "step": 3604
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00018095705480655842,
+      "loss": 0.9206,
+      "step": 3605
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00018094680923502157,
+      "loss": 1.0199,
+      "step": 3606
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00018093656119825177,
+      "loss": 0.9696,
+      "step": 3607
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00018092631069656113,
+      "loss": 0.8419,
+      "step": 3608
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0001809160577302618,
+      "loss": 0.922,
+      "step": 3609
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0001809058022996661,
+      "loss": 0.8865,
+      "step": 3610
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00018089554440508626,
+      "loss": 0.9845,
+      "step": 3611
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0001808852840468348,
+      "loss": 0.9169,
+      "step": 3612
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00018087502122522407,
+      "loss": 1.1185,
+      "step": 3613
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00018086475594056674,
+      "loss": 1.112,
+      "step": 3614
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00018085448819317535,
+      "loss": 0.911,
+      "step": 3615
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00018084421798336266,
+      "loss": 1.0229,
+      "step": 3616
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00018083394531144146,
+      "loss": 0.9511,
+      "step": 3617
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00018082367017772458,
+      "loss": 0.9441,
+      "step": 3618
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0001808133925825249,
+      "loss": 0.9246,
+      "step": 3619
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00018080311252615546,
+      "loss": 1.0581,
+      "step": 3620
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0001807928300089294,
+      "loss": 0.915,
+      "step": 3621
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00018078254503115982,
+      "loss": 1.0232,
+      "step": 3622
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00018077225759315993,
+      "loss": 0.9033,
+      "step": 3623
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00018076196769524304,
+      "loss": 0.8919,
+      "step": 3624
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00018075167533772256,
+      "loss": 0.8696,
+      "step": 3625
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00018074138052091195,
+      "loss": 0.9939,
+      "step": 3626
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00018073108324512472,
+      "loss": 1.1388,
+      "step": 3627
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00018072078351067444,
+      "loss": 0.9846,
+      "step": 3628
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00018071048131787482,
+      "loss": 1.0443,
+      "step": 3629
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00018070017666703961,
+      "loss": 0.9431,
+      "step": 3630
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00018068986955848264,
+      "loss": 0.9016,
+      "step": 3631
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00018067955999251782,
+      "loss": 0.8772,
+      "step": 3632
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.0001806692479694591,
+      "loss": 0.9322,
+      "step": 3633
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00018065893348962055,
+      "loss": 0.9338,
+      "step": 3634
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00018064861655331632,
+      "loss": 0.9885,
+      "step": 3635
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.0001806382971608606,
+      "loss": 0.9347,
+      "step": 3636
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00018062797531256762,
+      "loss": 1.0665,
+      "step": 3637
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00018061765100875176,
+      "loss": 0.8618,
+      "step": 3638
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.0001806073242497275,
+      "loss": 1.0358,
+      "step": 3639
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00018059699503580924,
+      "loss": 0.9058,
+      "step": 3640
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00018058666336731165,
+      "loss": 0.9686,
+      "step": 3641
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00018057632924454935,
+      "loss": 0.9673,
+      "step": 3642
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00018056599266783703,
+      "loss": 0.9373,
+      "step": 3643
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00018055565363748953,
+      "loss": 0.8195,
+      "step": 3644
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.0001805453121538217,
+      "loss": 0.9897,
+      "step": 3645
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00018053496821714854,
+      "loss": 0.9544,
+      "step": 3646
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.000180524621827785,
+      "loss": 0.9951,
+      "step": 3647
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00018051427298604623,
+      "loss": 0.8935,
+      "step": 3648
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00018050392169224736,
+      "loss": 0.9332,
+      "step": 3649
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00018049356794670374,
+      "loss": 0.8887,
+      "step": 3650
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00018048321174973054,
+      "loss": 1.1156,
+      "step": 3651
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00018047285310164327,
+      "loss": 0.9634,
+      "step": 3652
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.0001804624920027574,
+      "loss": 0.9403,
+      "step": 3653
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00018045212845338844,
+      "loss": 1.0341,
+      "step": 3654
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00018044176245385197,
+      "loss": 1.09,
+      "step": 3655
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.0001804313940044638,
+      "loss": 0.9615,
+      "step": 3656
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.0001804210231055396,
+      "loss": 1.0561,
+      "step": 3657
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00018041064975739523,
+      "loss": 0.9713,
+      "step": 3658
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00018040027396034667,
+      "loss": 0.8541,
+      "step": 3659
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00018038989571470988,
+      "loss": 1.0265,
+      "step": 3660
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.0001803795150208009,
+      "loss": 0.9412,
+      "step": 3661
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.0001803691318789359,
+      "loss": 1.0001,
+      "step": 3662
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00018035874628943111,
+      "loss": 1.0619,
+      "step": 3663
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00018034835825260284,
+      "loss": 1.0296,
+      "step": 3664
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.0001803379677687674,
+      "loss": 0.9775,
+      "step": 3665
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00018032757483824123,
+      "loss": 1.0712,
+      "step": 3666
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00018031717946134093,
+      "loss": 0.9167,
+      "step": 3667
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00018030678163838298,
+      "loss": 0.901,
+      "step": 3668
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00018029638136968416,
+      "loss": 0.9014,
+      "step": 3669
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00018028597865556113,
+      "loss": 1.0816,
+      "step": 3670
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00018027557349633072,
+      "loss": 0.9834,
+      "step": 3671
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00018026516589230983,
+      "loss": 1.0476,
+      "step": 3672
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00018025475584381543,
+      "loss": 0.9895,
+      "step": 3673
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00018024434335116456,
+      "loss": 1.0127,
+      "step": 3674
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.0001802339284146743,
+      "loss": 0.8752,
+      "step": 3675
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00018022351103466185,
+      "loss": 1.0525,
+      "step": 3676
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00018021309121144446,
+      "loss": 1.0404,
+      "step": 3677
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.0001802026689453395,
+      "loss": 0.9187,
+      "step": 3678
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00018019224423666437,
+      "loss": 0.9486,
+      "step": 3679
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00018018181708573657,
+      "loss": 0.9664,
+      "step": 3680
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.0001801713874928736,
+      "loss": 0.9814,
+      "step": 3681
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00018016095545839313,
+      "loss": 0.9798,
+      "step": 3682
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00018015052098261286,
+      "loss": 1.1809,
+      "step": 3683
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.0001801400840658506,
+      "loss": 1.0592,
+      "step": 3684
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00018012964470842416,
+      "loss": 1.063,
+      "step": 3685
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.0001801192029106515,
+      "loss": 0.9198,
+      "step": 3686
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00018010875867285065,
+      "loss": 0.93,
+      "step": 3687
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00018009831199533962,
+      "loss": 0.9862,
+      "step": 3688
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00018008786287843662,
+      "loss": 1.1241,
+      "step": 3689
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00018007741132245982,
+      "loss": 1.0411,
+      "step": 3690
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.0001800669573277276,
+      "loss": 1.0503,
+      "step": 3691
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.0001800565008945583,
+      "loss": 0.9602,
+      "step": 3692
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00018004604202327037,
+      "loss": 0.8753,
+      "step": 3693
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.0001800355807141823,
+      "loss": 1.054,
+      "step": 3694
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00018002511696761273,
+      "loss": 0.9768,
+      "step": 3695
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00018001465078388033,
+      "loss": 0.8944,
+      "step": 3696
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00018000418216330383,
+      "loss": 0.9442,
+      "step": 3697
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00017999371110620206,
+      "loss": 1.0152,
+      "step": 3698
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00017998323761289395,
+      "loss": 0.872,
+      "step": 3699
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.0001799727616836984,
+      "loss": 1.004,
+      "step": 3700
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.0001799622833189345,
+      "loss": 1.0524,
+      "step": 3701
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00017995180251892138,
+      "loss": 1.0933,
+      "step": 3702
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.0001799413192839782,
+      "loss": 1.0284,
+      "step": 3703
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00017993083361442425,
+      "loss": 0.8963,
+      "step": 3704
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00017992034551057883,
+      "loss": 0.9361,
+      "step": 3705
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.0001799098549727614,
+      "loss": 0.947,
+      "step": 3706
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00017989936200129146,
+      "loss": 1.0471,
+      "step": 3707
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00017988886659648853,
+      "loss": 1.003,
+      "step": 3708
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00017987836875867227,
+      "loss": 1.0316,
+      "step": 3709
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00017986786848816236,
+      "loss": 0.9394,
+      "step": 3710
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00017985736578527862,
+      "loss": 0.8616,
+      "step": 3711
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.0001798468606503409,
+      "loss": 0.9866,
+      "step": 3712
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00017983635308366916,
+      "loss": 0.9935,
+      "step": 3713
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00017982584308558335,
+      "loss": 0.9583,
+      "step": 3714
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00017981533065640362,
+      "loss": 0.8774,
+      "step": 3715
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00017980481579645005,
+      "loss": 0.9005,
+      "step": 3716
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00017979429850604293,
+      "loss": 0.9851,
+      "step": 3717
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00017978377878550254,
+      "loss": 0.9196,
+      "step": 3718
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.0001797732566351493,
+      "loss": 1.0356,
+      "step": 3719
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00017976273205530356,
+      "loss": 0.988,
+      "step": 3720
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00017975220504628595,
+      "loss": 1.0433,
+      "step": 3721
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.000179741675608417,
+      "loss": 1.032,
+      "step": 3722
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00017973114374201743,
+      "loss": 0.9239,
+      "step": 3723
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00017972060944740795,
+      "loss": 0.8542,
+      "step": 3724
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00017971007272490943,
+      "loss": 0.9528,
+      "step": 3725
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00017969953357484272,
+      "loss": 1.0172,
+      "step": 3726
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00017968899199752883,
+      "loss": 1.0021,
+      "step": 3727
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00017967844799328874,
+      "loss": 1.0239,
+      "step": 3728
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00017966790156244365,
+      "loss": 1.0437,
+      "step": 3729
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.0001796573527053147,
+      "loss": 0.9666,
+      "step": 3730
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00017964680142222315,
+      "loss": 1.0281,
+      "step": 3731
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00017963624771349034,
+      "loss": 1.0121,
+      "step": 3732
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00017962569157943772,
+      "loss": 0.9449,
+      "step": 3733
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00017961513302038675,
+      "loss": 0.8758,
+      "step": 3734
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00017960457203665896,
+      "loss": 0.9431,
+      "step": 3735
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00017959400862857606,
+      "loss": 0.9819,
+      "step": 3736
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.0001795834427964597,
+      "loss": 0.8586,
+      "step": 3737
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00017957287454063167,
+      "loss": 1.0063,
+      "step": 3738
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00017956230386141387,
+      "loss": 1.001,
+      "step": 3739
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00017955173075912817,
+      "loss": 0.9154,
+      "step": 3740
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00017954115523409657,
+      "loss": 0.8675,
+      "step": 3741
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00017953057728664122,
+      "loss": 1.0721,
+      "step": 3742
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.0001795199969170842,
+      "loss": 1.0271,
+      "step": 3743
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00017950941412574775,
+      "loss": 0.7502,
+      "step": 3744
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00017949882891295422,
+      "loss": 0.9442,
+      "step": 3745
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00017948824127902592,
+      "loss": 0.8288,
+      "step": 3746
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00017947765122428532,
+      "loss": 0.8865,
+      "step": 3747
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00017946705874905493,
+      "loss": 0.9369,
+      "step": 3748
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00017945646385365737,
+      "loss": 1.0022,
+      "step": 3749
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.0001794458665384153,
+      "loss": 0.9491,
+      "step": 3750
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00017943526680365142,
+      "loss": 0.8115,
+      "step": 3751
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00017942466464968858,
+      "loss": 0.9015,
+      "step": 3752
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00017941406007684964,
+      "loss": 0.9301,
+      "step": 3753
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00017940345308545762,
+      "loss": 1.1204,
+      "step": 3754
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.0001793928436758355,
+      "loss": 0.9384,
+      "step": 3755
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00017938223184830643,
+      "loss": 0.9337,
+      "step": 3756
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00017937161760319358,
+      "loss": 0.9621,
+      "step": 3757
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00017936100094082015,
+      "loss": 1.0032,
+      "step": 3758
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00017935038186150956,
+      "loss": 0.9007,
+      "step": 3759
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00017933976036558518,
+      "loss": 1.0344,
+      "step": 3760
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00017932913645337046,
+      "loss": 0.9468,
+      "step": 3761
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00017931851012518898,
+      "loss": 1.0411,
+      "step": 3762
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00017930788138136432,
+      "loss": 0.9639,
+      "step": 3763
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00017929725022222025,
+      "loss": 0.8659,
+      "step": 3764
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00017928661664808052,
+      "loss": 0.913,
+      "step": 3765
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00017927598065926894,
+      "loss": 0.8508,
+      "step": 3766
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00017926534225610943,
+      "loss": 0.8101,
+      "step": 3767
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00017925470143892604,
+      "loss": 1.0102,
+      "step": 3768
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.0001792440582080428,
+      "loss": 1.0804,
+      "step": 3769
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.0001792334125637838,
+      "loss": 0.9579,
+      "step": 3770
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00017922276450647333,
+      "loss": 0.9351,
+      "step": 3771
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00017921211403643564,
+      "loss": 0.8716,
+      "step": 3772
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00017920146115399513,
+      "loss": 0.8891,
+      "step": 3773
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00017919080585947615,
+      "loss": 0.8904,
+      "step": 3774
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00017918014815320328,
+      "loss": 0.9294,
+      "step": 3775
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00017916948803550108,
+      "loss": 0.9604,
+      "step": 3776
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00017915882550669418,
+      "loss": 0.9218,
+      "step": 3777
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00017914816056710734,
+      "loss": 0.9362,
+      "step": 3778
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00017913749321706534,
+      "loss": 0.8767,
+      "step": 3779
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00017912682345689308,
+      "loss": 1.0258,
+      "step": 3780
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00017911615128691547,
+      "loss": 0.9886,
+      "step": 3781
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00017910547670745757,
+      "loss": 1.0153,
+      "step": 3782
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00017909479971884445,
+      "loss": 0.9233,
+      "step": 3783
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00017908412032140126,
+      "loss": 1.0104,
+      "step": 3784
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00017907343851545325,
+      "loss": 1.0077,
+      "step": 3785
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.0001790627543013258,
+      "loss": 0.8929,
+      "step": 3786
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00017905206767934416,
+      "loss": 0.8973,
+      "step": 3787
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00017904137864983394,
+      "loss": 0.9054,
+      "step": 3788
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00017903068721312058,
+      "loss": 0.9656,
+      "step": 3789
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00017901999336952974,
+      "loss": 0.9592,
+      "step": 3790
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00017900929711938708,
+      "loss": 0.9028,
+      "step": 3791
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.0001789985984630183,
+      "loss": 1.0485,
+      "step": 3792
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00017898789740074933,
+      "loss": 0.8474,
+      "step": 3793
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.000178977193932906,
+      "loss": 0.9613,
+      "step": 3794
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.0001789664880598143,
+      "loss": 0.9342,
+      "step": 3795
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00017895577978180026,
+      "loss": 0.9543,
+      "step": 3796
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00017894506909919,
+      "loss": 0.9154,
+      "step": 3797
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00017893435601230978,
+      "loss": 1.0614,
+      "step": 3798
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00017892364052148583,
+      "loss": 0.9096,
+      "step": 3799
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00017891292262704446,
+      "loss": 0.8733,
+      "step": 3800
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.0001789022023293121,
+      "loss": 0.9392,
+      "step": 3801
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00017889147962861523,
+      "loss": 0.858,
+      "step": 3802
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00017888075452528044,
+      "loss": 0.9226,
+      "step": 3803
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.0001788700270196343,
+      "loss": 0.9629,
+      "step": 3804
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.0001788592971120036,
+      "loss": 1.0822,
+      "step": 3805
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00017884856480271507,
+      "loss": 1.0201,
+      "step": 3806
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00017883783009209556,
+      "loss": 0.9937,
+      "step": 3807
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.000178827092980472,
+      "loss": 1.0254,
+      "step": 3808
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00017881635346817142,
+      "loss": 0.9419,
+      "step": 3809
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00017880561155552083,
+      "loss": 0.9318,
+      "step": 3810
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.0001787948672428474,
+      "loss": 0.9963,
+      "step": 3811
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00017878412053047837,
+      "loss": 0.9254,
+      "step": 3812
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00017877337141874105,
+      "loss": 1.0121,
+      "step": 3813
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00017876261990796272,
+      "loss": 1.0132,
+      "step": 3814
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.0001787518659984709,
+      "loss": 0.9132,
+      "step": 3815
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00017874110969059306,
+      "loss": 1.0523,
+      "step": 3816
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00017873035098465678,
+      "loss": 0.9072,
+      "step": 3817
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00017871958988098974,
+      "loss": 0.7759,
+      "step": 3818
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00017870882637991965,
+      "loss": 1.1936,
+      "step": 3819
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00017869806048177432,
+      "loss": 0.8644,
+      "step": 3820
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00017868729218688162,
+      "loss": 0.8999,
+      "step": 3821
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00017867652149556953,
+      "loss": 1.0791,
+      "step": 3822
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.000178665748408166,
+      "loss": 0.9442,
+      "step": 3823
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.0001786549729249992,
+      "loss": 0.8944,
+      "step": 3824
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00017864419504639724,
+      "loss": 0.9895,
+      "step": 3825
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.0001786334147726884,
+      "loss": 0.9801,
+      "step": 3826
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00017862263210420096,
+      "loss": 0.9249,
+      "step": 3827
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00017861184704126334,
+      "loss": 0.9954,
+      "step": 3828
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.000178601059584204,
+      "loss": 0.9708,
+      "step": 3829
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00017859026973335141,
+      "loss": 0.8449,
+      "step": 3830
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00017857947748903423,
+      "loss": 1.0638,
+      "step": 3831
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00017856868285158112,
+      "loss": 0.9721,
+      "step": 3832
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00017855788582132087,
+      "loss": 0.8243,
+      "step": 3833
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00017854708639858225,
+      "loss": 0.9005,
+      "step": 3834
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00017853628458369415,
+      "loss": 0.831,
+      "step": 3835
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00017852548037698563,
+      "loss": 0.9801,
+      "step": 3836
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.0001785146737787856,
+      "loss": 0.9795,
+      "step": 3837
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00017850386478942328,
+      "loss": 1.0158,
+      "step": 3838
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00017849305340922778,
+      "loss": 1.0465,
+      "step": 3839
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00017848223963852844,
+      "loss": 1.0437,
+      "step": 3840
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00017847142347765453,
+      "loss": 1.0147,
+      "step": 3841
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00017846060492693547,
+      "loss": 0.9854,
+      "step": 3842
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00017844978398670077,
+      "loss": 0.8482,
+      "step": 3843
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00017843896065727994,
+      "loss": 0.9864,
+      "step": 3844
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00017842813493900263,
+      "loss": 0.9137,
+      "step": 3845
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.0001784173068321985,
+      "loss": 0.9554,
+      "step": 3846
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.0001784064763371974,
+      "loss": 0.8646,
+      "step": 3847
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.0001783956434543291,
+      "loss": 0.9281,
+      "step": 3848
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00017838480818392353,
+      "loss": 1.0063,
+      "step": 3849
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.0001783739705263107,
+      "loss": 1.0198,
+      "step": 3850
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.0001783631304818206,
+      "loss": 1.0194,
+      "step": 3851
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00017835228805078348,
+      "loss": 0.8931,
+      "step": 3852
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00017834144323352946,
+      "loss": 0.9791,
+      "step": 3853
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00017833059603038884,
+      "loss": 0.947,
+      "step": 3854
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00017831974644169198,
+      "loss": 0.9093,
+      "step": 3855
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.0001783088944677693,
+      "loss": 0.8984,
+      "step": 3856
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.0001782980401089513,
+      "loss": 0.9343,
+      "step": 3857
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.0001782871833655685,
+      "loss": 1.0041,
+      "step": 3858
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00017827632423795164,
+      "loss": 0.9826,
+      "step": 3859
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00017826546272643135,
+      "loss": 0.946,
+      "step": 3860
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00017825459883133847,
+      "loss": 0.8757,
+      "step": 3861
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00017824373255300382,
+      "loss": 0.9332,
+      "step": 3862
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00017823286389175835,
+      "loss": 0.9121,
+      "step": 3863
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00017822199284793307,
+      "loss": 0.9927,
+      "step": 3864
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00017821111942185904,
+      "loss": 0.845,
+      "step": 3865
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00017820024361386744,
+      "loss": 0.9588,
+      "step": 3866
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00017818936542428944,
+      "loss": 0.8574,
+      "step": 3867
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.0001781784848534564,
+      "loss": 0.8919,
+      "step": 3868
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00017816760190169964,
+      "loss": 1.0092,
+      "step": 3869
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.0001781567165693506,
+      "loss": 1.0174,
+      "step": 3870
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00017814582885674086,
+      "loss": 0.9973,
+      "step": 3871
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.0001781349387642019,
+      "loss": 0.9125,
+      "step": 3872
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00017812404629206547,
+      "loss": 1.0339,
+      "step": 3873
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00017811315144066324,
+      "loss": 0.9715,
+      "step": 3874
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00017810225421032702,
+      "loss": 1.1383,
+      "step": 3875
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.0001780913546013887,
+      "loss": 0.957,
+      "step": 3876
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00017808045261418024,
+      "loss": 0.8561,
+      "step": 3877
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00017806954824903365,
+      "loss": 1.01,
+      "step": 3878
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00017805864150628102,
+      "loss": 1.0431,
+      "step": 3879
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.0001780477323862545,
+      "loss": 0.9252,
+      "step": 3880
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00017803682088928633,
+      "loss": 0.9988,
+      "step": 3881
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00017802590701570885,
+      "loss": 0.9212,
+      "step": 3882
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00017801499076585442,
+      "loss": 1.1108,
+      "step": 3883
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00017800407214005548,
+      "loss": 1.0681,
+      "step": 3884
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00017799315113864456,
+      "loss": 0.8719,
+      "step": 3885
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00017798222776195427,
+      "loss": 0.9814,
+      "step": 3886
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.0001779713020103173,
+      "loss": 0.8766,
+      "step": 3887
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00017796037388406638,
+      "loss": 0.867,
+      "step": 3888
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.0001779494433835343,
+      "loss": 0.793,
+      "step": 3889
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.000177938510509054,
+      "loss": 1.0618,
+      "step": 3890
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00017792757526095837,
+      "loss": 0.9119,
+      "step": 3891
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00017791663763958048,
+      "loss": 0.8895,
+      "step": 3892
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00017790569764525345,
+      "loss": 0.9502,
+      "step": 3893
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00017789475527831044,
+      "loss": 0.9699,
+      "step": 3894
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.0001778838105390847,
+      "loss": 1.0607,
+      "step": 3895
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00017787286342790955,
+      "loss": 1.0614,
+      "step": 3896
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.0001778619139451184,
+      "loss": 1.0082,
+      "step": 3897
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.0001778509620910447,
+      "loss": 0.9608,
+      "step": 3898
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.000177840007866022,
+      "loss": 0.9294,
+      "step": 3899
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.0001778290512703839,
+      "loss": 0.996,
+      "step": 3900
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00017781809230446406,
+      "loss": 1.0582,
+      "step": 3901
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00017780713096859628,
+      "loss": 0.9456,
+      "step": 3902
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00017779616726311441,
+      "loss": 0.9673,
+      "step": 3903
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00017778520118835224,
+      "loss": 0.9172,
+      "step": 3904
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00017777423274464385,
+      "loss": 0.9072,
+      "step": 3905
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00017776326193232324,
+      "loss": 1.0204,
+      "step": 3906
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.0001777522887517245,
+      "loss": 0.9981,
+      "step": 3907
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00017774131320318187,
+      "loss": 0.9276,
+      "step": 3908
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.0001777303352870296,
+      "loss": 0.8912,
+      "step": 3909
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.000177719355003602,
+      "loss": 0.9682,
+      "step": 3910
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00017770837235323348,
+      "loss": 1.0863,
+      "step": 3911
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00017769738733625852,
+      "loss": 0.9151,
+      "step": 3912
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00017768639995301166,
+      "loss": 1.0366,
+      "step": 3913
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00017767541020382755,
+      "loss": 0.9137,
+      "step": 3914
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00017766441808904083,
+      "loss": 0.8975,
+      "step": 3915
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00017765342360898632,
+      "loss": 0.9011,
+      "step": 3916
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.0001776424267639988,
+      "loss": 1.0071,
+      "step": 3917
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00017763142755441324,
+      "loss": 0.8934,
+      "step": 3918
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.0001776204259805646,
+      "loss": 1.1554,
+      "step": 3919
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.0001776094220427879,
+      "loss": 0.864,
+      "step": 3920
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.0001775984157414183,
+      "loss": 1.038,
+      "step": 3921
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00017758740707679097,
+      "loss": 0.8711,
+      "step": 3922
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.0001775763960492412,
+      "loss": 0.9909,
+      "step": 3923
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00017756538265910435,
+      "loss": 0.9643,
+      "step": 3924
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00017755436690671577,
+      "loss": 1.0109,
+      "step": 3925
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.000177543348792411,
+      "loss": 1.0296,
+      "step": 3926
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00017753232831652557,
+      "loss": 0.8411,
+      "step": 3927
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00017752130547939513,
+      "loss": 1.027,
+      "step": 3928
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00017751028028135536,
+      "loss": 0.9419,
+      "step": 3929
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00017749925272274206,
+      "loss": 0.9264,
+      "step": 3930
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00017748822280389098,
+      "loss": 0.9486,
+      "step": 3931
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00017747719052513817,
+      "loss": 0.8744,
+      "step": 3932
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00017746615588681955,
+      "loss": 0.9452,
+      "step": 3933
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.0001774551188892712,
+      "loss": 0.9082,
+      "step": 3934
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.0001774440795328292,
+      "loss": 1.038,
+      "step": 3935
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.0001774330378178298,
+      "loss": 0.9712,
+      "step": 3936
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.0001774219937446093,
+      "loss": 0.9345,
+      "step": 3937
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00017741094731350398,
+      "loss": 1.0057,
+      "step": 3938
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.0001773998985248503,
+      "loss": 1.0346,
+      "step": 3939
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00017738884737898475,
+      "loss": 0.8498,
+      "step": 3940
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.00017737779387624384,
+      "loss": 1.1082,
+      "step": 3941
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.0001773667380169643,
+      "loss": 1.0784,
+      "step": 3942
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.00017735567980148276,
+      "loss": 1.0556,
+      "step": 3943
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.00017734461923013604,
+      "loss": 0.9296,
+      "step": 3944
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.00017733355630326095,
+      "loss": 0.8886,
+      "step": 3945
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.00017732249102119446,
+      "loss": 0.9366,
+      "step": 3946
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.0001773114233842735,
+      "loss": 0.9748,
+      "step": 3947
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.0001773003533928352,
+      "loss": 0.9884,
+      "step": 3948
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.00017728928104721666,
+      "loss": 1.0351,
+      "step": 3949
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.00017727820634775508,
+      "loss": 0.7973,
+      "step": 3950
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.00017726712929478774,
+      "loss": 0.9887,
+      "step": 3951
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.00017725604988865206,
+      "loss": 0.939,
+      "step": 3952
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.00017724496812968537,
+      "loss": 1.0253,
+      "step": 3953
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.0001772338840182252,
+      "loss": 0.9347,
+      "step": 3954
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.0001772227975546091,
+      "loss": 0.8774,
+      "step": 3955
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.00017721170873917478,
+      "loss": 0.9289,
+      "step": 3956
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.00017720061757225985,
+      "loss": 1.152,
+      "step": 3957
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.00017718952405420217,
+      "loss": 1.001,
+      "step": 3958
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.00017717842818533956,
+      "loss": 1.0007,
+      "step": 3959
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.00017716732996600994,
+      "loss": 0.9844,
+      "step": 3960
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.0001771562293965513,
+      "loss": 1.0136,
+      "step": 3961
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.00017714512647730172,
+      "loss": 0.9481,
+      "step": 3962
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.00017713402120859937,
+      "loss": 0.9352,
+      "step": 3963
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.0001771229135907824,
+      "loss": 1.0407,
+      "step": 3964
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.00017711180362418912,
+      "loss": 1.0053,
+      "step": 3965
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.00017710069130915786,
+      "loss": 0.9823,
+      "step": 3966
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.0001770895766460271,
+      "loss": 0.9703,
+      "step": 3967
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.00017707845963513532,
+      "loss": 0.9027,
+      "step": 3968
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.00017706734027682106,
+      "loss": 0.908,
+      "step": 3969
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.00017705621857142298,
+      "loss": 1.0586,
+      "step": 3970
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.00017704509451927977,
+      "loss": 1.0102,
+      "step": 3971
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.00017703396812073025,
+      "loss": 1.0377,
+      "step": 3972
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.00017702283937611323,
+      "loss": 0.9851,
+      "step": 3973
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.00017701170828576769,
+      "loss": 0.9909,
+      "step": 3974
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.00017700057485003256,
+      "loss": 0.9981,
+      "step": 3975
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.00017698943906924695,
+      "loss": 0.9784,
+      "step": 3976
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.00017697830094375,
+      "loss": 0.9241,
+      "step": 3977
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.00017696716047388094,
+      "loss": 1.0256,
+      "step": 3978
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.000176956017659979,
+      "loss": 0.9924,
+      "step": 3979
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.00017694487250238355,
+      "loss": 0.9396,
+      "step": 3980
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.00017693372500143405,
+      "loss": 0.9746,
+      "step": 3981
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.00017692257515747,
+      "loss": 0.8875,
+      "step": 3982
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.0001769114229708309,
+      "loss": 0.8912,
+      "step": 3983
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.00017690026844185643,
+      "loss": 0.901,
+      "step": 3984
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.00017688911157088633,
+      "loss": 1.0095,
+      "step": 3985
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00017687795235826035,
+      "loss": 1.0739,
+      "step": 3986
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00017686679080431832,
+      "loss": 0.9814,
+      "step": 3987
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00017685562690940022,
+      "loss": 1.0664,
+      "step": 3988
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00017684446067384602,
+      "loss": 1.0606,
+      "step": 3989
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00017683329209799577,
+      "loss": 0.958,
+      "step": 3990
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00017682212118218967,
+      "loss": 0.9883,
+      "step": 3991
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00017681094792676785,
+      "loss": 0.8766,
+      "step": 3992
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.0001767997723320706,
+      "loss": 0.9555,
+      "step": 3993
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00017678859439843834,
+      "loss": 1.026,
+      "step": 3994
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00017677741412621144,
+      "loss": 0.9784,
+      "step": 3995
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.0001767662315157304,
+      "loss": 0.9896,
+      "step": 3996
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.0001767550465673358,
+      "loss": 0.9588,
+      "step": 3997
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.0001767438592813683,
+      "loss": 0.7832,
+      "step": 3998
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00017673266965816855,
+      "loss": 0.9414,
+      "step": 3999
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00017672147769807737,
+      "loss": 0.9895,
+      "step": 4000
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00017671028340143557,
+      "loss": 0.9295,
+      "step": 4001
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00017669908676858412,
+      "loss": 0.9958,
+      "step": 4002
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00017668788779986403,
+      "loss": 0.8445,
+      "step": 4003
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.0001766766864956163,
+      "loss": 1.0238,
+      "step": 4004
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.0001766654828561821,
+      "loss": 0.8982,
+      "step": 4005
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00017665427688190263,
+      "loss": 0.9623,
+      "step": 4006
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00017664306857311917,
+      "loss": 1.1182,
+      "step": 4007
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00017663185793017307,
+      "loss": 1.0286,
+      "step": 4008
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00017662064495340576,
+      "loss": 0.9265,
+      "step": 4009
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00017660942964315869,
+      "loss": 0.9916,
+      "step": 4010
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00017659821199977345,
+      "loss": 1.0274,
+      "step": 4011
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00017658699202359172,
+      "loss": 0.9152,
+      "step": 4012
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00017657576971495513,
+      "loss": 0.8001,
+      "step": 4013
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00017656454507420546,
+      "loss": 1.0342,
+      "step": 4014
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00017655331810168462,
+      "loss": 0.9589,
+      "step": 4015
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00017654208879773443,
+      "loss": 1.1064,
+      "step": 4016
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.000176530857162697,
+      "loss": 0.856,
+      "step": 4017
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.0001765196231969143,
+      "loss": 0.8979,
+      "step": 4018
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00017650838690072847,
+      "loss": 0.9084,
+      "step": 4019
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00017649714827448174,
+      "loss": 1.0358,
+      "step": 4020
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00017648590731851637,
+      "loss": 0.9725,
+      "step": 4021
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00017647466403317467,
+      "loss": 0.9389,
+      "step": 4022
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00017646341841879911,
+      "loss": 0.8947,
+      "step": 4023
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00017645217047573215,
+      "loss": 0.8862,
+      "step": 4024
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00017644092020431632,
+      "loss": 0.872,
+      "step": 4025
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.0001764296676048943,
+      "loss": 0.9134,
+      "step": 4026
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00017641841267780877,
+      "loss": 0.9934,
+      "step": 4027
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00017640715542340245,
+      "loss": 0.9919,
+      "step": 4028
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00017639589584201825,
+      "loss": 1.0758,
+      "step": 4029
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00017638463393399901,
+      "loss": 0.9655,
+      "step": 4030
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.0001763733696996878,
+      "loss": 0.9916,
+      "step": 4031
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.00017636210313942758,
+      "loss": 1.1009,
+      "step": 4032
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.00017635083425356153,
+      "loss": 0.8042,
+      "step": 4033
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.0001763395630424328,
+      "loss": 0.9405,
+      "step": 4034
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.00017632828950638472,
+      "loss": 0.9717,
+      "step": 4035
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.00017631701364576057,
+      "loss": 1.0353,
+      "step": 4036
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.00017630573546090377,
+      "loss": 0.8882,
+      "step": 4037
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.0001762944549521578,
+      "loss": 0.9442,
+      "step": 4038
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.0001762831721198662,
+      "loss": 0.9466,
+      "step": 4039
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.0001762718869643726,
+      "loss": 0.9091,
+      "step": 4040
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.0001762605994860207,
+      "loss": 0.982,
+      "step": 4041
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.00017624930968515422,
+      "loss": 0.8357,
+      "step": 4042
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.00017623801756211702,
+      "loss": 0.9872,
+      "step": 4043
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.000176226723117253,
+      "loss": 0.9427,
+      "step": 4044
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.00017621542635090612,
+      "loss": 1.0244,
+      "step": 4045
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.00017620412726342046,
+      "loss": 0.8943,
+      "step": 4046
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.0001761928258551401,
+      "loss": 0.913,
+      "step": 4047
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.0001761815221264092,
+      "loss": 0.9524,
+      "step": 4048
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.00017617021607757206,
+      "loss": 0.9263,
+      "step": 4049
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.00017615890770897297,
+      "loss": 0.9915,
+      "step": 4050
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.0001761475970209564,
+      "loss": 0.9603,
+      "step": 4051
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.00017613628401386667,
+      "loss": 0.963,
+      "step": 4052
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.0001761249686880485,
+      "loss": 0.9618,
+      "step": 4053
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.00017611365104384634,
+      "loss": 1.0047,
+      "step": 4054
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.00017610233108160498,
+      "loss": 1.0316,
+      "step": 4055
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.00017609100880166907,
+      "loss": 0.8897,
+      "step": 4056
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.00017607968420438352,
+      "loss": 0.9183,
+      "step": 4057
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.00017606835729009322,
+      "loss": 0.9555,
+      "step": 4058
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.00017605702805914306,
+      "loss": 0.9656,
+      "step": 4059
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.0001760456965118781,
+      "loss": 1.0842,
+      "step": 4060
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.00017603436264864347,
+      "loss": 0.8449,
+      "step": 4061
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.00017602302646978427,
+      "loss": 0.8974,
+      "step": 4062
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.00017601168797564583,
+      "loss": 0.9172,
+      "step": 4063
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.00017600034716657342,
+      "loss": 0.7991,
+      "step": 4064
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.00017598900404291245,
+      "loss": 0.9299,
+      "step": 4065
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.00017597765860500838,
+      "loss": 0.9678,
+      "step": 4066
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.00017596631085320668,
+      "loss": 1.0014,
+      "step": 4067
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.00017595496078785295,
+      "loss": 1.1735,
+      "step": 4068
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.0001759436084092929,
+      "loss": 1.0027,
+      "step": 4069
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.00017593225371787224,
+      "loss": 0.7985,
+      "step": 4070
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.00017592089671393684,
+      "loss": 1.0709,
+      "step": 4071
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.00017590953739783248,
+      "loss": 0.8418,
+      "step": 4072
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.00017589817576990516,
+      "loss": 1.0013,
+      "step": 4073
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.00017588681183050087,
+      "loss": 0.9751,
+      "step": 4074
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.00017587544557996573,
+      "loss": 1.0283,
+      "step": 4075
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00017586407701864588,
+      "loss": 1.0537,
+      "step": 4076
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00017585270614688757,
+      "loss": 0.9685,
+      "step": 4077
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.0001758413329650371,
+      "loss": 0.9231,
+      "step": 4078
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00017582995747344077,
+      "loss": 0.9152,
+      "step": 4079
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00017581857967244514,
+      "loss": 0.9652,
+      "step": 4080
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00017580719956239663,
+      "loss": 1.0342,
+      "step": 4081
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00017579581714364183,
+      "loss": 0.8858,
+      "step": 4082
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00017578443241652743,
+      "loss": 0.8264,
+      "step": 4083
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.0001757730453814001,
+      "loss": 0.9928,
+      "step": 4084
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.0001757616560386067,
+      "loss": 0.968,
+      "step": 4085
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00017575026438849402,
+      "loss": 0.9761,
+      "step": 4086
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00017573887043140907,
+      "loss": 0.8206,
+      "step": 4087
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00017572747416769876,
+      "loss": 0.8446,
+      "step": 4088
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00017571607559771024,
+      "loss": 0.7684,
+      "step": 4089
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00017570467472179062,
+      "loss": 1.0795,
+      "step": 4090
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00017569327154028712,
+      "loss": 0.9684,
+      "step": 4091
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.000175681866053547,
+      "loss": 0.9275,
+      "step": 4092
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00017567045826191767,
+      "loss": 1.0101,
+      "step": 4093
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00017565904816574653,
+      "loss": 1.0311,
+      "step": 4094
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00017564763576538104,
+      "loss": 0.8849,
+      "step": 4095
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00017563622106116878,
+      "loss": 0.8194,
+      "step": 4096
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00017562480405345737,
+      "loss": 1.1313,
+      "step": 4097
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00017561338474259458,
+      "loss": 0.8723,
+      "step": 4098
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00017560196312892815,
+      "loss": 0.9354,
+      "step": 4099
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00017559053921280588,
+      "loss": 0.9292,
+      "step": 4100
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00017557911299457575,
+      "loss": 0.9929,
+      "step": 4101
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.0001755676844745857,
+      "loss": 1.0594,
+      "step": 4102
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.0001755562536531838,
+      "loss": 1.0932,
+      "step": 4103
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00017554482053071818,
+      "loss": 1.0562,
+      "step": 4104
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00017553338510753705,
+      "loss": 0.8865,
+      "step": 4105
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00017552194738398863,
+      "loss": 0.963,
+      "step": 4106
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00017551050736042134,
+      "loss": 0.9572,
+      "step": 4107
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.0001754990650371835,
+      "loss": 1.0094,
+      "step": 4108
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.0001754876204146236,
+      "loss": 0.9719,
+      "step": 4109
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.0001754761734930902,
+      "loss": 0.938,
+      "step": 4110
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00017546472427293192,
+      "loss": 0.9171,
+      "step": 4111
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00017545327275449748,
+      "loss": 0.9301,
+      "step": 4112
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00017544181893813557,
+      "loss": 0.9707,
+      "step": 4113
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00017543036282419502,
+      "loss": 1.0301,
+      "step": 4114
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00017541890441302477,
+      "loss": 0.9988,
+      "step": 4115
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.0001754074437049738,
+      "loss": 0.9501,
+      "step": 4116
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00017539598070039108,
+      "loss": 0.9486,
+      "step": 4117
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00017538451539962573,
+      "loss": 1.0065,
+      "step": 4118
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00017537304780302696,
+      "loss": 1.0659,
+      "step": 4119
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.000175361577910944,
+      "loss": 0.9077,
+      "step": 4120
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00017535010572372617,
+      "loss": 1.0285,
+      "step": 4121
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00017533863124172285,
+      "loss": 0.9278,
+      "step": 4122
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00017532715446528347,
+      "loss": 1.061,
+      "step": 4123
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.0001753156753947576,
+      "loss": 0.9754,
+      "step": 4124
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00017530419403049478,
+      "loss": 1.0479,
+      "step": 4125
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00017529271037284474,
+      "loss": 0.9816,
+      "step": 4126
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00017528122442215719,
+      "loss": 0.9298,
+      "step": 4127
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.0001752697361787819,
+      "loss": 1.0366,
+      "step": 4128
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00017525824564306876,
+      "loss": 0.865,
+      "step": 4129
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00017524675281536777,
+      "loss": 0.9012,
+      "step": 4130
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00017523525769602883,
+      "loss": 0.9168,
+      "step": 4131
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00017522376028540214,
+      "loss": 1.0599,
+      "step": 4132
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00017521226058383782,
+      "loss": 1.0548,
+      "step": 4133
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00017520075859168603,
+      "loss": 0.9954,
+      "step": 4134
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00017518925430929714,
+      "loss": 0.8689,
+      "step": 4135
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00017517774773702146,
+      "loss": 0.9599,
+      "step": 4136
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00017516623887520947,
+      "loss": 0.9727,
+      "step": 4137
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00017515472772421163,
+      "loss": 0.8413,
+      "step": 4138
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00017514321428437857,
+      "loss": 0.8427,
+      "step": 4139
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00017513169855606087,
+      "loss": 0.9482,
+      "step": 4140
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00017512018053960927,
+      "loss": 0.9008,
+      "step": 4141
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.0001751086602353745,
+      "loss": 0.9937,
+      "step": 4142
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00017509713764370747,
+      "loss": 0.9253,
+      "step": 4143
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00017508561276495912,
+      "loss": 0.9669,
+      "step": 4144
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00017507408559948038,
+      "loss": 0.9046,
+      "step": 4145
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00017506255614762234,
+      "loss": 1.0331,
+      "step": 4146
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.0001750510244097361,
+      "loss": 0.8915,
+      "step": 4147
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.0001750394903861729,
+      "loss": 0.8768,
+      "step": 4148
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.000175027954077284,
+      "loss": 1.0605,
+      "step": 4149
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.0001750164154834207,
+      "loss": 0.9542,
+      "step": 4150
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00017500487460493446,
+      "loss": 0.9214,
+      "step": 4151
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00017499333144217673,
+      "loss": 1.0135,
+      "step": 4152
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00017498178599549907,
+      "loss": 0.955,
+      "step": 4153
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00017497023826525305,
+      "loss": 0.9674,
+      "step": 4154
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.0001749586882517904,
+      "loss": 1.0081,
+      "step": 4155
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00017494713595546287,
+      "loss": 1.0148,
+      "step": 4156
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.0001749355813766223,
+      "loss": 1.1226,
+      "step": 4157
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00017492402451562057,
+      "loss": 0.9435,
+      "step": 4158
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00017491246537280962,
+      "loss": 0.9002,
+      "step": 4159
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.0001749009039485415,
+      "loss": 1.0582,
+      "step": 4160
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00017488934024316831,
+      "loss": 0.8601,
+      "step": 4161
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00017487777425704225,
+      "loss": 0.9301,
+      "step": 4162
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00017486620599051555,
+      "loss": 0.9243,
+      "step": 4163
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00017485463544394047,
+      "loss": 0.9411,
+      "step": 4164
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00017484306261766945,
+      "loss": 0.8645,
+      "step": 4165
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00017483148751205497,
+      "loss": 0.9397,
+      "step": 4166
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00017481991012744942,
+      "loss": 0.9609,
+      "step": 4167
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00017480833046420552,
+      "loss": 1.0506,
+      "step": 4168
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00017479674852267587,
+      "loss": 1.0439,
+      "step": 4169
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00017478516430321323,
+      "loss": 0.9251,
+      "step": 4170
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00017477357780617032,
+      "loss": 0.9218,
+      "step": 4171
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.0001747619890319001,
+      "loss": 1.0396,
+      "step": 4172
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00017475039798075547,
+      "loss": 0.9849,
+      "step": 4173
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00017473880465308943,
+      "loss": 0.9739,
+      "step": 4174
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00017472720904925502,
+      "loss": 0.9277,
+      "step": 4175
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00017471561116960547,
+      "loss": 0.9663,
+      "step": 4176
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00017470401101449392,
+      "loss": 0.8781,
+      "step": 4177
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00017469240858427365,
+      "loss": 0.9268,
+      "step": 4178
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00017468080387929806,
+      "loss": 1.0767,
+      "step": 4179
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00017466919689992057,
+      "loss": 0.9001,
+      "step": 4180
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.0001746575876464946,
+      "loss": 0.9422,
+      "step": 4181
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.0001746459761193738,
+      "loss": 1.013,
+      "step": 4182
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00017463436231891173,
+      "loss": 0.9715,
+      "step": 4183
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.0001746227462454621,
+      "loss": 0.9497,
+      "step": 4184
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00017461112789937875,
+      "loss": 0.8965,
+      "step": 4185
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00017459950728101543,
+      "loss": 0.8856,
+      "step": 4186
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00017458788439072607,
+      "loss": 0.969,
+      "step": 4187
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00017457625922886463,
+      "loss": 1.0753,
+      "step": 4188
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.0001745646317957852,
+      "loss": 0.9267,
+      "step": 4189
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00017455300209184187,
+      "loss": 0.9991,
+      "step": 4190
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00017454137011738879,
+      "loss": 1.066,
+      "step": 4191
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00017452973587278025,
+      "loss": 0.9171,
+      "step": 4192
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00017451809935837057,
+      "loss": 0.8958,
+      "step": 4193
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.0001745064605745141,
+      "loss": 1.0152,
+      "step": 4194
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00017449481952156536,
+      "loss": 0.9556,
+      "step": 4195
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00017448317619987887,
+      "loss": 1.0076,
+      "step": 4196
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00017447153060980917,
+      "loss": 0.9348,
+      "step": 4197
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00017445988275171094,
+      "loss": 0.9108,
+      "step": 4198
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00017444823262593895,
+      "loss": 1.0939,
+      "step": 4199
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.000174436580232848,
+      "loss": 0.8934,
+      "step": 4200
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00017442492557279295,
+      "loss": 1.0552,
+      "step": 4201
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00017441326864612875,
+      "loss": 0.9994,
+      "step": 4202
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.0001744016094532104,
+      "loss": 0.9695,
+      "step": 4203
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.000174389947994393,
+      "loss": 0.9346,
+      "step": 4204
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00017437828427003166,
+      "loss": 0.9224,
+      "step": 4205
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00017436661828048165,
+      "loss": 1.058,
+      "step": 4206
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.0001743549500260982,
+      "loss": 0.911,
+      "step": 4207
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00017434327950723675,
+      "loss": 1.01,
+      "step": 4208
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.0001743316067242526,
+      "loss": 0.9776,
+      "step": 4209
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00017431993167750138,
+      "loss": 0.9411,
+      "step": 4210
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00017430825436733858,
+      "loss": 0.9595,
+      "step": 4211
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.00017429657479411983,
+      "loss": 1.0409,
+      "step": 4212
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.00017428489295820084,
+      "loss": 0.9731,
+      "step": 4213
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.00017427320885993742,
+      "loss": 0.9785,
+      "step": 4214
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.00017426152249968535,
+      "loss": 0.9351,
+      "step": 4215
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.00017424983387780057,
+      "loss": 0.9422,
+      "step": 4216
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.000174238142994639,
+      "loss": 1.0302,
+      "step": 4217
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.0001742264498505568,
+      "loss": 0.931,
+      "step": 4218
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.00017421475444590997,
+      "loss": 0.9112,
+      "step": 4219
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.00017420305678105476,
+      "loss": 1.0466,
+      "step": 4220
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.0001741913568563474,
+      "loss": 1.063,
+      "step": 4221
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.00017417965467214424,
+      "loss": 0.9161,
+      "step": 4222
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.00017416795022880162,
+      "loss": 0.9503,
+      "step": 4223
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.00017415624352667603,
+      "loss": 0.9915,
+      "step": 4224
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.000174144534566124,
+      "loss": 0.903,
+      "step": 4225
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.0001741328233475021,
+      "loss": 0.9295,
+      "step": 4226
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.000174121109871167,
+      "loss": 0.9484,
+      "step": 4227
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.00017410939413747543,
+      "loss": 1.0536,
+      "step": 4228
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.00017409767614678424,
+      "loss": 1.0457,
+      "step": 4229
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.00017408595589945026,
+      "loss": 0.9628,
+      "step": 4230
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.00017407423339583047,
+      "loss": 1.075,
+      "step": 4231
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.0001740625086362818,
+      "loss": 0.9557,
+      "step": 4232
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.0001740507816211614,
+      "loss": 0.9687,
+      "step": 4233
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.00017403905235082636,
+      "loss": 0.9506,
+      "step": 4234
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.00017402732082563396,
+      "loss": 1.0387,
+      "step": 4235
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.00017401558704594142,
+      "loss": 1.0465,
+      "step": 4236
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.00017400385101210613,
+      "loss": 0.9187,
+      "step": 4237
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.0001739921127244855,
+      "loss": 0.9208,
+      "step": 4238
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.00017398037218343702,
+      "loss": 0.9026,
+      "step": 4239
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.0001739686293893182,
+      "loss": 0.9068,
+      "step": 4240
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.00017395688434248679,
+      "loss": 0.9903,
+      "step": 4241
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.00017394513704330038,
+      "loss": 0.8842,
+      "step": 4242
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.00017393338749211674,
+      "loss": 1.0244,
+      "step": 4243
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.00017392163568929374,
+      "loss": 1.0164,
+      "step": 4244
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.00017390988163518926,
+      "loss": 0.9028,
+      "step": 4245
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.0001738981253301613,
+      "loss": 0.904,
+      "step": 4246
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.00017388636677456785,
+      "loss": 0.9567,
+      "step": 4247
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.00017387460596876702,
+      "loss": 0.9645,
+      "step": 4248
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.00017386284291311704,
+      "loss": 0.9495,
+      "step": 4249
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.00017385107760797613,
+      "loss": 1.0191,
+      "step": 4250
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.00017383931005370253,
+      "loss": 1.0401,
+      "step": 4251
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.00017382754025065472,
+      "loss": 0.9926,
+      "step": 4252
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.00017381576819919112,
+      "loss": 0.9541,
+      "step": 4253
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.00017380399389967024,
+      "loss": 0.9845,
+      "step": 4254
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.0001737922173524506,
+      "loss": 1.0822,
+      "step": 4255
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.000173780438557891,
+      "loss": 0.9722,
+      "step": 4256
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.00017376865751635002,
+      "loss": 0.8709,
+      "step": 4257
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.00017375687422818654,
+      "loss": 0.9735,
+      "step": 4258
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.0001737450886937594,
+      "loss": 0.8537,
+      "step": 4259
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.00017373330091342747,
+      "loss": 1.0449,
+      "step": 4260
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.00017372151088754985,
+      "loss": 0.8554,
+      "step": 4261
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.00017370971861648552,
+      "loss": 1.0799,
+      "step": 4262
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.00017369792410059365,
+      "loss": 0.9422,
+      "step": 4263
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.00017368612734023343,
+      "loss": 0.8965,
+      "step": 4264
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.00017367432833576417,
+      "loss": 0.9715,
+      "step": 4265
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.00017366252708754515,
+      "loss": 0.7877,
+      "step": 4266
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.0001736507235959358,
+      "loss": 0.9784,
+      "step": 4267
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.00017363891786129562,
+      "loss": 0.88,
+      "step": 4268
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.00017362710988398408,
+      "loss": 0.8844,
+      "step": 4269
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.00017361529966436088,
+      "loss": 0.9509,
+      "step": 4270
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.00017360348720278564,
+      "loss": 0.9907,
+      "step": 4271
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.00017359167249961815,
+      "loss": 0.9633,
+      "step": 4272
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.00017357985555521818,
+      "loss": 1.0383,
+      "step": 4273
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.00017356803636994566,
+      "loss": 0.8815,
+      "step": 4274
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.0001735562149441605,
+      "loss": 0.9973,
+      "step": 4275
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.00017354439127822277,
+      "loss": 0.9789,
+      "step": 4276
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.00017353256537249253,
+      "loss": 0.899,
+      "step": 4277
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.00017352073722732994,
+      "loss": 0.8391,
+      "step": 4278
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.00017350890684309522,
+      "loss": 1.004,
+      "step": 4279
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.00017349707422014866,
+      "loss": 0.9319,
+      "step": 4280
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.0001734852393588507,
+      "loss": 1.0247,
+      "step": 4281
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.00017347340225956162,
+      "loss": 1.0276,
+      "step": 4282
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.00017346156292264204,
+      "loss": 1.0137,
+      "step": 4283
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.00017344972134845249,
+      "loss": 0.8879,
+      "step": 4284
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.0001734378775373536,
+      "loss": 0.963,
+      "step": 4285
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.0001734260314897061,
+      "loss": 0.9248,
+      "step": 4286
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.00017341418320587068,
+      "loss": 1.0399,
+      "step": 4287
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.0001734023326862083,
+      "loss": 0.9373,
+      "step": 4288
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.00017339047993107976,
+      "loss": 0.9394,
+      "step": 4289
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.00017337862494084608,
+      "loss": 0.9021,
+      "step": 4290
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.00017336676771586836,
+      "loss": 1.004,
+      "step": 4291
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.00017335490825650757,
+      "loss": 1.0588,
+      "step": 4292
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.00017334304656312503,
+      "loss": 1.0359,
+      "step": 4293
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.00017333118263608191,
+      "loss": 0.8176,
+      "step": 4294
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.0001733193164757396,
+      "loss": 0.9466,
+      "step": 4295
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.0001733074480824594,
+      "loss": 0.8748,
+      "step": 4296
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.00017329557745660273,
+      "loss": 0.9445,
+      "step": 4297
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.00017328370459853125,
+      "loss": 0.9585,
+      "step": 4298
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.00017327182950860643,
+      "loss": 1.0887,
+      "step": 4299
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.00017325995218719,
+      "loss": 1.0542,
+      "step": 4300
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.0001732480726346436,
+      "loss": 0.9036,
+      "step": 4301
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.0001732361908513291,
+      "loss": 0.9596,
+      "step": 4302
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.00017322430683760836,
+      "loss": 0.9294,
+      "step": 4303
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.00017321242059384324,
+      "loss": 0.9982,
+      "step": 4304
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.00017320053212039577,
+      "loss": 1.0988,
+      "step": 4305
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.00017318864141762803,
+      "loss": 0.8878,
+      "step": 4306
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.00017317674848590212,
+      "loss": 0.9146,
+      "step": 4307
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.00017316485332558027,
+      "loss": 0.9763,
+      "step": 4308
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.00017315295593702473,
+      "loss": 1.0391,
+      "step": 4309
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.00017314105632059782,
+      "loss": 0.9676,
+      "step": 4310
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.00017312915447666201,
+      "loss": 1.0322,
+      "step": 4311
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.00017311725040557966,
+      "loss": 1.0025,
+      "step": 4312
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.00017310534410771338,
+      "loss": 0.851,
+      "step": 4313
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.0001730934355834258,
+      "loss": 0.9338,
+      "step": 4314
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.00017308152483307953,
+      "loss": 0.9192,
+      "step": 4315
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.00017306961185703737,
+      "loss": 0.9599,
+      "step": 4316
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.00017305769665566204,
+      "loss": 1.0615,
+      "step": 4317
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.00017304577922931652,
+      "loss": 0.9316,
+      "step": 4318
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.00017303385957836368,
+      "loss": 0.9444,
+      "step": 4319
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.0001730219377031666,
+      "loss": 0.9692,
+      "step": 4320
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.0001730100136040883,
+      "loss": 0.9207,
+      "step": 4321
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.00017299808728149198,
+      "loss": 0.9029,
+      "step": 4322
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.0001729861587357408,
+      "loss": 0.8376,
+      "step": 4323
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.00017297422796719807,
+      "loss": 0.8972,
+      "step": 4324
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.00017296229497622715,
+      "loss": 0.8963,
+      "step": 4325
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.00017295035976319146,
+      "loss": 1.023,
+      "step": 4326
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.00017293842232845447,
+      "loss": 0.85,
+      "step": 4327
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.00017292648267237974,
+      "loss": 0.8157,
+      "step": 4328
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.00017291454079533088,
+      "loss": 0.922,
+      "step": 4329
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.0001729025966976716,
+      "loss": 0.9651,
+      "step": 4330
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.00017289065037976567,
+      "loss": 0.9245,
+      "step": 4331
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.0001728787018419769,
+      "loss": 0.8441,
+      "step": 4332
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.00017286675108466916,
+      "loss": 1.1064,
+      "step": 4333
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.00017285479810820643,
+      "loss": 0.9253,
+      "step": 4334
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.00017284284291295277,
+      "loss": 1.0011,
+      "step": 4335
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.00017283088549927217,
+      "loss": 0.9204,
+      "step": 4336
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.00017281892586752892,
+      "loss": 0.8517,
+      "step": 4337
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.0001728069640180872,
+      "loss": 1.1415,
+      "step": 4338
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.00017279499995131125,
+      "loss": 1.0398,
+      "step": 4339
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.00017278303366756554,
+      "loss": 0.9021,
+      "step": 4340
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.00017277106516721443,
+      "loss": 0.9603,
+      "step": 4341
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.0001727590944506224,
+      "loss": 0.8817,
+      "step": 4342
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.0001727471215181541,
+      "loss": 0.9249,
+      "step": 4343
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.00017273514637017415,
+      "loss": 0.964,
+      "step": 4344
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.0001727231690070472,
+      "loss": 1.085,
+      "step": 4345
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.00017271118942913805,
+      "loss": 0.9522,
+      "step": 4346
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.0001726992076368115,
+      "loss": 0.8965,
+      "step": 4347
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.0001726872236304325,
+      "loss": 1.0424,
+      "step": 4348
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00017267523741036603,
+      "loss": 0.9661,
+      "step": 4349
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00017266324897697715,
+      "loss": 0.9596,
+      "step": 4350
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00017265125833063088,
+      "loss": 1.0008,
+      "step": 4351
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00017263926547169244,
+      "loss": 0.9859,
+      "step": 4352
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.0001726272704005271,
+      "loss": 0.9283,
+      "step": 4353
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00017261527311750015,
+      "loss": 0.948,
+      "step": 4354
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.0001726032736229769,
+      "loss": 0.8878,
+      "step": 4355
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00017259127191732293,
+      "loss": 0.9397,
+      "step": 4356
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00017257926800090368,
+      "loss": 0.9754,
+      "step": 4357
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.0001725672618740847,
+      "loss": 1.0601,
+      "step": 4358
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00017255525353723165,
+      "loss": 0.9498,
+      "step": 4359
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.0001725432429907103,
+      "loss": 0.9096,
+      "step": 4360
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00017253123023488635,
+      "loss": 1.0664,
+      "step": 4361
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.0001725192152701257,
+      "loss": 0.8626,
+      "step": 4362
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00017250719809679428,
+      "loss": 1.0464,
+      "step": 4363
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.000172495178715258,
+      "loss": 0.9824,
+      "step": 4364
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00017248315712588296,
+      "loss": 1.0076,
+      "step": 4365
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00017247113332903528,
+      "loss": 0.8729,
+      "step": 4366
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00017245910732508114,
+      "loss": 0.9232,
+      "step": 4367
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.0001724470791143868,
+      "loss": 0.9567,
+      "step": 4368
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00017243504869731856,
+      "loss": 0.9013,
+      "step": 4369
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00017242301607424282,
+      "loss": 0.9739,
+      "step": 4370
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.000172410981245526,
+      "loss": 0.881,
+      "step": 4371
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.0001723989442115347,
+      "loss": 1.0425,
+      "step": 4372
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00017238690497263542,
+      "loss": 0.9235,
+      "step": 4373
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00017237486352919486,
+      "loss": 1.0451,
+      "step": 4374
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.0001723628198815797,
+      "loss": 1.0311,
+      "step": 4375
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00017235077403015677,
+      "loss": 1.0846,
+      "step": 4376
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00017233872597529292,
+      "loss": 1.0267,
+      "step": 4377
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00017232667571735508,
+      "loss": 0.8826,
+      "step": 4378
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00017231462325671023,
+      "loss": 0.8976,
+      "step": 4379
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00017230256859372543,
+      "loss": 1.0041,
+      "step": 4380
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00017229051172876778,
+      "loss": 0.9408,
+      "step": 4381
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.0001722784526622045,
+      "loss": 0.8834,
+      "step": 4382
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00017226639139440283,
+      "loss": 0.8647,
+      "step": 4383
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00017225432792573013,
+      "loss": 1.1379,
+      "step": 4384
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00017224226225655376,
+      "loss": 0.9385,
+      "step": 4385
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.0001722301943872412,
+      "loss": 1.0114,
+      "step": 4386
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00017221812431815993,
+      "loss": 0.9649,
+      "step": 4387
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00017220605204967764,
+      "loss": 0.8651,
+      "step": 4388
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.0001721939775821619,
+      "loss": 1.1144,
+      "step": 4389
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00017218190091598045,
+      "loss": 1.0143,
+      "step": 4390
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00017216982205150113,
+      "loss": 1.0865,
+      "step": 4391
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.0001721577409890917,
+      "loss": 0.9609,
+      "step": 4392
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.00017214565772912026,
+      "loss": 0.9974,
+      "step": 4393
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.00017213357227195466,
+      "loss": 1.1351,
+      "step": 4394
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.00017212148461796302,
+      "loss": 1.0509,
+      "step": 4395
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.00017210939476751345,
+      "loss": 0.9363,
+      "step": 4396
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.00017209730272097417,
+      "loss": 0.912,
+      "step": 4397
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.00017208520847871342,
+      "loss": 1.0008,
+      "step": 4398
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.00017207311204109953,
+      "loss": 0.8542,
+      "step": 4399
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.00017206101340850092,
+      "loss": 0.9343,
+      "step": 4400
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.00017204891258128604,
+      "loss": 0.9864,
+      "step": 4401
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.0001720368095598234,
+      "loss": 0.8981,
+      "step": 4402
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.00017202470434448164,
+      "loss": 0.9239,
+      "step": 4403
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.00017201259693562935,
+      "loss": 0.8556,
+      "step": 4404
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.00017200048733363534,
+      "loss": 1.1088,
+      "step": 4405
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.0001719883755388684,
+      "loss": 1.0062,
+      "step": 4406
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.00017197626155169738,
+      "loss": 0.966,
+      "step": 4407
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.00017196414537249115,
+      "loss": 0.9446,
+      "step": 4408
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.00017195202700161878,
+      "loss": 1.0227,
+      "step": 4409
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.00017193990643944935,
+      "loss": 0.8945,
+      "step": 4410
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.00017192778368635192,
+      "loss": 1.0442,
+      "step": 4411
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.00017191565874269572,
+      "loss": 1.0464,
+      "step": 4412
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.00017190353160885006,
+      "loss": 1.0376,
+      "step": 4413
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.00017189140228518422,
+      "loss": 0.9919,
+      "step": 4414
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.00017187927077206758,
+      "loss": 1.0186,
+      "step": 4415
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.00017186713706986964,
+      "loss": 1.0275,
+      "step": 4416
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.00017185500117895995,
+      "loss": 0.8955,
+      "step": 4417
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.0001718428630997081,
+      "loss": 1.0019,
+      "step": 4418
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.00017183072283248371,
+      "loss": 0.9871,
+      "step": 4419
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.00017181858037765654,
+      "loss": 0.9003,
+      "step": 4420
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.0001718064357355964,
+      "loss": 0.8859,
+      "step": 4421
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.00017179428890667318,
+      "loss": 1.111,
+      "step": 4422
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.00017178213989125672,
+      "loss": 1.0633,
+      "step": 4423
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.0001717699886897171,
+      "loss": 0.9654,
+      "step": 4424
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.00017175783530242432,
+      "loss": 0.9807,
+      "step": 4425
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.00017174567972974862,
+      "loss": 0.888,
+      "step": 4426
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.00017173352197206006,
+      "loss": 0.9875,
+      "step": 4427
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.000171721362029729,
+      "loss": 0.9514,
+      "step": 4428
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.00017170919990312574,
+      "loss": 0.9649,
+      "step": 4429
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.00017169703559262068,
+      "loss": 1.0029,
+      "step": 4430
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.0001716848690985843,
+      "loss": 0.9459,
+      "step": 4431
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.0001716727004213871,
+      "loss": 1.1021,
+      "step": 4432
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.0001716605295613997,
+      "loss": 0.8899,
+      "step": 4433
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.0001716483565189927,
+      "loss": 0.9989,
+      "step": 4434
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.00017163618129453694,
+      "loss": 1.0164,
+      "step": 4435
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.00017162400388840313,
+      "loss": 0.95,
+      "step": 4436
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.00017161182430096216,
+      "loss": 0.9139,
+      "step": 4437
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.00017159964253258498,
+      "loss": 0.9334,
+      "step": 4438
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.00017158745858364256,
+      "loss": 0.9559,
+      "step": 4439
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.000171575272454506,
+      "loss": 0.9135,
+      "step": 4440
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.0001715630841455463,
+      "loss": 0.8595,
+      "step": 4441
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.00017155089365713483,
+      "loss": 0.9392,
+      "step": 4442
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.00017153870098964277,
+      "loss": 0.8969,
+      "step": 4443
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.00017152650614344142,
+      "loss": 1.0602,
+      "step": 4444
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.00017151430911890224,
+      "loss": 0.8414,
+      "step": 4445
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.00017150210991639662,
+      "loss": 1.0434,
+      "step": 4446
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.00017148990853629612,
+      "loss": 0.9968,
+      "step": 4447
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.00017147770497897232,
+      "loss": 0.8316,
+      "step": 4448
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.00017146549924479694,
+      "loss": 1.0504,
+      "step": 4449
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.0001714532913341416,
+      "loss": 0.9093,
+      "step": 4450
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.0001714410812473782,
+      "loss": 1.0416,
+      "step": 4451
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.00017142886898487852,
+      "loss": 0.7567,
+      "step": 4452
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.0001714166545470145,
+      "loss": 1.0492,
+      "step": 4453
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.00017140443793415815,
+      "loss": 1.0146,
+      "step": 4454
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.00017139221914668152,
+      "loss": 1.0767,
+      "step": 4455
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.00017137999818495674,
+      "loss": 0.955,
+      "step": 4456
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.00017136777504935595,
+      "loss": 0.9418,
+      "step": 4457
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.00017135554974025146,
+      "loss": 0.8553,
+      "step": 4458
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.0001713433222580156,
+      "loss": 0.9837,
+      "step": 4459
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.00017133109260302073,
+      "loss": 0.8959,
+      "step": 4460
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.00017131886077563927,
+      "loss": 0.9901,
+      "step": 4461
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.0001713066267762438,
+      "loss": 1.0063,
+      "step": 4462
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.0001712943906052069,
+      "loss": 0.8398,
+      "step": 4463
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.0001712821522629012,
+      "loss": 0.8137,
+      "step": 4464
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.00017126991174969938,
+      "loss": 0.8841,
+      "step": 4465
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.00017125766906597433,
+      "loss": 0.9703,
+      "step": 4466
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.00017124542421209882,
+      "loss": 0.8974,
+      "step": 4467
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.00017123317718844577,
+      "loss": 0.9602,
+      "step": 4468
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.0001712209279953882,
+      "loss": 0.9603,
+      "step": 4469
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.00017120867663329912,
+      "loss": 1.0492,
+      "step": 4470
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.00017119642310255167,
+      "loss": 0.9507,
+      "step": 4471
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.000171184167403519,
+      "loss": 1.0367,
+      "step": 4472
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.00017117190953657438,
+      "loss": 0.9741,
+      "step": 4473
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.00017115964950209115,
+      "loss": 1.0576,
+      "step": 4474
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.00017114738730044266,
+      "loss": 1.091,
+      "step": 4475
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.00017113512293200235,
+      "loss": 0.9718,
+      "step": 4476
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.00017112285639714374,
+      "loss": 0.8763,
+      "step": 4477
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.0001711105876962404,
+      "loss": 0.9206,
+      "step": 4478
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.000171098316829666,
+      "loss": 1.0776,
+      "step": 4479
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.00017108604379779418,
+      "loss": 1.0268,
+      "step": 4480
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.00017107376860099878,
+      "loss": 1.0332,
+      "step": 4481
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00017106149123965366,
+      "loss": 0.9623,
+      "step": 4482
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00017104921171413263,
+      "loss": 0.8396,
+      "step": 4483
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00017103693002480975,
+      "loss": 0.8593,
+      "step": 4484
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00017102464617205906,
+      "loss": 0.9957,
+      "step": 4485
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00017101236015625463,
+      "loss": 0.918,
+      "step": 4486
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0001710000719777706,
+      "loss": 0.9842,
+      "step": 4487
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00017098778163698124,
+      "loss": 0.8827,
+      "step": 4488
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0001709754891342609,
+      "loss": 0.9324,
+      "step": 4489
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00017096319446998387,
+      "loss": 0.956,
+      "step": 4490
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0001709508976445246,
+      "loss": 0.87,
+      "step": 4491
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00017093859865825763,
+      "loss": 1.0233,
+      "step": 4492
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0001709262975115575,
+      "loss": 0.9791,
+      "step": 4493
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0001709139942047988,
+      "loss": 0.9873,
+      "step": 4494
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00017090168873835633,
+      "loss": 1.0038,
+      "step": 4495
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00017088938111260474,
+      "loss": 1.0338,
+      "step": 4496
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00017087707132791894,
+      "loss": 0.9596,
+      "step": 4497
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00017086475938467378,
+      "loss": 0.8968,
+      "step": 4498
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00017085244528324423,
+      "loss": 0.8783,
+      "step": 4499
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00017084012902400533,
+      "loss": 1.0171,
+      "step": 4500
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00017082781060733215,
+      "loss": 0.9052,
+      "step": 4501
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00017081549003359988,
+      "loss": 0.9499,
+      "step": 4502
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00017080316730318367,
+      "loss": 1.0719,
+      "step": 4503
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00017079084241645892,
+      "loss": 0.9362,
+      "step": 4504
+    },
+    {
+      "epoch": 1.0,
+      "eval_loss": 0.5923583507537842,
+      "eval_runtime": 322.1825,
+      "eval_samples_per_second": 56.3,
+      "eval_steps_per_second": 28.152,
+      "step": 4504
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00017077851537380086,
+      "loss": 0.8743,
+      "step": 4505
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00017076618617558502,
+      "loss": 0.835,
+      "step": 4506
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0001707538548221868,
+      "loss": 0.9492,
+      "step": 4507
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0001707415213139818,
+      "loss": 0.907,
+      "step": 4508
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00017072918565134564,
+      "loss": 1.0297,
+      "step": 4509
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.000170716847834654,
+      "loss": 0.8893,
+      "step": 4510
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0001707045078642826,
+      "loss": 0.8949,
+      "step": 4511
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00017069216574060726,
+      "loss": 0.849,
+      "step": 4512
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00017067982146400386,
+      "loss": 0.8681,
+      "step": 4513
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00017066747503484834,
+      "loss": 0.863,
+      "step": 4514
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00017065512645351676,
+      "loss": 1.0757,
+      "step": 4515
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00017064277572038513,
+      "loss": 1.0262,
+      "step": 4516
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00017063042283582967,
+      "loss": 0.953,
+      "step": 4517
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0001706180678002265,
+      "loss": 0.8829,
+      "step": 4518
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00017060571061395193,
+      "loss": 0.8862,
+      "step": 4519
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0001705933512773823,
+      "loss": 1.0063,
+      "step": 4520
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00017058098979089398,
+      "loss": 1.0245,
+      "step": 4521
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00017056862615486348,
+      "loss": 0.9483,
+      "step": 4522
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00017055626036966735,
+      "loss": 0.9552,
+      "step": 4523
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00017054389243568215,
+      "loss": 1.1389,
+      "step": 4524
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00017053152235328457,
+      "loss": 0.8752,
+      "step": 4525
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00017051915012285132,
+      "loss": 0.9553,
+      "step": 4526
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00017050677574475918,
+      "loss": 0.9369,
+      "step": 4527
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00017049439921938507,
+      "loss": 1.0688,
+      "step": 4528
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00017048202054710586,
+      "loss": 0.8319,
+      "step": 4529
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00017046963972829857,
+      "loss": 0.8728,
+      "step": 4530
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00017045725676334024,
+      "loss": 1.0214,
+      "step": 4531
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.000170444871652608,
+      "loss": 0.9312,
+      "step": 4532
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00017043248439647907,
+      "loss": 0.887,
+      "step": 4533
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.0001704200949953307,
+      "loss": 0.9366,
+      "step": 4534
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00017040770344954012,
+      "loss": 0.98,
+      "step": 4535
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00017039530975948482,
+      "loss": 1.0605,
+      "step": 4536
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.0001703829139255422,
+      "loss": 0.9384,
+      "step": 4537
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00017037051594808976,
+      "loss": 0.9765,
+      "step": 4538
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00017035811582750515,
+      "loss": 0.9701,
+      "step": 4539
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00017034571356416595,
+      "loss": 0.9308,
+      "step": 4540
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00017033330915844987,
+      "loss": 0.99,
+      "step": 4541
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00017032090261073473,
+      "loss": 0.8937,
+      "step": 4542
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00017030849392139834,
+      "loss": 0.9294,
+      "step": 4543
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00017029608309081858,
+      "loss": 1.0307,
+      "step": 4544
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.0001702836701193735,
+      "loss": 0.9831,
+      "step": 4545
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.0001702712550074411,
+      "loss": 0.9434,
+      "step": 4546
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00017025883775539943,
+      "loss": 0.8932,
+      "step": 4547
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.0001702464183636267,
+      "loss": 0.9592,
+      "step": 4548
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.0001702339968325012,
+      "loss": 0.9217,
+      "step": 4549
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.0001702215731624011,
+      "loss": 1.02,
+      "step": 4550
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.0001702091473537049,
+      "loss": 1.0004,
+      "step": 4551
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00017019671940679093,
+      "loss": 0.9581,
+      "step": 4552
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.0001701842893220377,
+      "loss": 0.8743,
+      "step": 4553
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00017017185709982378,
+      "loss": 0.8922,
+      "step": 4554
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00017015942274052777,
+      "loss": 0.9453,
+      "step": 4555
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00017014698624452843,
+      "loss": 1.0454,
+      "step": 4556
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00017013454761220445,
+      "loss": 0.9581,
+      "step": 4557
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00017012210684393467,
+      "loss": 0.8176,
+      "step": 4558
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00017010966394009796,
+      "loss": 1.1113,
+      "step": 4559
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00017009721890107326,
+      "loss": 0.915,
+      "step": 4560
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00017008477172723963,
+      "loss": 0.9617,
+      "step": 4561
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00017007232241897607,
+      "loss": 0.9784,
+      "step": 4562
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00017005987097666183,
+      "loss": 0.9377,
+      "step": 4563
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.000170047417400676,
+      "loss": 1.011,
+      "step": 4564
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00017003496169139794,
+      "loss": 0.9688,
+      "step": 4565
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00017002250384920696,
+      "loss": 0.9055,
+      "step": 4566
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00017001004387448245,
+      "loss": 0.8667,
+      "step": 4567
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00016999758176760392,
+      "loss": 0.8844,
+      "step": 4568
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00016998511752895087,
+      "loss": 0.9916,
+      "step": 4569
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00016997265115890287,
+      "loss": 0.9619,
+      "step": 4570
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00016996018265783965,
+      "loss": 0.8683,
+      "step": 4571
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.0001699477120261409,
+      "loss": 0.8758,
+      "step": 4572
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.00016993523926418636,
+      "loss": 0.9557,
+      "step": 4573
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00016992276437235602,
+      "loss": 0.9633,
+      "step": 4574
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0001699102873510297,
+      "loss": 0.9115,
+      "step": 4575
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0001698978082005874,
+      "loss": 0.845,
+      "step": 4576
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00016988532692140918,
+      "loss": 1.0184,
+      "step": 4577
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00016987284351387518,
+      "loss": 0.8577,
+      "step": 4578
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00016986035797836557,
+      "loss": 0.846,
+      "step": 4579
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00016984787031526058,
+      "loss": 0.9443,
+      "step": 4580
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00016983538052494054,
+      "loss": 0.9897,
+      "step": 4581
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00016982288860778583,
+      "loss": 0.9705,
+      "step": 4582
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00016981039456417682,
+      "loss": 0.9233,
+      "step": 4583
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00016979789839449412,
+      "loss": 0.8244,
+      "step": 4584
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00016978540009911827,
+      "loss": 0.9605,
+      "step": 4585
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00016977289967842987,
+      "loss": 0.908,
+      "step": 4586
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00016976039713280966,
+      "loss": 0.9135,
+      "step": 4587
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0001697478924626384,
+      "loss": 0.9827,
+      "step": 4588
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00016973538566829685,
+      "loss": 1.0027,
+      "step": 4589
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.000169722876750166,
+      "loss": 0.8497,
+      "step": 4590
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00016971036570862674,
+      "loss": 0.9556,
+      "step": 4591
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00016969785254406015,
+      "loss": 1.0476,
+      "step": 4592
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0001696853372568473,
+      "loss": 0.9725,
+      "step": 4593
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00016967281984736933,
+      "loss": 0.8629,
+      "step": 4594
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0001696603003160074,
+      "loss": 0.7442,
+      "step": 4595
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00016964777866314293,
+      "loss": 0.8593,
+      "step": 4596
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00016963525488915716,
+      "loss": 0.9362,
+      "step": 4597
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00016962272899443152,
+      "loss": 0.873,
+      "step": 4598
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00016961020097934747,
+      "loss": 0.8823,
+      "step": 4599
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00016959767084428662,
+      "loss": 0.8969,
+      "step": 4600
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00016958513858963053,
+      "loss": 0.8789,
+      "step": 4601
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00016957260421576086,
+      "loss": 0.9027,
+      "step": 4602
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00016956006772305936,
+      "loss": 0.9824,
+      "step": 4603
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00016954752911190783,
+      "loss": 0.9995,
+      "step": 4604
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00016953498838268808,
+      "loss": 0.9908,
+      "step": 4605
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00016952244553578214,
+      "loss": 0.9151,
+      "step": 4606
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.0001695099005715719,
+      "loss": 0.9277,
+      "step": 4607
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.0001694973534904395,
+      "loss": 1.0287,
+      "step": 4608
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00016948480429276704,
+      "loss": 0.9752,
+      "step": 4609
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00016947225297893662,
+      "loss": 0.8142,
+      "step": 4610
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.0001694596995493306,
+      "loss": 0.7425,
+      "step": 4611
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00016944714400433125,
+      "loss": 1.0007,
+      "step": 4612
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00016943458634432096,
+      "loss": 0.8197,
+      "step": 4613
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00016942202656968216,
+      "loss": 0.8917,
+      "step": 4614
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00016940946468079737,
+      "loss": 0.9173,
+      "step": 4615
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00016939690067804914,
+      "loss": 0.9408,
+      "step": 4616
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00016938433456182016,
+      "loss": 0.8678,
+      "step": 4617
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00016937176633249302,
+      "loss": 0.9649,
+      "step": 4618
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.0001693591959904506,
+      "loss": 0.8041,
+      "step": 4619
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.0001693466235360757,
+      "loss": 0.8292,
+      "step": 4620
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00016933404896975118,
+      "loss": 0.9036,
+      "step": 4621
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00016932147229186003,
+      "loss": 0.9405,
+      "step": 4622
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00016930889350278525,
+      "loss": 0.8861,
+      "step": 4623
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00016929631260290993,
+      "loss": 0.9082,
+      "step": 4624
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00016928372959261725,
+      "loss": 0.8377,
+      "step": 4625
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.0001692711444722904,
+      "loss": 0.8147,
+      "step": 4626
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00016925855724231266,
+      "loss": 1.0219,
+      "step": 4627
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00016924596790306734,
+      "loss": 0.8456,
+      "step": 4628
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00016923337645493794,
+      "loss": 0.908,
+      "step": 4629
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00016922078289830785,
+      "loss": 1.0005,
+      "step": 4630
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00016920818723356062,
+      "loss": 1.0061,
+      "step": 4631
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.0001691955894610799,
+      "loss": 0.8789,
+      "step": 4632
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.0001691829895812493,
+      "loss": 0.8356,
+      "step": 4633
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00016917038759445255,
+      "loss": 1.0141,
+      "step": 4634
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.0001691577835010735,
+      "loss": 0.8496,
+      "step": 4635
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00016914517730149596,
+      "loss": 1.0738,
+      "step": 4636
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00016913256899610385,
+      "loss": 0.9297,
+      "step": 4637
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00016911995858528116,
+      "loss": 0.9437,
+      "step": 4638
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00016910734606941197,
+      "loss": 0.8953,
+      "step": 4639
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00016909473144888034,
+      "loss": 0.9018,
+      "step": 4640
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.00016908211472407047,
+      "loss": 0.8998,
+      "step": 4641
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.00016906949589536662,
+      "loss": 0.8516,
+      "step": 4642
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.0001690568749631531,
+      "loss": 0.9862,
+      "step": 4643
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.00016904425192781425,
+      "loss": 0.9246,
+      "step": 4644
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.00016903162678973455,
+      "loss": 0.8899,
+      "step": 4645
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.00016901899954929842,
+      "loss": 0.7987,
+      "step": 4646
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.0001690063702068905,
+      "loss": 0.8825,
+      "step": 4647
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.00016899373876289538,
+      "loss": 0.9183,
+      "step": 4648
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.00016898110521769777,
+      "loss": 0.9094,
+      "step": 4649
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.0001689684695716824,
+      "loss": 0.882,
+      "step": 4650
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.00016895583182523408,
+      "loss": 0.9318,
+      "step": 4651
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.00016894319197873774,
+      "loss": 0.9186,
+      "step": 4652
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.0001689305500325783,
+      "loss": 0.9783,
+      "step": 4653
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.00016891790598714076,
+      "loss": 0.9244,
+      "step": 4654
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.0001689052598428102,
+      "loss": 0.9402,
+      "step": 4655
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.00016889261159997175,
+      "loss": 0.9067,
+      "step": 4656
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.00016887996125901063,
+      "loss": 0.9138,
+      "step": 4657
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.0001688673088203121,
+      "loss": 0.8751,
+      "step": 4658
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.0001688546542842615,
+      "loss": 0.8194,
+      "step": 4659
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.00016884199765124417,
+      "loss": 0.9539,
+      "step": 4660
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.00016882933892164566,
+      "loss": 0.9666,
+      "step": 4661
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.00016881667809585143,
+      "loss": 0.8816,
+      "step": 4662
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.00016880401517424705,
+      "loss": 0.9163,
+      "step": 4663
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.0001687913501572182,
+      "loss": 0.897,
+      "step": 4664
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.0001687786830451506,
+      "loss": 0.8388,
+      "step": 4665
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.00016876601383842998,
+      "loss": 0.8781,
+      "step": 4666
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.00016875334253744226,
+      "loss": 0.9761,
+      "step": 4667
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.00016874066914257326,
+      "loss": 0.9269,
+      "step": 4668
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.000168727993654209,
+      "loss": 0.9307,
+      "step": 4669
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.0001687153160727355,
+      "loss": 0.9138,
+      "step": 4670
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.00016870263639853886,
+      "loss": 0.9174,
+      "step": 4671
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.0001686899546320052,
+      "loss": 0.9814,
+      "step": 4672
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.0001686772707735208,
+      "loss": 0.9456,
+      "step": 4673
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.00016866458482347194,
+      "loss": 0.9138,
+      "step": 4674
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.0001686518967822449,
+      "loss": 0.8915,
+      "step": 4675
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.00016863920665022617,
+      "loss": 0.8274,
+      "step": 4676
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.00016862651442780223,
+      "loss": 0.8673,
+      "step": 4677
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.00016861382011535957,
+      "loss": 0.8448,
+      "step": 4678
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.0001686011237132848,
+      "loss": 0.8758,
+      "step": 4679
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.00016858842522196463,
+      "loss": 0.9091,
+      "step": 4680
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.00016857572464178578,
+      "loss": 0.8633,
+      "step": 4681
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.00016856302197313503,
+      "loss": 0.8989,
+      "step": 4682
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.00016855031721639923,
+      "loss": 0.8909,
+      "step": 4683
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.0001685376103719653,
+      "loss": 0.8566,
+      "step": 4684
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.0001685249014402203,
+      "loss": 0.8947,
+      "step": 4685
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.0001685121904215512,
+      "loss": 0.9797,
+      "step": 4686
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.00016849947731634513,
+      "loss": 0.9095,
+      "step": 4687
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.00016848676212498928,
+      "loss": 0.7535,
+      "step": 4688
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.00016847404484787091,
+      "loss": 0.9417,
+      "step": 4689
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.00016846132548537724,
+      "loss": 0.7921,
+      "step": 4690
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.00016844860403789576,
+      "loss": 0.8807,
+      "step": 4691
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.00016843588050581383,
+      "loss": 0.7918,
+      "step": 4692
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.00016842315488951895,
+      "loss": 0.9717,
+      "step": 4693
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.00016841042718939868,
+      "loss": 0.8684,
+      "step": 4694
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.00016839769740584063,
+      "loss": 0.9034,
+      "step": 4695
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.00016838496553923254,
+      "loss": 0.9008,
+      "step": 4696
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.0001683722315899621,
+      "loss": 0.9282,
+      "step": 4697
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.00016835949555841712,
+      "loss": 1.02,
+      "step": 4698
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.00016834675744498552,
+      "loss": 1.0698,
+      "step": 4699
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.0001683340172500552,
+      "loss": 0.9488,
+      "step": 4700
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.0001683212749740142,
+      "loss": 0.9572,
+      "step": 4701
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.00016830853061725054,
+      "loss": 1.0167,
+      "step": 4702
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.0001682957841801524,
+      "loss": 0.9517,
+      "step": 4703
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.0001682830356631079,
+      "loss": 0.9703,
+      "step": 4704
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.0001682702850665054,
+      "loss": 0.8919,
+      "step": 4705
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.00016825753239073311,
+      "loss": 0.9408,
+      "step": 4706
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.00016824477763617953,
+      "loss": 0.9117,
+      "step": 4707
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.00016823202080323298,
+      "loss": 0.8821,
+      "step": 4708
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.00016821926189228203,
+      "loss": 0.9671,
+      "step": 4709
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.00016820650090371527,
+      "loss": 0.9264,
+      "step": 4710
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.0001681937378379213,
+      "loss": 0.8862,
+      "step": 4711
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.00016818097269528885,
+      "loss": 0.9631,
+      "step": 4712
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.00016816820547620664,
+      "loss": 0.9547,
+      "step": 4713
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.00016815543618106354,
+      "loss": 1.1046,
+      "step": 4714
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.00016814266481024841,
+      "loss": 0.8654,
+      "step": 4715
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.0001681298913641502,
+      "loss": 0.9466,
+      "step": 4716
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.00016811711584315797,
+      "loss": 0.863,
+      "step": 4717
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.00016810433824766076,
+      "loss": 0.87,
+      "step": 4718
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.00016809155857804768,
+      "loss": 0.8879,
+      "step": 4719
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.000168078776834708,
+      "loss": 0.8388,
+      "step": 4720
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.00016806599301803096,
+      "loss": 0.8449,
+      "step": 4721
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.00016805320712840586,
+      "loss": 0.8506,
+      "step": 4722
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.00016804041916622216,
+      "loss": 0.8836,
+      "step": 4723
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.0001680276291318693,
+      "loss": 0.9007,
+      "step": 4724
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.00016801483702573672,
+      "loss": 0.816,
+      "step": 4725
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.0001680020428482141,
+      "loss": 0.9445,
+      "step": 4726
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.00016798924659969103,
+      "loss": 0.9241,
+      "step": 4727
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.00016797644828055725,
+      "loss": 0.8282,
+      "step": 4728
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.00016796364789120252,
+      "loss": 1.0588,
+      "step": 4729
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.0001679508454320167,
+      "loss": 0.7958,
+      "step": 4730
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.00016793804090338963,
+      "loss": 0.9293,
+      "step": 4731
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.00016792523430571134,
+      "loss": 0.8825,
+      "step": 4732
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.00016791242563937177,
+      "loss": 0.7912,
+      "step": 4733
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.0001678996149047611,
+      "loss": 1.0153,
+      "step": 4734
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.00016788680210226946,
+      "loss": 0.898,
+      "step": 4735
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.000167873987232287,
+      "loss": 0.818,
+      "step": 4736
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.00016786117029520407,
+      "loss": 0.8583,
+      "step": 4737
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.00016784835129141096,
+      "loss": 0.8693,
+      "step": 4738
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.0001678355302212981,
+      "loss": 0.7606,
+      "step": 4739
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.00016782270708525597,
+      "loss": 1.0542,
+      "step": 4740
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.00016780988188367505,
+      "loss": 1.0003,
+      "step": 4741
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.00016779705461694594,
+      "loss": 0.8346,
+      "step": 4742
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.00016778422528545935,
+      "loss": 0.8882,
+      "step": 4743
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.00016777139388960593,
+      "loss": 0.9492,
+      "step": 4744
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.00016775856042977648,
+      "loss": 0.8845,
+      "step": 4745
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.00016774572490636187,
+      "loss": 0.8186,
+      "step": 4746
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.00016773288731975302,
+      "loss": 1.0257,
+      "step": 4747
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.0001677200476703408,
+      "loss": 0.8491,
+      "step": 4748
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.00016770720595851634,
+      "loss": 0.9079,
+      "step": 4749
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.00016769436218467068,
+      "loss": 0.8429,
+      "step": 4750
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.00016768151634919502,
+      "loss": 0.9424,
+      "step": 4751
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.00016766866845248054,
+      "loss": 0.9307,
+      "step": 4752
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.00016765581849491855,
+      "loss": 0.8544,
+      "step": 4753
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.00016764296647690037,
+      "loss": 1.0117,
+      "step": 4754
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.00016763011239881743,
+      "loss": 0.9535,
+      "step": 4755
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.0001676172562610612,
+      "loss": 0.9497,
+      "step": 4756
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.0001676043980640232,
+      "loss": 0.7876,
+      "step": 4757
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.00016759153780809502,
+      "loss": 0.9425,
+      "step": 4758
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.00016757867549366834,
+      "loss": 0.7908,
+      "step": 4759
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.00016756581112113488,
+      "loss": 0.9937,
+      "step": 4760
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.00016755294469088637,
+      "loss": 0.936,
+      "step": 4761
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.00016754007620331475,
+      "loss": 1.0272,
+      "step": 4762
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.0001675272056588119,
+      "loss": 0.9602,
+      "step": 4763
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.00016751433305776974,
+      "loss": 0.9729,
+      "step": 4764
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.00016750145840058035,
+      "loss": 0.8634,
+      "step": 4765
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.00016748858168763578,
+      "loss": 0.8739,
+      "step": 4766
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.00016747570291932826,
+      "loss": 0.9136,
+      "step": 4767
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.00016746282209604996,
+      "loss": 0.919,
+      "step": 4768
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.0001674499392181932,
+      "loss": 0.8686,
+      "step": 4769
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.0001674370542861503,
+      "loss": 0.9364,
+      "step": 4770
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.00016742416730031371,
+      "loss": 0.9976,
+      "step": 4771
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.00016741127826107584,
+      "loss": 1.0521,
+      "step": 4772
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.0001673983871688293,
+      "loss": 0.6675,
+      "step": 4773
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.0001673854940239666,
+      "loss": 0.9907,
+      "step": 4774
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.00016737259882688047,
+      "loss": 0.8087,
+      "step": 4775
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.00016735970157796365,
+      "loss": 0.8423,
+      "step": 4776
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.00016734680227760884,
+      "loss": 0.8578,
+      "step": 4777
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.00016733390092620894,
+      "loss": 0.9566,
+      "step": 4778
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.00016732099752415685,
+      "loss": 0.9104,
+      "step": 4779
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.00016730809207184555,
+      "loss": 1.0377,
+      "step": 4780
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.00016729518456966806,
+      "loss": 0.8888,
+      "step": 4781
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.00016728227501801752,
+      "loss": 0.869,
+      "step": 4782
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.00016726936341728704,
+      "loss": 1.0265,
+      "step": 4783
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.00016725644976786988,
+      "loss": 0.7783,
+      "step": 4784
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.0001672435340701593,
+      "loss": 0.7569,
+      "step": 4785
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.00016723061632454864,
+      "loss": 0.848,
+      "step": 4786
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.00016721769653143132,
+      "loss": 0.9613,
+      "step": 4787
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.00016720477469120084,
+      "loss": 0.919,
+      "step": 4788
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.0001671918508042507,
+      "loss": 0.9356,
+      "step": 4789
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.00016717892487097454,
+      "loss": 0.9826,
+      "step": 4790
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.00016716599689176593,
+      "loss": 0.8941,
+      "step": 4791
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.0001671530668670187,
+      "loss": 0.9571,
+      "step": 4792
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.00016714013479712656,
+      "loss": 0.9316,
+      "step": 4793
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.00016712720068248334,
+      "loss": 0.8981,
+      "step": 4794
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.00016711426452348303,
+      "loss": 0.9238,
+      "step": 4795
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.00016710132632051953,
+      "loss": 0.9681,
+      "step": 4796
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.0001670883860739869,
+      "loss": 0.8486,
+      "step": 4797
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.00016707544378427927,
+      "loss": 0.99,
+      "step": 4798
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.00016706249945179074,
+      "loss": 0.8257,
+      "step": 4799
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.00016704955307691554,
+      "loss": 0.9181,
+      "step": 4800
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.00016703660466004796,
+      "loss": 0.8523,
+      "step": 4801
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.00016702365420158234,
+      "loss": 0.934,
+      "step": 4802
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.00016701070170191312,
+      "loss": 0.981,
+      "step": 4803
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.0001669977471614347,
+      "loss": 0.853,
+      "step": 4804
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.0001669847905805417,
+      "loss": 0.9376,
+      "step": 4805
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.0001669718319596286,
+      "loss": 0.8936,
+      "step": 4806
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.00016695887129909015,
+      "loss": 0.9025,
+      "step": 4807
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.00016694590859932102,
+      "loss": 0.8426,
+      "step": 4808
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.00016693294386071602,
+      "loss": 0.9132,
+      "step": 4809
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.00016691997708366996,
+      "loss": 0.9022,
+      "step": 4810
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.00016690700826857774,
+      "loss": 0.9437,
+      "step": 4811
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.00016689403741583435,
+      "loss": 1.0152,
+      "step": 4812
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.00016688106452583478,
+      "loss": 0.9906,
+      "step": 4813
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.00016686808959897418,
+      "loss": 0.7783,
+      "step": 4814
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.00016685511263564768,
+      "loss": 1.0114,
+      "step": 4815
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.00016684213363625045,
+      "loss": 0.8581,
+      "step": 4816
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.00016682915260117783,
+      "loss": 1.0618,
+      "step": 4817
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.00016681616953082506,
+      "loss": 0.9188,
+      "step": 4818
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.00016680318442558766,
+      "loss": 0.8263,
+      "step": 4819
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.00016679019728586098,
+      "loss": 0.9865,
+      "step": 4820
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.00016677720811204063,
+      "loss": 1.0927,
+      "step": 4821
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.00016676421690452214,
+      "loss": 0.9797,
+      "step": 4822
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.00016675122366370122,
+      "loss": 0.955,
+      "step": 4823
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.00016673822838997348,
+      "loss": 0.7891,
+      "step": 4824
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.00016672523108373478,
+      "loss": 0.9309,
+      "step": 4825
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.0001667122317453809,
+      "loss": 0.9481,
+      "step": 4826
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.00016669923037530775,
+      "loss": 0.9519,
+      "step": 4827
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.0001666862269739113,
+      "loss": 1.0124,
+      "step": 4828
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.00016667322154158756,
+      "loss": 0.8547,
+      "step": 4829
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.0001666602140787326,
+      "loss": 0.8693,
+      "step": 4830
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.0001666472045857426,
+      "loss": 1.0069,
+      "step": 4831
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.0001666341930630137,
+      "loss": 0.9537,
+      "step": 4832
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.0001666211795109422,
+      "loss": 0.9219,
+      "step": 4833
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.00016660816392992443,
+      "loss": 1.0174,
+      "step": 4834
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.00016659514632035678,
+      "loss": 1.011,
+      "step": 4835
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.00016658212668263568,
+      "loss": 0.977,
+      "step": 4836
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.00016656910501715771,
+      "loss": 0.8827,
+      "step": 4837
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.00016655608132431934,
+      "loss": 1.0139,
+      "step": 4838
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.00016654305560451728,
+      "loss": 0.8284,
+      "step": 4839
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.0001665300278581482,
+      "loss": 0.8965,
+      "step": 4840
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.0001665169980856089,
+      "loss": 1.0486,
+      "step": 4841
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.00016650396628729614,
+      "loss": 0.9208,
+      "step": 4842
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.00016649093246360686,
+      "loss": 0.8542,
+      "step": 4843
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.00016647789661493795,
+      "loss": 1.0441,
+      "step": 4844
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.00016646485874168644,
+      "loss": 0.8968,
+      "step": 4845
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.0001664518188442494,
+      "loss": 0.8146,
+      "step": 4846
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.00016643877692302397,
+      "loss": 0.9493,
+      "step": 4847
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.00016642573297840735,
+      "loss": 0.9441,
+      "step": 4848
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.00016641268701079675,
+      "loss": 1.0986,
+      "step": 4849
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.00016639963902058953,
+      "loss": 0.8049,
+      "step": 4850
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.00016638658900818302,
+      "loss": 0.9262,
+      "step": 4851
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.0001663735369739747,
+      "loss": 0.9021,
+      "step": 4852
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.00016636048291836206,
+      "loss": 0.8809,
+      "step": 4853
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.00016634742684174265,
+      "loss": 0.7877,
+      "step": 4854
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.00016633436874451406,
+      "loss": 0.8464,
+      "step": 4855
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.00016632130862707407,
+      "loss": 0.9926,
+      "step": 4856
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.00016630824648982036,
+      "loss": 0.8673,
+      "step": 4857
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.0001662951823331507,
+      "loss": 0.8943,
+      "step": 4858
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.00016628211615746303,
+      "loss": 1.0579,
+      "step": 4859
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.00016626904796315526,
+      "loss": 0.8815,
+      "step": 4860
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.00016625597775062536,
+      "loss": 0.9473,
+      "step": 4861
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.0001662429055202714,
+      "loss": 0.9227,
+      "step": 4862
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.00016622983127249149,
+      "loss": 0.9812,
+      "step": 4863
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.00016621675500768382,
+      "loss": 0.9444,
+      "step": 4864
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.00016620367672624662,
+      "loss": 0.8942,
+      "step": 4865
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.00016619059642857814,
+      "loss": 0.8507,
+      "step": 4866
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.0001661775141150768,
+      "loss": 1.0231,
+      "step": 4867
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.000166164429786141,
+      "loss": 0.8326,
+      "step": 4868
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.00016615134344216928,
+      "loss": 0.8906,
+      "step": 4869
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.0001661382550835601,
+      "loss": 0.8909,
+      "step": 4870
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.00016612516471071207,
+      "loss": 0.9713,
+      "step": 4871
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.00016611207232402388,
+      "loss": 0.9986,
+      "step": 4872
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.0001660989779238943,
+      "loss": 0.9205,
+      "step": 4873
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.00016608588151072206,
+      "loss": 0.9971,
+      "step": 4874
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.00016607278308490605,
+      "loss": 0.953,
+      "step": 4875
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.00016605968264684514,
+      "loss": 0.8968,
+      "step": 4876
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.00016604658019693836,
+      "loss": 0.8811,
+      "step": 4877
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.0001660334757355847,
+      "loss": 0.8963,
+      "step": 4878
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.0001660203692631833,
+      "loss": 0.8175,
+      "step": 4879
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.00016600726078013324,
+      "loss": 0.8588,
+      "step": 4880
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.0001659941502868338,
+      "loss": 0.889,
+      "step": 4881
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.00016598103778368424,
+      "loss": 0.9685,
+      "step": 4882
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.00016596792327108393,
+      "loss": 0.8689,
+      "step": 4883
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.00016595480674943224,
+      "loss": 1.0329,
+      "step": 4884
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.00016594168821912862,
+      "loss": 0.9325,
+      "step": 4885
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.00016592856768057264,
+      "loss": 0.9373,
+      "step": 4886
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.00016591544513416384,
+      "loss": 0.9264,
+      "step": 4887
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.00016590232058030189,
+      "loss": 0.8013,
+      "step": 4888
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.0001658891940193865,
+      "loss": 0.8566,
+      "step": 4889
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.00016587606545181744,
+      "loss": 0.8558,
+      "step": 4890
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.00016586293487799454,
+      "loss": 0.865,
+      "step": 4891
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.00016584980229831768,
+      "loss": 0.9668,
+      "step": 4892
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.00016583666771318681,
+      "loss": 1.0024,
+      "step": 4893
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.00016582353112300197,
+      "loss": 0.9204,
+      "step": 4894
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.00016581039252816323,
+      "loss": 0.9456,
+      "step": 4895
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.0001657972519290707,
+      "loss": 0.8357,
+      "step": 4896
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.00016578410932612458,
+      "loss": 0.9183,
+      "step": 4897
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.00016577096471972512,
+      "loss": 0.8601,
+      "step": 4898
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.0001657578181102727,
+      "loss": 0.9316,
+      "step": 4899
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.00016574466949816763,
+      "loss": 0.9473,
+      "step": 4900
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.00016573151888381037,
+      "loss": 1.0102,
+      "step": 4901
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.00016571836626760145,
+      "loss": 1.0571,
+      "step": 4902
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.0001657052116499414,
+      "loss": 0.8928,
+      "step": 4903
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.00016569205503123082,
+      "loss": 0.9366,
+      "step": 4904
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.00016567889641187045,
+      "loss": 1.0015,
+      "step": 4905
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.00016566573579226103,
+      "loss": 0.7754,
+      "step": 4906
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.00016565257317280335,
+      "loss": 0.8878,
+      "step": 4907
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.0001656394085538983,
+      "loss": 1.0502,
+      "step": 4908
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.00016562624193594674,
+      "loss": 0.8462,
+      "step": 4909
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.00016561307331934973,
+      "loss": 0.9247,
+      "step": 4910
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.00016559990270450828,
+      "loss": 0.973,
+      "step": 4911
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.00016558673009182351,
+      "loss": 0.8,
+      "step": 4912
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.00016557355548169658,
+      "loss": 0.9057,
+      "step": 4913
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.0001655603788745288,
+      "loss": 0.9151,
+      "step": 4914
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.00016554720027072136,
+      "loss": 0.9491,
+      "step": 4915
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.00016553401967067566,
+      "loss": 0.8538,
+      "step": 4916
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.0001655208370747931,
+      "loss": 0.8287,
+      "step": 4917
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.00016550765248347517,
+      "loss": 0.8644,
+      "step": 4918
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.00016549446589712342,
+      "loss": 0.8284,
+      "step": 4919
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.00016548127731613944,
+      "loss": 1.1122,
+      "step": 4920
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.00016546808674092485,
+      "loss": 0.8635,
+      "step": 4921
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.00016545489417188141,
+      "loss": 0.9231,
+      "step": 4922
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.00016544169960941088,
+      "loss": 0.887,
+      "step": 4923
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.0001654285030539151,
+      "loss": 0.8327,
+      "step": 4924
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.000165415304505796,
+      "loss": 0.9392,
+      "step": 4925
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.0001654021039654555,
+      "loss": 0.8926,
+      "step": 4926
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.00016538890143329564,
+      "loss": 0.8537,
+      "step": 4927
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.00016537569690971851,
+      "loss": 0.9122,
+      "step": 4928
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.00016536249039512623,
+      "loss": 0.8248,
+      "step": 4929
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.00016534928188992104,
+      "loss": 0.9071,
+      "step": 4930
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.00016533607139450517,
+      "loss": 0.8896,
+      "step": 4931
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.000165322858909281,
+      "loss": 0.7462,
+      "step": 4932
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.0001653096444346508,
+      "loss": 0.8899,
+      "step": 4933
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.00016529642797101718,
+      "loss": 0.8602,
+      "step": 4934
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.0001652832095187825,
+      "loss": 1.0211,
+      "step": 4935
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.00016526998907834942,
+      "loss": 0.9452,
+      "step": 4936
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.00016525676665012052,
+      "loss": 0.9297,
+      "step": 4937
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.0001652435422344985,
+      "loss": 0.8774,
+      "step": 4938
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.00016523031583188613,
+      "loss": 0.8279,
+      "step": 4939
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.0001652170874426862,
+      "loss": 0.9529,
+      "step": 4940
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.00016520385706730157,
+      "loss": 0.8528,
+      "step": 4941
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.0001651906247061352,
+      "loss": 0.942,
+      "step": 4942
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.00016517739035959005,
+      "loss": 0.9401,
+      "step": 4943
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.0001651641540280692,
+      "loss": 0.9835,
+      "step": 4944
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.00016515091571197573,
+      "loss": 0.9372,
+      "step": 4945
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.00016513767541171285,
+      "loss": 0.9229,
+      "step": 4946
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.00016512443312768378,
+      "loss": 0.8236,
+      "step": 4947
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.0001651111888602918,
+      "loss": 0.9289,
+      "step": 4948
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.0001650979426099403,
+      "loss": 1.0919,
+      "step": 4949
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.0001650846943770326,
+      "loss": 0.9899,
+      "step": 4950
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.0001650714441619723,
+      "loss": 0.947,
+      "step": 4951
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.00016505819196516283,
+      "loss": 0.9385,
+      "step": 4952
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.00016504493778700787,
+      "loss": 1.0129,
+      "step": 4953
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.00016503168162791101,
+      "loss": 0.8226,
+      "step": 4954
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.00016501842348827603,
+      "loss": 0.8556,
+      "step": 4955
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.00016500516336850663,
+      "loss": 0.9408,
+      "step": 4956
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.0001649919012690067,
+      "loss": 0.869,
+      "step": 4957
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.00016497863719018013,
+      "loss": 1.0017,
+      "step": 4958
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.00016496537113243088,
+      "loss": 0.9297,
+      "step": 4959
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.00016495210309616293,
+      "loss": 0.9145,
+      "step": 4960
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.0001649388330817804,
+      "loss": 0.9199,
+      "step": 4961
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.0001649255610896874,
+      "loss": 1.0252,
+      "step": 4962
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.00016491228712028815,
+      "loss": 0.9013,
+      "step": 4963
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.0001648990111739869,
+      "loss": 0.9682,
+      "step": 4964
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.000164885733251188,
+      "loss": 0.8653,
+      "step": 4965
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.00016487245335229574,
+      "loss": 0.9242,
+      "step": 4966
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.00016485917147771462,
+      "loss": 0.9358,
+      "step": 4967
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.00016484588762784918,
+      "loss": 0.8516,
+      "step": 4968
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.0001648326018031039,
+      "loss": 0.9266,
+      "step": 4969
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.00016481931400388344,
+      "loss": 0.7864,
+      "step": 4970
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.00016480602423059247,
+      "loss": 0.8564,
+      "step": 4971
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.00016479273248363576,
+      "loss": 0.9316,
+      "step": 4972
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.00016477943876341806,
+      "loss": 0.8881,
+      "step": 4973
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.00016476614307034426,
+      "loss": 0.9408,
+      "step": 4974
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.00016475284540481925,
+      "loss": 0.7831,
+      "step": 4975
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.00016473954576724803,
+      "loss": 0.9008,
+      "step": 4976
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.0001647262441580357,
+      "loss": 0.8698,
+      "step": 4977
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.00016471294057758727,
+      "loss": 0.8256,
+      "step": 4978
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.0001646996350263079,
+      "loss": 0.8876,
+      "step": 4979
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.00016468632750460288,
+      "loss": 0.9207,
+      "step": 4980
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.00016467301801287743,
+      "loss": 0.8813,
+      "step": 4981
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.00016465970655153695,
+      "loss": 0.9718,
+      "step": 4982
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.0001646463931209868,
+      "loss": 0.9264,
+      "step": 4983
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.0001646330777216324,
+      "loss": 0.9799,
+      "step": 4984
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.00016461976035387935,
+      "loss": 0.9667,
+      "step": 4985
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.0001646064410181332,
+      "loss": 0.9252,
+      "step": 4986
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.00016459311971479958,
+      "loss": 0.8658,
+      "step": 4987
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.00016457979644428418,
+      "loss": 0.9885,
+      "step": 4988
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.0001645664712069928,
+      "loss": 0.9661,
+      "step": 4989
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.0001645531440033312,
+      "loss": 0.8751,
+      "step": 4990
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.00016453981483370532,
+      "loss": 0.8632,
+      "step": 4991
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.00016452648369852106,
+      "loss": 1.065,
+      "step": 4992
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.00016451315059818442,
+      "loss": 0.8474,
+      "step": 4993
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.0001644998155331015,
+      "loss": 0.8515,
+      "step": 4994
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.00016448647850367837,
+      "loss": 0.7781,
+      "step": 4995
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.0001644731395103212,
+      "loss": 0.8797,
+      "step": 4996
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.0001644597985534363,
+      "loss": 0.8845,
+      "step": 4997
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.00016444645563342994,
+      "loss": 0.9748,
+      "step": 4998
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.00016443311075070845,
+      "loss": 0.8855,
+      "step": 4999
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.00016441976390567828,
+      "loss": 0.9543,
+      "step": 5000
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.00016440641509874587,
+      "loss": 0.9597,
+      "step": 5001
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.00016439306433031775,
+      "loss": 0.8874,
+      "step": 5002
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.00016437971160080062,
+      "loss": 0.8174,
+      "step": 5003
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.000164366356910601,
+      "loss": 0.9341,
+      "step": 5004
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.00016435300026012566,
+      "loss": 0.9185,
+      "step": 5005
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.00016433964164978146,
+      "loss": 0.8816,
+      "step": 5006
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.0001643262810799751,
+      "loss": 0.9448,
+      "step": 5007
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.00016431291855111354,
+      "loss": 0.8544,
+      "step": 5008
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.00016429955406360375,
+      "loss": 0.9828,
+      "step": 5009
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.00016428618761785274,
+      "loss": 0.8322,
+      "step": 5010
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.00016427281921426755,
+      "loss": 0.8812,
+      "step": 5011
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.00016425944885325536,
+      "loss": 0.9245,
+      "step": 5012
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.00016424607653522334,
+      "loss": 0.9796,
+      "step": 5013
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.00016423270226057873,
+      "loss": 0.9065,
+      "step": 5014
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.00016421932602972886,
+      "loss": 0.9412,
+      "step": 5015
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.0001642059478430811,
+      "loss": 0.9164,
+      "step": 5016
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.0001641925677010429,
+      "loss": 1.0103,
+      "step": 5017
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.0001641791856040217,
+      "loss": 0.8366,
+      "step": 5018
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.00016416580155242513,
+      "loss": 0.9172,
+      "step": 5019
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.00016415241554666073,
+      "loss": 1.0281,
+      "step": 5020
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.0001641390275871362,
+      "loss": 0.9307,
+      "step": 5021
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.0001641256376742593,
+      "loss": 1.0285,
+      "step": 5022
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.00016411224580843776,
+      "loss": 0.9295,
+      "step": 5023
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.00016409885199007945,
+      "loss": 1.009,
+      "step": 5024
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.0001640854562195923,
+      "loss": 0.9413,
+      "step": 5025
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.00016407205849738426,
+      "loss": 1.1001,
+      "step": 5026
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.00016405865882386335,
+      "loss": 0.998,
+      "step": 5027
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.00016404525719943767,
+      "loss": 0.8773,
+      "step": 5028
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.0001640318536245154,
+      "loss": 0.9854,
+      "step": 5029
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.00016401844809950467,
+      "loss": 0.9512,
+      "step": 5030
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.0001640050406248138,
+      "loss": 0.8919,
+      "step": 5031
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.00016399163120085105,
+      "loss": 0.9864,
+      "step": 5032
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.0001639782198280249,
+      "loss": 0.8828,
+      "step": 5033
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.00016396480650674374,
+      "loss": 0.7505,
+      "step": 5034
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.00016395139123741607,
+      "loss": 0.8748,
+      "step": 5035
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.00016393797402045046,
+      "loss": 0.8929,
+      "step": 5036
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.00016392455485625553,
+      "loss": 0.8304,
+      "step": 5037
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.00016391113374523994,
+      "loss": 0.8668,
+      "step": 5038
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.00016389771068781248,
+      "loss": 0.9094,
+      "step": 5039
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.0001638842856843819,
+      "loss": 0.7862,
+      "step": 5040
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.0001638708587353571,
+      "loss": 0.8622,
+      "step": 5041
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.00016385742984114694,
+      "loss": 0.9814,
+      "step": 5042
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.00016384399900216042,
+      "loss": 0.9903,
+      "step": 5043
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.00016383056621880663,
+      "loss": 0.9535,
+      "step": 5044
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.0001638171314914946,
+      "loss": 0.9573,
+      "step": 5045
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.00016380369482063353,
+      "loss": 0.8406,
+      "step": 5046
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.00016379025620663257,
+      "loss": 0.9053,
+      "step": 5047
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.00016377681564990105,
+      "loss": 0.9059,
+      "step": 5048
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.00016376337315084828,
+      "loss": 1.056,
+      "step": 5049
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.00016374992870988363,
+      "loss": 0.842,
+      "step": 5050
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.0001637364823274166,
+      "loss": 1.0277,
+      "step": 5051
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.0001637230340038567,
+      "loss": 0.7776,
+      "step": 5052
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.00016370958373961345,
+      "loss": 0.9049,
+      "step": 5053
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.0001636961315350965,
+      "loss": 0.8693,
+      "step": 5054
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.00016368267739071552,
+      "loss": 0.919,
+      "step": 5055
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.00016366922130688026,
+      "loss": 0.8449,
+      "step": 5056
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.00016365576328400058,
+      "loss": 0.862,
+      "step": 5057
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.00016364230332248627,
+      "loss": 0.9411,
+      "step": 5058
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.0001636288414227473,
+      "loss": 0.9241,
+      "step": 5059
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.00016361537758519362,
+      "loss": 0.8409,
+      "step": 5060
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.0001636019118102353,
+      "loss": 0.7925,
+      "step": 5061
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.0001635884440982824,
+      "loss": 0.9284,
+      "step": 5062
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.00016357497444974514,
+      "loss": 0.9457,
+      "step": 5063
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.00016356150286503366,
+      "loss": 0.8836,
+      "step": 5064
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.00016354802934455832,
+      "loss": 0.8488,
+      "step": 5065
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.00016353455388872938,
+      "loss": 0.9472,
+      "step": 5066
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.0001635210764979573,
+      "loss": 0.9156,
+      "step": 5067
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.00016350759717265247,
+      "loss": 0.8425,
+      "step": 5068
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.00016349411591322545,
+      "loss": 0.8932,
+      "step": 5069
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.0001634806327200868,
+      "loss": 0.8581,
+      "step": 5070
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.00016346714759364717,
+      "loss": 0.8666,
+      "step": 5071
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.0001634536605343172,
+      "loss": 0.994,
+      "step": 5072
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.00016344017154250762,
+      "loss": 0.8683,
+      "step": 5073
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.00016342668061862934,
+      "loss": 0.8212,
+      "step": 5074
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.00016341318776309316,
+      "loss": 0.8029,
+      "step": 5075
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.00016339969297631,
+      "loss": 0.9393,
+      "step": 5076
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.00016338619625869084,
+      "loss": 0.8483,
+      "step": 5077
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.00016337269761064675,
+      "loss": 0.9184,
+      "step": 5078
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.0001633591970325888,
+      "loss": 0.8745,
+      "step": 5079
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.00016334569452492823,
+      "loss": 0.8451,
+      "step": 5080
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.00016333219008807617,
+      "loss": 0.9363,
+      "step": 5081
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.0001633186837224439,
+      "loss": 0.8417,
+      "step": 5082
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.00016330517542844278,
+      "loss": 0.9481,
+      "step": 5083
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.0001632916652064842,
+      "loss": 0.9654,
+      "step": 5084
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.00016327815305697962,
+      "loss": 0.8743,
+      "step": 5085
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.00016326463898034058,
+      "loss": 0.9425,
+      "step": 5086
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.0001632511229769786,
+      "loss": 0.9375,
+      "step": 5087
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.00016323760504730533,
+      "loss": 0.8824,
+      "step": 5088
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.00016322408519173242,
+      "loss": 0.8679,
+      "step": 5089
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.00016321056341067172,
+      "loss": 0.947,
+      "step": 5090
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.0001631970397045349,
+      "loss": 0.8409,
+      "step": 5091
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.00016318351407373395,
+      "loss": 1.0941,
+      "step": 5092
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.00016316998651868075,
+      "loss": 0.9172,
+      "step": 5093
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.00016315645703978722,
+      "loss": 0.8628,
+      "step": 5094
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.00016314292563746548,
+      "loss": 1.1036,
+      "step": 5095
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.00016312939231212756,
+      "loss": 0.9848,
+      "step": 5096
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.00016311585706418568,
+      "loss": 0.9555,
+      "step": 5097
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.00016310231989405204,
+      "loss": 0.7529,
+      "step": 5098
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.0001630887808021389,
+      "loss": 0.8652,
+      "step": 5099
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.0001630752397888586,
+      "loss": 0.9287,
+      "step": 5100
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.0001630616968546235,
+      "loss": 1.0323,
+      "step": 5101
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.0001630481519998461,
+      "loss": 0.9381,
+      "step": 5102
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.00016303460522493888,
+      "loss": 0.8477,
+      "step": 5103
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.00016302105653031443,
+      "loss": 0.9583,
+      "step": 5104
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.00016300750591638532,
+      "loss": 0.9794,
+      "step": 5105
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.0001629939533835643,
+      "loss": 0.772,
+      "step": 5106
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.00016298039893226406,
+      "loss": 0.8765,
+      "step": 5107
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.00016296684256289745,
+      "loss": 0.8367,
+      "step": 5108
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.00016295328427587728,
+      "loss": 0.8989,
+      "step": 5109
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.00016293972407161651,
+      "loss": 0.9276,
+      "step": 5110
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.00016292616195052808,
+      "loss": 0.8766,
+      "step": 5111
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.00016291259791302506,
+      "loss": 0.8787,
+      "step": 5112
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.0001628990319595205,
+      "loss": 0.9215,
+      "step": 5113
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.00016288546409042756,
+      "loss": 1.0399,
+      "step": 5114
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.0001628718943061595,
+      "loss": 0.8813,
+      "step": 5115
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.00016285832260712953,
+      "loss": 0.857,
+      "step": 5116
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.000162844748993751,
+      "loss": 0.9809,
+      "step": 5117
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.00016283117346643726,
+      "loss": 0.8969,
+      "step": 5118
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.0001628175960256018,
+      "loss": 1.0723,
+      "step": 5119
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.00016280401667165812,
+      "loss": 0.8855,
+      "step": 5120
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.00016279043540501974,
+      "loss": 0.8283,
+      "step": 5121
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.00016277685222610032,
+      "loss": 0.9394,
+      "step": 5122
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.0001627632671353135,
+      "loss": 0.9641,
+      "step": 5123
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.000162749680133073,
+      "loss": 0.9773,
+      "step": 5124
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.0001627360912197927,
+      "loss": 0.932,
+      "step": 5125
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.00016272250039588634,
+      "loss": 0.7409,
+      "step": 5126
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.00016270890766176786,
+      "loss": 0.7345,
+      "step": 5127
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.0001626953130178513,
+      "loss": 0.9082,
+      "step": 5128
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.00016268171646455058,
+      "loss": 0.8687,
+      "step": 5129
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.00016266811800227986,
+      "loss": 0.8097,
+      "step": 5130
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.00016265451763145325,
+      "loss": 0.8752,
+      "step": 5131
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.00016264091535248495,
+      "loss": 0.8855,
+      "step": 5132
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.0001626273111657892,
+      "loss": 0.9397,
+      "step": 5133
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.00016261370507178032,
+      "loss": 0.9211,
+      "step": 5134
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.0001626000970708727,
+      "loss": 0.8363,
+      "step": 5135
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.0001625864871634808,
+      "loss": 0.9399,
+      "step": 5136
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.00016257287535001907,
+      "loss": 0.9603,
+      "step": 5137
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.00016255926163090206,
+      "loss": 0.9046,
+      "step": 5138
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.00016254564600654437,
+      "loss": 0.9296,
+      "step": 5139
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.00016253202847736067,
+      "loss": 0.9335,
+      "step": 5140
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.00016251840904376566,
+      "loss": 0.9805,
+      "step": 5141
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.0001625047877061742,
+      "loss": 0.8406,
+      "step": 5142
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.00016249116446500104,
+      "loss": 0.9214,
+      "step": 5143
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.0001624775393206611,
+      "loss": 0.9501,
+      "step": 5144
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.00016246391227356933,
+      "loss": 1.1267,
+      "step": 5145
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.00016245028332414078,
+      "loss": 0.8786,
+      "step": 5146
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.0001624366524727905,
+      "loss": 0.8953,
+      "step": 5147
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.00016242301971993359,
+      "loss": 0.8499,
+      "step": 5148
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.00016240938506598525,
+      "loss": 0.8635,
+      "step": 5149
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.00016239574851136071,
+      "loss": 0.9604,
+      "step": 5150
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.0001623821100564753,
+      "loss": 1.0755,
+      "step": 5151
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.00016236846970174438,
+      "loss": 0.9794,
+      "step": 5152
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.00016235482744758332,
+      "loss": 0.7947,
+      "step": 5153
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.00016234118329440763,
+      "loss": 0.8405,
+      "step": 5154
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.0001623275372426329,
+      "loss": 0.9903,
+      "step": 5155
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.0001623138892926746,
+      "loss": 0.9256,
+      "step": 5156
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.00016230023944494845,
+      "loss": 0.944,
+      "step": 5157
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.00016228658769987014,
+      "loss": 0.9817,
+      "step": 5158
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.00016227293405785543,
+      "loss": 0.9308,
+      "step": 5159
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.00016225927851932018,
+      "loss": 0.9234,
+      "step": 5160
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.00016224562108468023,
+      "loss": 0.6952,
+      "step": 5161
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.00016223196175435152,
+      "loss": 0.9336,
+      "step": 5162
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.00016221830052875002,
+      "loss": 0.8151,
+      "step": 5163
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.0001622046374082918,
+      "loss": 0.9877,
+      "step": 5164
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.00016219097239339304,
+      "loss": 0.9122,
+      "step": 5165
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.00016217730548446978,
+      "loss": 0.9529,
+      "step": 5166
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.0001621636366819384,
+      "loss": 0.9382,
+      "step": 5167
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.00016214996598621503,
+      "loss": 0.8235,
+      "step": 5168
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.0001621362933977161,
+      "loss": 0.8564,
+      "step": 5169
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.00016212261891685795,
+      "loss": 0.9999,
+      "step": 5170
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.00016210894254405713,
+      "loss": 0.9478,
+      "step": 5171
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.00016209526427973006,
+      "loss": 0.8337,
+      "step": 5172
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.00016208158412429335,
+      "loss": 0.8802,
+      "step": 5173
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.00016206790207816364,
+      "loss": 0.9198,
+      "step": 5174
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.00016205421814175755,
+      "loss": 0.9219,
+      "step": 5175
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.0001620405323154919,
+      "loss": 0.8761,
+      "step": 5176
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.00016202684459978351,
+      "loss": 0.8486,
+      "step": 5177
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.00016201315499504915,
+      "loss": 0.8945,
+      "step": 5178
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.00016199946350170584,
+      "loss": 0.9848,
+      "step": 5179
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.00016198577012017045,
+      "loss": 1.0246,
+      "step": 5180
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.00016197207485086002,
+      "loss": 0.7662,
+      "step": 5181
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.00016195837769419173,
+      "loss": 0.9528,
+      "step": 5182
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.00016194467865058266,
+      "loss": 0.8249,
+      "step": 5183
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.00016193097772045006,
+      "loss": 0.998,
+      "step": 5184
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.0001619172749042111,
+      "loss": 1.0146,
+      "step": 5185
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.00016190357020228323,
+      "loss": 0.9217,
+      "step": 5186
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.00016188986361508368,
+      "loss": 0.9879,
+      "step": 5187
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.00016187615514303,
+      "loss": 0.9225,
+      "step": 5188
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.0001618624447865396,
+      "loss": 0.8201,
+      "step": 5189
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.0001618487325460301,
+      "loss": 1.0815,
+      "step": 5190
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.00016183501842191909,
+      "loss": 0.9783,
+      "step": 5191
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.0001618213024146242,
+      "loss": 0.9638,
+      "step": 5192
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.00016180758452456314,
+      "loss": 0.9329,
+      "step": 5193
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.00016179386475215376,
+      "loss": 0.8926,
+      "step": 5194
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.00016178014309781383,
+      "loss": 0.9259,
+      "step": 5195
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.00016176641956196125,
+      "loss": 0.9461,
+      "step": 5196
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.000161752694145014,
+      "loss": 0.9784,
+      "step": 5197
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.00016173896684739005,
+      "loss": 0.9957,
+      "step": 5198
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.00016172523766950752,
+      "loss": 0.7969,
+      "step": 5199
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.00016171150661178446,
+      "loss": 0.8376,
+      "step": 5200
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.00016169777367463915,
+      "loss": 0.9482,
+      "step": 5201
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.00016168403885848971,
+      "loss": 0.8955,
+      "step": 5202
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.00016167030216375448,
+      "loss": 0.997,
+      "step": 5203
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.00016165656359085184,
+      "loss": 0.8427,
+      "step": 5204
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.00016164282314020016,
+      "loss": 0.8894,
+      "step": 5205
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.00016162908081221791,
+      "loss": 0.887,
+      "step": 5206
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.00016161533660732363,
+      "loss": 0.8219,
+      "step": 5207
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.00016160159052593592,
+      "loss": 0.9379,
+      "step": 5208
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.00016158784256847333,
+      "loss": 0.8509,
+      "step": 5209
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.00016157409273535466,
+      "loss": 0.8336,
+      "step": 5210
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.00016156034102699857,
+      "loss": 0.7816,
+      "step": 5211
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.00016154658744382396,
+      "loss": 0.9197,
+      "step": 5212
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.00016153283198624963,
+      "loss": 0.8931,
+      "step": 5213
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.00016151907465469448,
+      "loss": 0.8003,
+      "step": 5214
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.00016150531544957756,
+      "loss": 0.9047,
+      "step": 5215
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.00016149155437131788,
+      "loss": 0.9058,
+      "step": 5216
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.0001614777914203345,
+      "loss": 0.9245,
+      "step": 5217
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.0001614640265970466,
+      "loss": 0.8389,
+      "step": 5218
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.0001614502599018734,
+      "loss": 1.0523,
+      "step": 5219
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.00016143649133523415,
+      "loss": 0.8963,
+      "step": 5220
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.00016142272089754816,
+      "loss": 0.8809,
+      "step": 5221
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.00016140894858923481,
+      "loss": 1.0101,
+      "step": 5222
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.00016139517441071351,
+      "loss": 1.0192,
+      "step": 5223
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.00016138139836240385,
+      "loss": 0.8497,
+      "step": 5224
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.0001613676204447253,
+      "loss": 0.8024,
+      "step": 5225
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.00016135384065809745,
+      "loss": 0.8835,
+      "step": 5226
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.00016134005900294,
+      "loss": 0.9902,
+      "step": 5227
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.00016132627547967268,
+      "loss": 0.9085,
+      "step": 5228
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.00016131249008871523,
+      "loss": 0.8834,
+      "step": 5229
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.00016129870283048753,
+      "loss": 0.8284,
+      "step": 5230
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.0001612849137054094,
+      "loss": 0.7867,
+      "step": 5231
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.00016127112271390088,
+      "loss": 0.9674,
+      "step": 5232
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.0001612573298563819,
+      "loss": 0.8745,
+      "step": 5233
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.00016124353513327256,
+      "loss": 1.0666,
+      "step": 5234
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.00016122973854499294,
+      "loss": 0.8248,
+      "step": 5235
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.00016121594009196327,
+      "loss": 0.9697,
+      "step": 5236
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.00016120213977460375,
+      "loss": 0.821,
+      "step": 5237
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.00016118833759333465,
+      "loss": 1.0557,
+      "step": 5238
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.00016117453354857634,
+      "loss": 1.0296,
+      "step": 5239
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.00016116072764074923,
+      "loss": 0.9988,
+      "step": 5240
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.00016114691987027376,
+      "loss": 0.8867,
+      "step": 5241
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.00016113311023757042,
+      "loss": 0.9955,
+      "step": 5242
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.00016111929874305987,
+      "loss": 1.0315,
+      "step": 5243
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.0001611054853871626,
+      "loss": 1.0327,
+      "step": 5244
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.00016109167017029944,
+      "loss": 0.8989,
+      "step": 5245
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.00016107785309289106,
+      "loss": 0.9118,
+      "step": 5246
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.00016106403415535827,
+      "loss": 0.8365,
+      "step": 5247
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.0001610502133581219,
+      "loss": 0.9723,
+      "step": 5248
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.00016103639070160291,
+      "loss": 0.9363,
+      "step": 5249
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.00016102256618622223,
+      "loss": 0.906,
+      "step": 5250
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.0001610087398124009,
+      "loss": 0.8797,
+      "step": 5251
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.00016099491158055998,
+      "loss": 0.8525,
+      "step": 5252
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.00016098108149112066,
+      "loss": 0.9894,
+      "step": 5253
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.00016096724954450407,
+      "loss": 0.949,
+      "step": 5254
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.00016095341574113152,
+      "loss": 0.9506,
+      "step": 5255
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.0001609395800814243,
+      "loss": 0.9127,
+      "step": 5256
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.00016092574256580376,
+      "loss": 0.8104,
+      "step": 5257
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.0001609119031946913,
+      "loss": 0.8158,
+      "step": 5258
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.0001608980619685085,
+      "loss": 0.9544,
+      "step": 5259
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.00016088421888767676,
+      "loss": 0.985,
+      "step": 5260
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.00016087037395261776,
+      "loss": 0.8301,
+      "step": 5261
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.00016085652716375307,
+      "loss": 0.9136,
+      "step": 5262
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.0001608426785215045,
+      "loss": 0.8376,
+      "step": 5263
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.00016082882802629373,
+      "loss": 0.8658,
+      "step": 5264
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.0001608149756785426,
+      "loss": 0.9984,
+      "step": 5265
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.00016080112147867295,
+      "loss": 0.943,
+      "step": 5266
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.0001607872654271068,
+      "loss": 0.9047,
+      "step": 5267
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.00016077340752426603,
+      "loss": 0.9695,
+      "step": 5268
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.00016075954777057275,
+      "loss": 0.976,
+      "step": 5269
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.00016074568616644905,
+      "loss": 0.8969,
+      "step": 5270
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.00016073182271231706,
+      "loss": 0.9857,
+      "step": 5271
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.00016071795740859901,
+      "loss": 0.9781,
+      "step": 5272
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.00016070409025571714,
+      "loss": 0.9179,
+      "step": 5273
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.0001606902212540938,
+      "loss": 0.8919,
+      "step": 5274
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.0001606763504041514,
+      "loss": 0.8824,
+      "step": 5275
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.0001606624777063123,
+      "loss": 0.9153,
+      "step": 5276
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.00016064860316099906,
+      "loss": 0.891,
+      "step": 5277
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.00016063472676863415,
+      "loss": 0.8261,
+      "step": 5278
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.00016062084852964029,
+      "loss": 0.8928,
+      "step": 5279
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.00016060696844444005,
+      "loss": 0.926,
+      "step": 5280
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.00016059308651345617,
+      "loss": 0.8576,
+      "step": 5281
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.00016057920273711144,
+      "loss": 1.0166,
+      "step": 5282
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.0001605653171158287,
+      "loss": 0.8636,
+      "step": 5283
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.0001605514296500308,
+      "loss": 0.8673,
+      "step": 5284
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.0001605375403401407,
+      "loss": 0.88,
+      "step": 5285
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.0001605236491865814,
+      "loss": 0.9749,
+      "step": 5286
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.00016050975618977598,
+      "loss": 0.8403,
+      "step": 5287
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.00016049586135014746,
+      "loss": 0.9109,
+      "step": 5288
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.00016048196466811912,
+      "loss": 0.9378,
+      "step": 5289
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.00016046806614411412,
+      "loss": 1.0463,
+      "step": 5290
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.00016045416577855574,
+      "loss": 0.7129,
+      "step": 5291
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.00016044026357186732,
+      "loss": 0.9652,
+      "step": 5292
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.00016042635952447228,
+      "loss": 0.899,
+      "step": 5293
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.00016041245363679406,
+      "loss": 0.9587,
+      "step": 5294
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.0001603985459092561,
+      "loss": 0.7884,
+      "step": 5295
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.00016038463634228208,
+      "loss": 0.9822,
+      "step": 5296
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.0001603707249362955,
+      "loss": 1.0292,
+      "step": 5297
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.00016035681169172008,
+      "loss": 0.9448,
+      "step": 5298
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.00016034289660897956,
+      "loss": 0.7097,
+      "step": 5299
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.0001603289796884977,
+      "loss": 0.9778,
+      "step": 5300
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.00016031506093069835,
+      "loss": 0.8649,
+      "step": 5301
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.00016030114033600539,
+      "loss": 0.8775,
+      "step": 5302
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.00016028721790484278,
+      "loss": 0.9589,
+      "step": 5303
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.00016027329363763455,
+      "loss": 0.867,
+      "step": 5304
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.00016025936753480473,
+      "loss": 0.9462,
+      "step": 5305
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.00016024543959677746,
+      "loss": 0.7878,
+      "step": 5306
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.0001602315098239769,
+      "loss": 0.958,
+      "step": 5307
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.0001602175782168273,
+      "loss": 0.9909,
+      "step": 5308
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.00016020364477575293,
+      "loss": 0.898,
+      "step": 5309
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.00016018970950117814,
+      "loss": 0.9689,
+      "step": 5310
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.00016017577239352732,
+      "loss": 1.0439,
+      "step": 5311
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.00016016183345322495,
+      "loss": 0.8769,
+      "step": 5312
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.0001601478926806955,
+      "loss": 0.8519,
+      "step": 5313
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.00016013395007636358,
+      "loss": 1.0841,
+      "step": 5314
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.00016012000564065377,
+      "loss": 0.8847,
+      "step": 5315
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.00016010605937399078,
+      "loss": 0.9419,
+      "step": 5316
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.00016009211127679932,
+      "loss": 0.8075,
+      "step": 5317
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.0001600781613495042,
+      "loss": 0.9315,
+      "step": 5318
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.00016006420959253023,
+      "loss": 0.9882,
+      "step": 5319
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.00016005025600630233,
+      "loss": 0.9364,
+      "step": 5320
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.00016003630059124548,
+      "loss": 0.9024,
+      "step": 5321
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.00016002234334778467,
+      "loss": 0.9966,
+      "step": 5322
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.00016000838427634496,
+      "loss": 0.8859,
+      "step": 5323
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.0001599944233773515,
+      "loss": 0.81,
+      "step": 5324
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.00015998046065122942,
+      "loss": 0.9073,
+      "step": 5325
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.00015996649609840402,
+      "loss": 0.9665,
+      "step": 5326
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.00015995252971930053,
+      "loss": 0.8819,
+      "step": 5327
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.0001599385615143443,
+      "loss": 0.9263,
+      "step": 5328
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.0001599245914839608,
+      "loss": 0.8689,
+      "step": 5329
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.00015991061962857542,
+      "loss": 0.9123,
+      "step": 5330
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.00015989664594861372,
+      "loss": 0.9818,
+      "step": 5331
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.0001598826704445012,
+      "loss": 1.0513,
+      "step": 5332
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.00015986869311666353,
+      "loss": 0.7967,
+      "step": 5333
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.0001598547139655264,
+      "loss": 0.9167,
+      "step": 5334
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.0001598407329915155,
+      "loss": 0.9262,
+      "step": 5335
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.00015982675019505667,
+      "loss": 0.8556,
+      "step": 5336
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.00015981276557657572,
+      "loss": 0.8347,
+      "step": 5337
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.00015979877913649857,
+      "loss": 0.8458,
+      "step": 5338
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.00015978479087525117,
+      "loss": 1.0757,
+      "step": 5339
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.00015977080079325955,
+      "loss": 0.9221,
+      "step": 5340
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.00015975680889094974,
+      "loss": 0.8828,
+      "step": 5341
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.00015974281516874787,
+      "loss": 0.8667,
+      "step": 5342
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.00015972881962708017,
+      "loss": 0.9916,
+      "step": 5343
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.00015971482226637278,
+      "loss": 0.8532,
+      "step": 5344
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.0001597008230870521,
+      "loss": 0.9242,
+      "step": 5345
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.0001596868220895444,
+      "loss": 1.0165,
+      "step": 5346
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.0001596728192742761,
+      "loss": 0.9427,
+      "step": 5347
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.00015965881464167367,
+      "loss": 0.9031,
+      "step": 5348
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.00015964480819216357,
+      "loss": 0.9663,
+      "step": 5349
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.00015963079992617244,
+      "loss": 0.9457,
+      "step": 5350
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.00015961678984412685,
+      "loss": 0.9755,
+      "step": 5351
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.0001596027779464535,
+      "loss": 0.9224,
+      "step": 5352
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.00015958876423357907,
+      "loss": 0.979,
+      "step": 5353
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.00015957474870593042,
+      "loss": 0.8691,
+      "step": 5354
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.00015956073136393436,
+      "loss": 0.7227,
+      "step": 5355
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.0001595467122080178,
+      "loss": 0.9139,
+      "step": 5356
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.00015953269123860768,
+      "loss": 0.9658,
+      "step": 5357
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.00015951866845613098,
+      "loss": 0.9789,
+      "step": 5358
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.00015950464386101483,
+      "loss": 0.9876,
+      "step": 5359
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.0001594906174536863,
+      "loss": 0.9084,
+      "step": 5360
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.00015947658923457257,
+      "loss": 0.8512,
+      "step": 5361
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.00015946255920410088,
+      "loss": 0.9015,
+      "step": 5362
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.00015944852736269854,
+      "loss": 0.9151,
+      "step": 5363
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.0001594344937107928,
+      "loss": 0.9138,
+      "step": 5364
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.00015942045824881115,
+      "loss": 0.8733,
+      "step": 5365
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.00015940642097718102,
+      "loss": 0.9549,
+      "step": 5366
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.0001593923818963299,
+      "loss": 0.895,
+      "step": 5367
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.00015937834100668532,
+      "loss": 0.9055,
+      "step": 5368
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.0001593642983086749,
+      "loss": 0.8958,
+      "step": 5369
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.00015935025380272636,
+      "loss": 0.8605,
+      "step": 5370
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.0001593362074892674,
+      "loss": 0.8319,
+      "step": 5371
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.0001593221593687258,
+      "loss": 0.8435,
+      "step": 5372
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.00015930810944152942,
+      "loss": 0.9086,
+      "step": 5373
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.00015929405770810607,
+      "loss": 0.8567,
+      "step": 5374
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.00015928000416888378,
+      "loss": 0.9928,
+      "step": 5375
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.0001592659488242905,
+      "loss": 0.8915,
+      "step": 5376
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.00015925189167475433,
+      "loss": 0.8991,
+      "step": 5377
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.00015923783272070335,
+      "loss": 0.8549,
+      "step": 5378
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.00015922377196256573,
+      "loss": 0.8364,
+      "step": 5379
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.00015920970940076968,
+      "loss": 1.0051,
+      "step": 5380
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.00015919564503574346,
+      "loss": 0.9394,
+      "step": 5381
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.00015918157886791548,
+      "loss": 0.9214,
+      "step": 5382
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.00015916751089771402,
+      "loss": 0.9477,
+      "step": 5383
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.0001591534411255676,
+      "loss": 0.8964,
+      "step": 5384
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.00015913936955190468,
+      "loss": 0.84,
+      "step": 5385
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.0001591252961771538,
+      "loss": 1.0252,
+      "step": 5386
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.00015911122100174357,
+      "loss": 0.958,
+      "step": 5387
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.00015909714402610267,
+      "loss": 0.9238,
+      "step": 5388
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.0001590830652506598,
+      "loss": 0.9286,
+      "step": 5389
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.0001590689846758437,
+      "loss": 0.933,
+      "step": 5390
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.00015905490230208325,
+      "loss": 0.8368,
+      "step": 5391
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.0001590408181298073,
+      "loss": 0.902,
+      "step": 5392
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.0001590267321594448,
+      "loss": 0.9439,
+      "step": 5393
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.0001590126443914247,
+      "loss": 0.9529,
+      "step": 5394
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.00015899855482617606,
+      "loss": 0.9156,
+      "step": 5395
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.00015898446346412798,
+      "loss": 1.0174,
+      "step": 5396
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.00015897037030570963,
+      "loss": 0.9425,
+      "step": 5397
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.00015895627535135017,
+      "loss": 0.9681,
+      "step": 5398
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.0001589421786014789,
+      "loss": 0.8936,
+      "step": 5399
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.00015892808005652515,
+      "loss": 0.7582,
+      "step": 5400
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.00015891397971691827,
+      "loss": 0.903,
+      "step": 5401
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.00015889987758308766,
+      "loss": 0.9748,
+      "step": 5402
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.00015888577365546285,
+      "loss": 0.9666,
+      "step": 5403
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.00015887166793447332,
+      "loss": 0.8631,
+      "step": 5404
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.0001588575604205487,
+      "loss": 0.8299,
+      "step": 5405
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.00015884345111411862,
+      "loss": 0.9312,
+      "step": 5406
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.0001588293400156128,
+      "loss": 1.0005,
+      "step": 5407
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.00015881522712546097,
+      "loss": 0.7655,
+      "step": 5408
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.00015880111244409294,
+      "loss": 0.9887,
+      "step": 5409
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.00015878699597193853,
+      "loss": 0.8659,
+      "step": 5410
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.00015877287770942777,
+      "loss": 0.9673,
+      "step": 5411
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.0001587587576569905,
+      "loss": 1.0595,
+      "step": 5412
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.00015874463581505682,
+      "loss": 0.9206,
+      "step": 5413
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.00015873051218405685,
+      "loss": 0.8724,
+      "step": 5414
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.0001587163867644206,
+      "loss": 0.8969,
+      "step": 5415
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.0001587022595565784,
+      "loss": 0.8577,
+      "step": 5416
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.00015868813056096038,
+      "loss": 0.9605,
+      "step": 5417
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.0001586739997779969,
+      "loss": 0.8199,
+      "step": 5418
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.0001586598672081183,
+      "loss": 0.9354,
+      "step": 5419
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.00015864573285175499,
+      "loss": 0.7815,
+      "step": 5420
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.0001586315967093374,
+      "loss": 0.851,
+      "step": 5421
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.00015861745878129608,
+      "loss": 0.8738,
+      "step": 5422
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.0001586033190680616,
+      "loss": 1.0524,
+      "step": 5423
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.00015858917757006458,
+      "loss": 1.0104,
+      "step": 5424
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.00015857503428773567,
+      "loss": 1.0179,
+      "step": 5425
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.00015856088922150567,
+      "loss": 0.916,
+      "step": 5426
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.0001585467423718053,
+      "loss": 0.9161,
+      "step": 5427
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.00015853259373906542,
+      "loss": 0.9178,
+      "step": 5428
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.00015851844332371695,
+      "loss": 0.9197,
+      "step": 5429
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.0001585042911261908,
+      "loss": 0.9609,
+      "step": 5430
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.00015849013714691802,
+      "loss": 0.8598,
+      "step": 5431
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.00015847598138632967,
+      "loss": 0.7256,
+      "step": 5432
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.0001584618238448568,
+      "loss": 1.0257,
+      "step": 5433
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.00015844766452293064,
+      "loss": 0.8913,
+      "step": 5434
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.00015843350342098237,
+      "loss": 1.036,
+      "step": 5435
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.00015841934053944333,
+      "loss": 0.8074,
+      "step": 5436
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.00015840517587874478,
+      "loss": 0.9608,
+      "step": 5437
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.00015839100943931814,
+      "loss": 0.9288,
+      "step": 5438
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.00015837684122159483,
+      "loss": 0.8922,
+      "step": 5439
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.00015836267122600634,
+      "loss": 0.8484,
+      "step": 5440
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.00015834849945298426,
+      "loss": 0.9977,
+      "step": 5441
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.00015833432590296013,
+      "loss": 0.9757,
+      "step": 5442
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.00015832015057636566,
+      "loss": 0.8564,
+      "step": 5443
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.0001583059734736325,
+      "loss": 1.0416,
+      "step": 5444
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.0001582917945951925,
+      "loss": 0.8187,
+      "step": 5445
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.0001582776139414774,
+      "loss": 0.7608,
+      "step": 5446
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.00015826343151291908,
+      "loss": 1.1916,
+      "step": 5447
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.00015824924730994952,
+      "loss": 0.9525,
+      "step": 5448
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.00015823506133300064,
+      "loss": 0.9046,
+      "step": 5449
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.0001582208735825045,
+      "loss": 0.9304,
+      "step": 5450
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.00015820668405889317,
+      "loss": 0.905,
+      "step": 5451
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.00015819249276259882,
+      "loss": 0.9668,
+      "step": 5452
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.00015817829969405362,
+      "loss": 0.9244,
+      "step": 5453
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.00015816410485368983,
+      "loss": 1.0589,
+      "step": 5454
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.00015814990824193978,
+      "loss": 0.9034,
+      "step": 5455
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.00015813570985923577,
+      "loss": 0.8793,
+      "step": 5456
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.00015812150970601025,
+      "loss": 0.906,
+      "step": 5457
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.0001581073077826957,
+      "loss": 0.8787,
+      "step": 5458
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.00015809310408972457,
+      "loss": 0.9307,
+      "step": 5459
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.00015807889862752954,
+      "loss": 0.9044,
+      "step": 5460
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.00015806469139654312,
+      "loss": 0.9842,
+      "step": 5461
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.00015805048239719805,
+      "loss": 0.9051,
+      "step": 5462
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.0001580362716299271,
+      "loss": 1.1024,
+      "step": 5463
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.00015802205909516302,
+      "loss": 0.7969,
+      "step": 5464
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.0001580078447933386,
+      "loss": 0.9051,
+      "step": 5465
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.00015799362872488684,
+      "loss": 0.9185,
+      "step": 5466
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.00015797941089024065,
+      "loss": 0.865,
+      "step": 5467
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.00015796519128983297,
+      "loss": 0.9167,
+      "step": 5468
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.00015795096992409695,
+      "loss": 0.8361,
+      "step": 5469
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.00015793674679346565,
+      "loss": 0.9166,
+      "step": 5470
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.00015792252189837224,
+      "loss": 0.9535,
+      "step": 5471
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.00015790829523924997,
+      "loss": 0.9725,
+      "step": 5472
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.00015789406681653205,
+      "loss": 0.7862,
+      "step": 5473
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.00015787983663065187,
+      "loss": 0.9102,
+      "step": 5474
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.00015786560468204278,
+      "loss": 1.0002,
+      "step": 5475
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.00015785137097113822,
+      "loss": 0.8736,
+      "step": 5476
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.00015783713549837166,
+      "loss": 0.8985,
+      "step": 5477
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.0001578228982641767,
+      "loss": 0.9423,
+      "step": 5478
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.00015780865926898685,
+      "loss": 0.8983,
+      "step": 5479
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.00015779441851323579,
+      "loss": 0.7906,
+      "step": 5480
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.00015778017599735726,
+      "loss": 0.9076,
+      "step": 5481
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.000157765931721785,
+      "loss": 0.8334,
+      "step": 5482
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.00015775168568695278,
+      "loss": 0.9439,
+      "step": 5483
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.00015773743789329446,
+      "loss": 1.0933,
+      "step": 5484
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.00015772318834124404,
+      "loss": 0.8588,
+      "step": 5485
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.0001577089370312354,
+      "loss": 0.9073,
+      "step": 5486
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.00015769468396370262,
+      "loss": 0.9832,
+      "step": 5487
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.00015768042913907974,
+      "loss": 0.9537,
+      "step": 5488
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.00015766617255780088,
+      "loss": 0.8951,
+      "step": 5489
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.0001576519142203003,
+      "loss": 0.9398,
+      "step": 5490
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.00015763765412701215,
+      "loss": 0.9599,
+      "step": 5491
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.00015762339227837078,
+      "loss": 1.0122,
+      "step": 5492
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.0001576091286748105,
+      "loss": 0.8847,
+      "step": 5493
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.00015759486331676573,
+      "loss": 0.8693,
+      "step": 5494
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.00015758059620467088,
+      "loss": 0.995,
+      "step": 5495
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.00015756632733896048,
+      "loss": 0.8746,
+      "step": 5496
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.0001575520567200691,
+      "loss": 0.7847,
+      "step": 5497
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.00015753778434843136,
+      "loss": 0.8516,
+      "step": 5498
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.0001575235102244819,
+      "loss": 0.8561,
+      "step": 5499
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.00015750923434865546,
+      "loss": 0.9095,
+      "step": 5500
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.00015749495672138679,
+      "loss": 0.8852,
+      "step": 5501
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.0001574806773431107,
+      "loss": 0.8455,
+      "step": 5502
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.00015746639621426212,
+      "loss": 0.9562,
+      "step": 5503
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.00015745211333527593,
+      "loss": 0.9242,
+      "step": 5504
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.00015743782870658715,
+      "loss": 0.8965,
+      "step": 5505
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.0001574235423286308,
+      "loss": 0.8726,
+      "step": 5506
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.000157409254201842,
+      "loss": 0.9021,
+      "step": 5507
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.00015739496432665582,
+      "loss": 0.8603,
+      "step": 5508
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.00015738067270350755,
+      "loss": 0.9419,
+      "step": 5509
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.00015736637933283238,
+      "loss": 0.8902,
+      "step": 5510
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.00015735208421506562,
+      "loss": 0.9582,
+      "step": 5511
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.00015733778735064265,
+      "loss": 0.8971,
+      "step": 5512
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.0001573234887399989,
+      "loss": 0.893,
+      "step": 5513
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.0001573091883835698,
+      "loss": 0.9505,
+      "step": 5514
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.0001572948862817908,
+      "loss": 0.9869,
+      "step": 5515
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.0001572805824350976,
+      "loss": 0.8882,
+      "step": 5516
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.00015726627684392578,
+      "loss": 0.8669,
+      "step": 5517
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.00015725196950871099,
+      "loss": 0.9454,
+      "step": 5518
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.00015723766042988896,
+      "loss": 0.9631,
+      "step": 5519
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.00015722334960789546,
+      "loss": 0.9498,
+      "step": 5520
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.0001572090370431664,
+      "loss": 0.8466,
+      "step": 5521
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.00015719472273613758,
+      "loss": 0.9855,
+      "step": 5522
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.000157180406687245,
+      "loss": 0.9099,
+      "step": 5523
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.00015716608889692463,
+      "loss": 0.976,
+      "step": 5524
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.00015715176936561255,
+      "loss": 0.9025,
+      "step": 5525
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.0001571374480937448,
+      "loss": 0.8477,
+      "step": 5526
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.00015712312508175757,
+      "loss": 1.0102,
+      "step": 5527
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.00015710880033008708,
+      "loss": 0.9322,
+      "step": 5528
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.00015709447383916957,
+      "loss": 0.988,
+      "step": 5529
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.00015708014560944134,
+      "loss": 0.865,
+      "step": 5530
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.0001570658156413388,
+      "loss": 0.8225,
+      "step": 5531
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.0001570514839352983,
+      "loss": 0.8613,
+      "step": 5532
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.0001570371504917564,
+      "loss": 1.0098,
+      "step": 5533
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.00015702281531114958,
+      "loss": 0.8177,
+      "step": 5534
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.0001570084783939144,
+      "loss": 0.932,
+      "step": 5535
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.0001569941397404875,
+      "loss": 0.9517,
+      "step": 5536
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.00015697979935130557,
+      "loss": 0.9353,
+      "step": 5537
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.00015696545722680534,
+      "loss": 0.9335,
+      "step": 5538
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.0001569511133674236,
+      "loss": 1.0008,
+      "step": 5539
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.00015693676777359724,
+      "loss": 0.9644,
+      "step": 5540
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.0001569224204457631,
+      "loss": 0.9847,
+      "step": 5541
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.00015690807138435812,
+      "loss": 0.9018,
+      "step": 5542
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.00015689372058981933,
+      "loss": 1.0134,
+      "step": 5543
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.00015687936806258373,
+      "loss": 0.8884,
+      "step": 5544
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.00015686501380308854,
+      "loss": 0.9139,
+      "step": 5545
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.0001568506578117708,
+      "loss": 0.9019,
+      "step": 5546
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.0001568363000890678,
+      "loss": 0.9643,
+      "step": 5547
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.00015682194063541675,
+      "loss": 0.8847,
+      "step": 5548
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.00015680757945125498,
+      "loss": 0.8256,
+      "step": 5549
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.00015679321653701988,
+      "loss": 0.9046,
+      "step": 5550
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.00015677885189314887,
+      "loss": 0.9307,
+      "step": 5551
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.00015676448552007943,
+      "loss": 0.8909,
+      "step": 5552
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.00015675011741824902,
+      "loss": 0.8938,
+      "step": 5553
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.00015673574758809532,
+      "loss": 0.9889,
+      "step": 5554
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.00015672137603005594,
+      "loss": 1.0526,
+      "step": 5555
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.00015670700274456848,
+      "loss": 0.8619,
+      "step": 5556
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.00015669262773207075,
+      "loss": 0.8867,
+      "step": 5557
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.00015667825099300057,
+      "loss": 0.9122,
+      "step": 5558
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.0001566638725277957,
+      "loss": 0.9208,
+      "step": 5559
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.0001566494923368941,
+      "loss": 0.8343,
+      "step": 5560
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.0001566351104207337,
+      "loss": 0.9991,
+      "step": 5561
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.00015662072677975248,
+      "loss": 1.0607,
+      "step": 5562
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.0001566063414143885,
+      "loss": 0.8123,
+      "step": 5563
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.00015659195432507993,
+      "loss": 0.956,
+      "step": 5564
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.00015657756551226487,
+      "loss": 0.915,
+      "step": 5565
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.0001565631749763815,
+      "loss": 0.9148,
+      "step": 5566
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.00015654878271786815,
+      "loss": 0.8955,
+      "step": 5567
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.0001565343887371631,
+      "loss": 0.9764,
+      "step": 5568
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.00015651999303470472,
+      "loss": 0.9294,
+      "step": 5569
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.00015650559561093143,
+      "loss": 0.9901,
+      "step": 5570
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.00015649119646628172,
+      "loss": 0.934,
+      "step": 5571
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.0001564767956011941,
+      "loss": 0.9394,
+      "step": 5572
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.00015646239301610713,
+      "loss": 0.8453,
+      "step": 5573
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.00015644798871145948,
+      "loss": 0.9925,
+      "step": 5574
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.00015643358268768983,
+      "loss": 0.9105,
+      "step": 5575
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.0001564191749452369,
+      "loss": 0.8028,
+      "step": 5576
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.00015640476548453946,
+      "loss": 1.0011,
+      "step": 5577
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.00015639035430603637,
+      "loss": 0.9552,
+      "step": 5578
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.0001563759414101665,
+      "loss": 1.0339,
+      "step": 5579
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.00015636152679736887,
+      "loss": 0.8627,
+      "step": 5580
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.0001563471104680824,
+      "loss": 0.897,
+      "step": 5581
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.00015633269242274614,
+      "loss": 0.946,
+      "step": 5582
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.00015631827266179924,
+      "loss": 0.9486,
+      "step": 5583
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.00015630385118568077,
+      "loss": 0.8412,
+      "step": 5584
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.00015628942799483003,
+      "loss": 0.9922,
+      "step": 5585
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.00015627500308968624,
+      "loss": 1.053,
+      "step": 5586
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.0001562605764706887,
+      "loss": 0.9462,
+      "step": 5587
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.00015624614813827679,
+      "loss": 0.8849,
+      "step": 5588
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.00015623171809288987,
+      "loss": 0.9318,
+      "step": 5589
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.0001562172863349675,
+      "loss": 0.9051,
+      "step": 5590
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.00015620285286494916,
+      "loss": 1.0084,
+      "step": 5591
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.00015618841768327436,
+      "loss": 0.8614,
+      "step": 5592
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.0001561739807903828,
+      "loss": 0.8843,
+      "step": 5593
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.00015615954218671412,
+      "loss": 0.8957,
+      "step": 5594
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.0001561451018727081,
+      "loss": 0.9959,
+      "step": 5595
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.0001561306598488044,
+      "loss": 0.806,
+      "step": 5596
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.000156116216115443,
+      "loss": 0.8164,
+      "step": 5597
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.00015610177067306368,
+      "loss": 0.9999,
+      "step": 5598
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.00015608732352210641,
+      "loss": 0.9912,
+      "step": 5599
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.00015607287466301117,
+      "loss": 0.8446,
+      "step": 5600
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.000156058424096218,
+      "loss": 0.9167,
+      "step": 5601
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.00015604397182216703,
+      "loss": 0.8708,
+      "step": 5602
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.00015602951784129835,
+      "loss": 0.8857,
+      "step": 5603
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.00015601506215405216,
+      "loss": 1.0167,
+      "step": 5604
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.00015600060476086874,
+      "loss": 0.9177,
+      "step": 5605
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.00015598614566218833,
+      "loss": 0.8127,
+      "step": 5606
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.00015597168485845136,
+      "loss": 0.9679,
+      "step": 5607
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.0001559572223500982,
+      "loss": 0.8329,
+      "step": 5608
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.00015594275813756928,
+      "loss": 1.0816,
+      "step": 5609
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.00015592829222130513,
+      "loss": 1.0455,
+      "step": 5610
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.00015591382460174632,
+      "loss": 0.7496,
+      "step": 5611
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.0001558993552793335,
+      "loss": 0.8126,
+      "step": 5612
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.0001558848842545072,
+      "loss": 0.9344,
+      "step": 5613
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.00015587041152770824,
+      "loss": 0.9263,
+      "step": 5614
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.0001558559370993774,
+      "loss": 0.8858,
+      "step": 5615
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.00015584146096995542,
+      "loss": 0.885,
+      "step": 5616
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.00015582698313988322,
+      "loss": 1.0627,
+      "step": 5617
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.00015581250360960174,
+      "loss": 0.992,
+      "step": 5618
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.00015579802237955187,
+      "loss": 0.9254,
+      "step": 5619
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.0001557835394501747,
+      "loss": 0.8487,
+      "step": 5620
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.00015576905482191138,
+      "loss": 1.0229,
+      "step": 5621
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.00015575456849520293,
+      "loss": 0.8241,
+      "step": 5622
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.0001557400804704905,
+      "loss": 0.9781,
+      "step": 5623
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.00015572559074821543,
+      "loss": 0.8563,
+      "step": 5624
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.00015571109932881893,
+      "loss": 0.8052,
+      "step": 5625
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.00015569660621274237,
+      "loss": 0.8875,
+      "step": 5626
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.00015568211140042716,
+      "loss": 0.9918,
+      "step": 5627
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.0001556676148923147,
+      "loss": 0.8976,
+      "step": 5628
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.00015565311668884647,
+      "loss": 0.874,
+      "step": 5629
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.00015563861679046404,
+      "loss": 0.8138,
+      "step": 5630
+    },
+    {
+      "epoch": 1.23,
+      "eval_loss": 0.5862651467323303,
+      "eval_runtime": 321.9919,
+      "eval_samples_per_second": 56.334,
+      "eval_steps_per_second": 28.168,
+      "step": 5630
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.000155624115197609,
+      "loss": 0.9215,
+      "step": 5631
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.00015560961191072296,
+      "loss": 0.9262,
+      "step": 5632
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.00015559510693024771,
+      "loss": 1.0256,
+      "step": 5633
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.0001555806002566249,
+      "loss": 0.7909,
+      "step": 5634
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.00015556609189029636,
+      "loss": 1.106,
+      "step": 5635
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.00015555158183170397,
+      "loss": 0.9208,
+      "step": 5636
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.0001555370700812896,
+      "loss": 0.8376,
+      "step": 5637
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.00015552255663949522,
+      "loss": 0.9253,
+      "step": 5638
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.00015550804150676285,
+      "loss": 0.9464,
+      "step": 5639
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.0001554935246835345,
+      "loss": 0.8994,
+      "step": 5640
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.00015547900617025233,
+      "loss": 0.8354,
+      "step": 5641
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.0001554644859673585,
+      "loss": 0.8171,
+      "step": 5642
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.00015544996407529518,
+      "loss": 1.0066,
+      "step": 5643
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.0001554354404945047,
+      "loss": 0.8743,
+      "step": 5644
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.0001554209152254293,
+      "loss": 0.9675,
+      "step": 5645
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.00015540638826851136,
+      "loss": 0.9203,
+      "step": 5646
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.00015539185962419336,
+      "loss": 0.8939,
+      "step": 5647
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.0001553773292929177,
+      "loss": 1.1121,
+      "step": 5648
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.00015536279727512695,
+      "loss": 0.9194,
+      "step": 5649
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.00015534826357126367,
+      "loss": 0.9445,
+      "step": 5650
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.00015533372818177048,
+      "loss": 0.9756,
+      "step": 5651
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.00015531919110709001,
+      "loss": 0.9368,
+      "step": 5652
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.00015530465234766505,
+      "loss": 0.9654,
+      "step": 5653
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.00015529011190393836,
+      "loss": 0.9205,
+      "step": 5654
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.00015527556977635275,
+      "loss": 1.0137,
+      "step": 5655
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.00015526102596535112,
+      "loss": 0.8851,
+      "step": 5656
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.0001552464804713764,
+      "loss": 0.9419,
+      "step": 5657
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.00015523193329487157,
+      "loss": 0.9371,
+      "step": 5658
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.00015521738443627963,
+      "loss": 1.0188,
+      "step": 5659
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.00015520283389604372,
+      "loss": 0.9354,
+      "step": 5660
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.00015518828167460698,
+      "loss": 1.1186,
+      "step": 5661
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.00015517372777241252,
+      "loss": 0.9402,
+      "step": 5662
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.00015515917218990368,
+      "loss": 0.9515,
+      "step": 5663
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.00015514461492752366,
+      "loss": 1.0708,
+      "step": 5664
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.00015513005598571583,
+      "loss": 1.0091,
+      "step": 5665
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.00015511549536492364,
+      "loss": 0.9284,
+      "step": 5666
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.00015510093306559045,
+      "loss": 0.9708,
+      "step": 5667
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.0001550863690881598,
+      "loss": 0.9175,
+      "step": 5668
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.0001550718034330752,
+      "loss": 0.9287,
+      "step": 5669
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.00015505723610078031,
+      "loss": 0.9267,
+      "step": 5670
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.0001550426670917187,
+      "loss": 0.9727,
+      "step": 5671
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.00015502809640633415,
+      "loss": 0.8616,
+      "step": 5672
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.00015501352404507037,
+      "loss": 0.8137,
+      "step": 5673
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.00015499895000837115,
+      "loss": 0.9965,
+      "step": 5674
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.00015498437429668034,
+      "loss": 0.9579,
+      "step": 5675
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.00015496979691044182,
+      "loss": 0.8894,
+      "step": 5676
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.0001549552178500996,
+      "loss": 0.9183,
+      "step": 5677
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.0001549406371160977,
+      "loss": 1.0409,
+      "step": 5678
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.00015492605470888014,
+      "loss": 0.9812,
+      "step": 5679
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.00015491147062889098,
+      "loss": 0.9613,
+      "step": 5680
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.00015489688487657443,
+      "loss": 0.9246,
+      "step": 5681
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.0001548822974523747,
+      "loss": 0.9384,
+      "step": 5682
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.00015486770835673605,
+      "loss": 0.8849,
+      "step": 5683
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.0001548531175901028,
+      "loss": 0.9415,
+      "step": 5684
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.00015483852515291928,
+      "loss": 0.9752,
+      "step": 5685
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.0001548239310456299,
+      "loss": 0.8645,
+      "step": 5686
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.0001548093352686791,
+      "loss": 0.8526,
+      "step": 5687
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.0001547947378225115,
+      "loss": 0.9116,
+      "step": 5688
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.00015478013870757158,
+      "loss": 0.8279,
+      "step": 5689
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.000154765537924304,
+      "loss": 1.0268,
+      "step": 5690
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.00015475093547315333,
+      "loss": 0.9931,
+      "step": 5691
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.00015473633135456444,
+      "loss": 0.8177,
+      "step": 5692
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.00015472172556898196,
+      "loss": 1.0039,
+      "step": 5693
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.00015470711811685077,
+      "loss": 0.9128,
+      "step": 5694
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.00015469250899861575,
+      "loss": 0.9164,
+      "step": 5695
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.0001546778982147218,
+      "loss": 0.9661,
+      "step": 5696
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.0001546632857656139,
+      "loss": 0.9537,
+      "step": 5697
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.000154648671651737,
+      "loss": 0.8737,
+      "step": 5698
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.00015463405587353633,
+      "loss": 0.9862,
+      "step": 5699
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.00015461943843145688,
+      "loss": 0.8895,
+      "step": 5700
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.00015460481932594387,
+      "loss": 0.9932,
+      "step": 5701
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.00015459019855744254,
+      "loss": 0.8936,
+      "step": 5702
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.00015457557612639814,
+      "loss": 0.9275,
+      "step": 5703
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.00015456095203325595,
+      "loss": 0.9763,
+      "step": 5704
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.00015454632627846145,
+      "loss": 1.0055,
+      "step": 5705
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.00015453169886246002,
+      "loss": 0.8877,
+      "step": 5706
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.00015451706978569712,
+      "loss": 1.0514,
+      "step": 5707
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.00015450243904861828,
+      "loss": 0.8837,
+      "step": 5708
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.0001544878066516691,
+      "loss": 0.8627,
+      "step": 5709
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.0001544731725952952,
+      "loss": 0.9424,
+      "step": 5710
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.00015445853687994223,
+      "loss": 0.9313,
+      "step": 5711
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.00015444389950605597,
+      "loss": 0.8611,
+      "step": 5712
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.0001544292604740822,
+      "loss": 0.8401,
+      "step": 5713
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.0001544146197844667,
+      "loss": 0.7889,
+      "step": 5714
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.0001543999774376554,
+      "loss": 0.9191,
+      "step": 5715
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.00015438533343409424,
+      "loss": 0.9421,
+      "step": 5716
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.00015437068777422912,
+      "loss": 0.8789,
+      "step": 5717
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.00015435604045850614,
+      "loss": 0.8968,
+      "step": 5718
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.00015434139148737144,
+      "loss": 0.995,
+      "step": 5719
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.00015432674086127103,
+      "loss": 0.9059,
+      "step": 5720
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.00015431208858065115,
+      "loss": 0.8206,
+      "step": 5721
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.00015429743464595808,
+      "loss": 0.9301,
+      "step": 5722
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.000154282779057638,
+      "loss": 0.8678,
+      "step": 5723
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.00015426812181613736,
+      "loss": 0.8164,
+      "step": 5724
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.00015425346292190247,
+      "loss": 1.0249,
+      "step": 5725
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.0001542388023753798,
+      "loss": 0.8561,
+      "step": 5726
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.00015422414017701583,
+      "loss": 0.9349,
+      "step": 5727
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.0001542094763272571,
+      "loss": 1.0299,
+      "step": 5728
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.00015419481082655016,
+      "loss": 0.8454,
+      "step": 5729
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.00015418014367534172,
+      "loss": 0.8653,
+      "step": 5730
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.0001541654748740784,
+      "loss": 0.8524,
+      "step": 5731
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.000154150804423207,
+      "loss": 0.8673,
+      "step": 5732
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.0001541361323231742,
+      "loss": 0.896,
+      "step": 5733
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.00015412145857442698,
+      "loss": 0.973,
+      "step": 5734
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.00015410678317741213,
+      "loss": 0.8654,
+      "step": 5735
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.00015409210613257662,
+      "loss": 0.9102,
+      "step": 5736
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.00015407742744036744,
+      "loss": 0.8375,
+      "step": 5737
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.00015406274710123163,
+      "loss": 0.884,
+      "step": 5738
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.0001540480651156163,
+      "loss": 0.9892,
+      "step": 5739
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.0001540333814839685,
+      "loss": 0.9569,
+      "step": 5740
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.00015401869620673553,
+      "loss": 0.9906,
+      "step": 5741
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.00015400400928436458,
+      "loss": 1.0403,
+      "step": 5742
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.00015398932071730297,
+      "loss": 1.0618,
+      "step": 5743
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.00015397463050599797,
+      "loss": 0.9794,
+      "step": 5744
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.000153959938650897,
+      "loss": 0.8684,
+      "step": 5745
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.00015394524515244758,
+      "loss": 0.8521,
+      "step": 5746
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.0001539305500110971,
+      "loss": 0.9872,
+      "step": 5747
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.00015391585322729317,
+      "loss": 0.917,
+      "step": 5748
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.00015390115480148328,
+      "loss": 0.8823,
+      "step": 5749
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.00015388645473411516,
+      "loss": 0.8982,
+      "step": 5750
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.00015387175302563648,
+      "loss": 0.9408,
+      "step": 5751
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.00015385704967649496,
+      "loss": 0.9459,
+      "step": 5752
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.00015384234468713846,
+      "loss": 0.8333,
+      "step": 5753
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.00015382763805801472,
+      "loss": 0.7788,
+      "step": 5754
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.00015381292978957167,
+      "loss": 0.8818,
+      "step": 5755
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.00015379821988225728,
+      "loss": 0.8337,
+      "step": 5756
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.0001537835083365195,
+      "loss": 0.8458,
+      "step": 5757
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.0001537687951528064,
+      "loss": 0.9072,
+      "step": 5758
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.00015375408033156606,
+      "loss": 0.9323,
+      "step": 5759
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.0001537393638732466,
+      "loss": 1.046,
+      "step": 5760
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.00015372464577829623,
+      "loss": 0.9242,
+      "step": 5761
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.00015370992604716313,
+      "loss": 0.8764,
+      "step": 5762
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.00015369520468029571,
+      "loss": 0.9338,
+      "step": 5763
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.00015368048167814223,
+      "loss": 0.8002,
+      "step": 5764
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.00015366575704115108,
+      "loss": 0.8774,
+      "step": 5765
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.0001536510307697707,
+      "loss": 0.9768,
+      "step": 5766
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.0001536363028644496,
+      "loss": 0.8969,
+      "step": 5767
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.00015362157332563629,
+      "loss": 0.8752,
+      "step": 5768
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.0001536068421537794,
+      "loss": 0.8869,
+      "step": 5769
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.0001535921093493275,
+      "loss": 0.9356,
+      "step": 5770
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.00015357737491272934,
+      "loss": 0.8854,
+      "step": 5771
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.00015356263884443364,
+      "loss": 0.9687,
+      "step": 5772
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.00015354790114488918,
+      "loss": 0.8929,
+      "step": 5773
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.00015353316181454475,
+      "loss": 0.7583,
+      "step": 5774
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.00015351842085384933,
+      "loss": 1.0115,
+      "step": 5775
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.00015350367826325178,
+      "loss": 1.0119,
+      "step": 5776
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.00015348893404320112,
+      "loss": 0.8549,
+      "step": 5777
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.0001534741881941464,
+      "loss": 0.8372,
+      "step": 5778
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.00015345944071653667,
+      "loss": 0.9524,
+      "step": 5779
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.00015344469161082108,
+      "loss": 0.8987,
+      "step": 5780
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.0001534299408774488,
+      "loss": 0.9986,
+      "step": 5781
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.00015341518851686906,
+      "loss": 0.8296,
+      "step": 5782
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.0001534004345295312,
+      "loss": 0.8508,
+      "step": 5783
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.00015338567891588447,
+      "loss": 0.7575,
+      "step": 5784
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.0001533709216763783,
+      "loss": 1.0146,
+      "step": 5785
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.0001533561628114621,
+      "loss": 0.9839,
+      "step": 5786
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.00015334140232158534,
+      "loss": 0.8647,
+      "step": 5787
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.00015332664020719763,
+      "loss": 0.8659,
+      "step": 5788
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.00015331187646874846,
+      "loss": 0.882,
+      "step": 5789
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.0001532971111066875,
+      "loss": 0.9436,
+      "step": 5790
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.0001532823441214644,
+      "loss": 0.9777,
+      "step": 5791
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.00015326757551352893,
+      "loss": 1.0238,
+      "step": 5792
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.00015325280528333084,
+      "loss": 0.9096,
+      "step": 5793
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.00015323803343131998,
+      "loss": 0.8924,
+      "step": 5794
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.0001532232599579462,
+      "loss": 0.835,
+      "step": 5795
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.00015320848486365942,
+      "loss": 0.981,
+      "step": 5796
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.00015319370814890966,
+      "loss": 0.9552,
+      "step": 5797
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.00015317892981414686,
+      "loss": 0.9124,
+      "step": 5798
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.00015316414985982118,
+      "loss": 0.9106,
+      "step": 5799
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.00015314936828638272,
+      "loss": 0.8507,
+      "step": 5800
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.00015313458509428162,
+      "loss": 1.0037,
+      "step": 5801
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.00015311980028396817,
+      "loss": 0.9585,
+      "step": 5802
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.00015310501385589257,
+      "loss": 0.8501,
+      "step": 5803
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.00015309022581050513,
+      "loss": 0.839,
+      "step": 5804
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.00015307543614825634,
+      "loss": 1.1894,
+      "step": 5805
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.00015306064486959647,
+      "loss": 0.876,
+      "step": 5806
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.00015304585197497604,
+      "loss": 0.9507,
+      "step": 5807
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.00015303105746484566,
+      "loss": 0.8894,
+      "step": 5808
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.00015301626133965574,
+      "loss": 0.9361,
+      "step": 5809
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.00015300146359985698,
+      "loss": 1.0151,
+      "step": 5810
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.0001529866642459001,
+      "loss": 0.8953,
+      "step": 5811
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.00015297186327823566,
+      "loss": 0.8945,
+      "step": 5812
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.00015295706069731459,
+      "loss": 0.9308,
+      "step": 5813
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.0001529422565035876,
+      "loss": 0.9372,
+      "step": 5814
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.00015292745069750557,
+      "loss": 0.9022,
+      "step": 5815
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.00015291264327951942,
+      "loss": 0.9167,
+      "step": 5816
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.0001528978342500801,
+      "loss": 0.9716,
+      "step": 5817
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.00015288302360963866,
+      "loss": 0.9713,
+      "step": 5818
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.0001528682113586461,
+      "loss": 0.8801,
+      "step": 5819
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.00015285339749755353,
+      "loss": 1.0393,
+      "step": 5820
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.00015283858202681217,
+      "loss": 0.9122,
+      "step": 5821
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.00015282376494687316,
+      "loss": 0.9253,
+      "step": 5822
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.0001528089462581878,
+      "loss": 0.9207,
+      "step": 5823
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.00015279412596120733,
+      "loss": 1.0474,
+      "step": 5824
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.00015277930405638315,
+      "loss": 0.7897,
+      "step": 5825
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.00015276448054416666,
+      "loss": 0.9547,
+      "step": 5826
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.00015274965542500928,
+      "loss": 0.9689,
+      "step": 5827
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.00015273482869936254,
+      "loss": 1.001,
+      "step": 5828
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.000152720000367678,
+      "loss": 0.9153,
+      "step": 5829
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.00015270517043040717,
+      "loss": 0.887,
+      "step": 5830
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.00015269033888800178,
+      "loss": 0.95,
+      "step": 5831
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.00015267550574091357,
+      "loss": 0.7966,
+      "step": 5832
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.00015266067098959415,
+      "loss": 0.9292,
+      "step": 5833
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.0001526458346344954,
+      "loss": 0.877,
+      "step": 5834
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.00015263099667606913,
+      "loss": 0.8935,
+      "step": 5835
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.00015261615711476725,
+      "loss": 0.8639,
+      "step": 5836
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.00015260131595104168,
+      "loss": 1.0268,
+      "step": 5837
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.0001525864731853444,
+      "loss": 0.896,
+      "step": 5838
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.00015257162881812742,
+      "loss": 0.9201,
+      "step": 5839
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.00015255678284984296,
+      "loss": 0.9629,
+      "step": 5840
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.000152541935280943,
+      "loss": 0.9732,
+      "step": 5841
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.00015252708611187976,
+      "loss": 1.027,
+      "step": 5842
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.00015251223534310554,
+      "loss": 1.0308,
+      "step": 5843
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.00015249738297507255,
+      "loss": 0.9605,
+      "step": 5844
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.00015248252900823314,
+      "loss": 1.0967,
+      "step": 5845
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.00015246767344303969,
+      "loss": 0.86,
+      "step": 5846
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.00015245281627994462,
+      "loss": 0.8747,
+      "step": 5847
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.00015243795751940042,
+      "loss": 0.9409,
+      "step": 5848
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.00015242309716185956,
+      "loss": 0.8666,
+      "step": 5849
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.00015240823520777468,
+      "loss": 0.9355,
+      "step": 5850
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.0001523933716575984,
+      "loss": 0.8001,
+      "step": 5851
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.0001523785065117833,
+      "loss": 1.0216,
+      "step": 5852
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.0001523636397707822,
+      "loss": 0.8509,
+      "step": 5853
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.00015234877143504783,
+      "loss": 0.989,
+      "step": 5854
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.00015233390150503294,
+      "loss": 1.0194,
+      "step": 5855
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.0001523190299811905,
+      "loss": 0.8167,
+      "step": 5856
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.00015230415686397341,
+      "loss": 0.8698,
+      "step": 5857
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.00015228928215383456,
+      "loss": 1.0211,
+      "step": 5858
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.000152274405851227,
+      "loss": 0.9869,
+      "step": 5859
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.00015225952795660375,
+      "loss": 0.8696,
+      "step": 5860
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.00015224464847041796,
+      "loss": 0.9926,
+      "step": 5861
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.0001522297673931228,
+      "loss": 0.9873,
+      "step": 5862
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.00015221488472517145,
+      "loss": 0.9273,
+      "step": 5863
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.00015220000046701712,
+      "loss": 0.9718,
+      "step": 5864
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.00015218511461911316,
+      "loss": 0.9255,
+      "step": 5865
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.0001521702271819129,
+      "loss": 0.8601,
+      "step": 5866
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.00015215533815586974,
+      "loss": 0.7611,
+      "step": 5867
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.00015214044754143714,
+      "loss": 0.9116,
+      "step": 5868
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.0001521255553390686,
+      "loss": 0.923,
+      "step": 5869
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.0001521106615492176,
+      "loss": 0.899,
+      "step": 5870
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.00015209576617233777,
+      "loss": 0.9934,
+      "step": 5871
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.0001520808692088828,
+      "loss": 1.0526,
+      "step": 5872
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.00015206597065930624,
+      "loss": 0.8672,
+      "step": 5873
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.00015205107052406196,
+      "loss": 0.94,
+      "step": 5874
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.0001520361688036037,
+      "loss": 1.1988,
+      "step": 5875
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.00015202126549838528,
+      "loss": 0.91,
+      "step": 5876
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.00015200636060886056,
+      "loss": 0.9715,
+      "step": 5877
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.00015199145413548353,
+      "loss": 0.9581,
+      "step": 5878
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.0001519765460787081,
+      "loss": 0.8665,
+      "step": 5879
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.00015196163643898833,
+      "loss": 0.8685,
+      "step": 5880
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.00015194672521677828,
+      "loss": 1.0632,
+      "step": 5881
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.00015193181241253208,
+      "loss": 1.0275,
+      "step": 5882
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.00015191689802670393,
+      "loss": 0.9913,
+      "step": 5883
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.00015190198205974797,
+      "loss": 0.989,
+      "step": 5884
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.0001518870645121185,
+      "loss": 1.0222,
+      "step": 5885
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.00015187214538426992,
+      "loss": 0.9919,
+      "step": 5886
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.00015185722467665644,
+      "loss": 0.9077,
+      "step": 5887
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.00015184230238973254,
+      "loss": 0.8053,
+      "step": 5888
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.00015182737852395274,
+      "loss": 0.844,
+      "step": 5889
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.0001518124530797714,
+      "loss": 0.9431,
+      "step": 5890
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.00015179752605764323,
+      "loss": 0.9963,
+      "step": 5891
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.00015178259745802277,
+      "loss": 0.8494,
+      "step": 5892
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.00015176766728136466,
+      "loss": 0.9017,
+      "step": 5893
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.00015175273552812355,
+      "loss": 0.9226,
+      "step": 5894
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.00015173780219875427,
+      "loss": 0.8885,
+      "step": 5895
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.00015172286729371156,
+      "loss": 0.8565,
+      "step": 5896
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.00015170793081345035,
+      "loss": 1.0106,
+      "step": 5897
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.00015169299275842542,
+      "loss": 0.975,
+      "step": 5898
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.00015167805312909174,
+      "loss": 0.8249,
+      "step": 5899
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.0001516631119259043,
+      "loss": 0.9161,
+      "step": 5900
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.00015164816914931816,
+      "loss": 0.7783,
+      "step": 5901
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.0001516332247997884,
+      "loss": 1.0379,
+      "step": 5902
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.00015161827887777014,
+      "loss": 0.8684,
+      "step": 5903
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.00015160333138371853,
+      "loss": 0.8489,
+      "step": 5904
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.0001515883823180888,
+      "loss": 1.0474,
+      "step": 5905
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.00015157343168133627,
+      "loss": 0.9467,
+      "step": 5906
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.0001515584794739162,
+      "loss": 0.9981,
+      "step": 5907
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.00015154352569628403,
+      "loss": 1.0018,
+      "step": 5908
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.00015152857034889512,
+      "loss": 1.0055,
+      "step": 5909
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.00015151361343220497,
+      "loss": 0.8348,
+      "step": 5910
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.00015149865494666905,
+      "loss": 0.8786,
+      "step": 5911
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.00015148369489274296,
+      "loss": 0.9507,
+      "step": 5912
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.00015146873327088228,
+      "loss": 0.9381,
+      "step": 5913
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.00015145377008154268,
+      "loss": 0.8822,
+      "step": 5914
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.00015143880532517986,
+      "loss": 0.9331,
+      "step": 5915
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.0001514238390022496,
+      "loss": 0.825,
+      "step": 5916
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.00015140887111320765,
+      "loss": 0.9501,
+      "step": 5917
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.0001513939016585099,
+      "loss": 1.0096,
+      "step": 5918
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.00015137893063861218,
+      "loss": 0.9376,
+      "step": 5919
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.00015136395805397053,
+      "loss": 1.1435,
+      "step": 5920
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.00015134898390504087,
+      "loss": 0.9452,
+      "step": 5921
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.0001513340081922792,
+      "loss": 0.8762,
+      "step": 5922
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.00015131903091614174,
+      "loss": 0.9579,
+      "step": 5923
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.00015130405207708448,
+      "loss": 0.811,
+      "step": 5924
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.00015128907167556369,
+      "loss": 0.8762,
+      "step": 5925
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.00015127408971203553,
+      "loss": 0.7164,
+      "step": 5926
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.0001512591061869564,
+      "loss": 0.9935,
+      "step": 5927
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.00015124412110078246,
+      "loss": 0.8492,
+      "step": 5928
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.00015122913445397016,
+      "loss": 0.9496,
+      "step": 5929
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.00015121414624697592,
+      "loss": 0.9408,
+      "step": 5930
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.00015119915648025622,
+      "loss": 0.9237,
+      "step": 5931
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.00015118416515426758,
+      "loss": 0.9143,
+      "step": 5932
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.00015116917226946648,
+      "loss": 0.8642,
+      "step": 5933
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.00015115417782630958,
+      "loss": 0.8662,
+      "step": 5934
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.00015113918182525356,
+      "loss": 0.9093,
+      "step": 5935
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.0001511241842667551,
+      "loss": 0.9482,
+      "step": 5936
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.00015110918515127096,
+      "loss": 0.8917,
+      "step": 5937
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.0001510941844792579,
+      "loss": 0.9611,
+      "step": 5938
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.00015107918225117282,
+      "loss": 1.0089,
+      "step": 5939
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.0001510641784674726,
+      "loss": 0.9122,
+      "step": 5940
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.00015104917312861412,
+      "loss": 0.9687,
+      "step": 5941
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.00015103416623505443,
+      "loss": 0.8403,
+      "step": 5942
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.00015101915778725056,
+      "loss": 0.8433,
+      "step": 5943
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.00015100414778565954,
+      "loss": 0.9477,
+      "step": 5944
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.00015098913623073855,
+      "loss": 0.9685,
+      "step": 5945
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.0001509741231229448,
+      "loss": 0.8993,
+      "step": 5946
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.0001509591084627354,
+      "loss": 0.8023,
+      "step": 5947
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.00015094409225056775,
+      "loss": 0.9188,
+      "step": 5948
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.0001509290744868991,
+      "loss": 0.9655,
+      "step": 5949
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.00015091405517218676,
+      "loss": 0.9366,
+      "step": 5950
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.0001508990343068883,
+      "loss": 0.9235,
+      "step": 5951
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.000150884011891461,
+      "loss": 0.9974,
+      "step": 5952
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.0001508689879263625,
+      "loss": 0.8605,
+      "step": 5953
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.0001508539624120503,
+      "loss": 0.9064,
+      "step": 5954
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.000150838935348982,
+      "loss": 0.9165,
+      "step": 5955
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.0001508239067376153,
+      "loss": 0.9073,
+      "step": 5956
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.0001508088765784078,
+      "loss": 0.7927,
+      "step": 5957
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.00015079384487181731,
+      "loss": 0.9805,
+      "step": 5958
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.00015077881161830163,
+      "loss": 0.9404,
+      "step": 5959
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.00015076377681831857,
+      "loss": 0.7592,
+      "step": 5960
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.000150748740472326,
+      "loss": 0.9244,
+      "step": 5961
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.00015073370258078186,
+      "loss": 0.9445,
+      "step": 5962
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.0001507186631441442,
+      "loss": 0.801,
+      "step": 5963
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.00015070362216287092,
+      "loss": 0.8221,
+      "step": 5964
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.00015068857963742018,
+      "loss": 1.0671,
+      "step": 5965
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.0001506735355682501,
+      "loss": 0.9312,
+      "step": 5966
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.0001506584899558188,
+      "loss": 0.9447,
+      "step": 5967
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.0001506434428005845,
+      "loss": 1.0105,
+      "step": 5968
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.0001506283941030055,
+      "loss": 0.854,
+      "step": 5969
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.00015061334386354012,
+      "loss": 0.8989,
+      "step": 5970
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.00015059829208264664,
+      "loss": 0.8857,
+      "step": 5971
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.0001505832387607835,
+      "loss": 0.947,
+      "step": 5972
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.00015056818389840918,
+      "loss": 0.8151,
+      "step": 5973
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.0001505531274959821,
+      "loss": 0.9414,
+      "step": 5974
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.00015053806955396093,
+      "loss": 0.9779,
+      "step": 5975
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.00015052301007280412,
+      "loss": 0.8945,
+      "step": 5976
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.00015050794905297035,
+      "loss": 0.9326,
+      "step": 5977
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.00015049288649491836,
+      "loss": 1.0636,
+      "step": 5978
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.0001504778223991068,
+      "loss": 0.8288,
+      "step": 5979
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.0001504627567659945,
+      "loss": 0.8699,
+      "step": 5980
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.00015044768959604027,
+      "loss": 0.9561,
+      "step": 5981
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.00015043262088970297,
+      "loss": 0.7515,
+      "step": 5982
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.00015041755064744154,
+      "loss": 1.0842,
+      "step": 5983
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.00015040247886971493,
+      "loss": 0.9247,
+      "step": 5984
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.00015038740555698213,
+      "loss": 0.9109,
+      "step": 5985
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.00015037233070970223,
+      "loss": 0.8012,
+      "step": 5986
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.0001503572543283343,
+      "loss": 0.9479,
+      "step": 5987
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.00015034217641333753,
+      "loss": 0.77,
+      "step": 5988
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.00015032709696517112,
+      "loss": 0.9326,
+      "step": 5989
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.00015031201598429423,
+      "loss": 0.8489,
+      "step": 5990
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.00015029693347116624,
+      "loss": 0.9919,
+      "step": 5991
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.00015028184942624648,
+      "loss": 1.0328,
+      "step": 5992
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.0001502667638499943,
+      "loss": 1.0827,
+      "step": 5993
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.00015025167674286916,
+      "loss": 0.927,
+      "step": 5994
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.0001502365881053305,
+      "loss": 0.9417,
+      "step": 5995
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.00015022149793783788,
+      "loss": 0.8816,
+      "step": 5996
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.0001502064062408509,
+      "loss": 0.8922,
+      "step": 5997
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.00015019131301482908,
+      "loss": 0.9441,
+      "step": 5998
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.00015017621826023215,
+      "loss": 0.9412,
+      "step": 5999
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.0001501611219775198,
+      "loss": 1.001,
+      "step": 6000
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.00015014602416715185,
+      "loss": 1.0734,
+      "step": 6001
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.000150130924829588,
+      "loss": 0.9285,
+      "step": 6002
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.00015011582396528818,
+      "loss": 0.8593,
+      "step": 6003
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.00015010072157471222,
+      "loss": 0.9142,
+      "step": 6004
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.00015008561765832013,
+      "loss": 0.8894,
+      "step": 6005
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.00015007051221657187,
+      "loss": 0.9458,
+      "step": 6006
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.00015005540524992746,
+      "loss": 0.8314,
+      "step": 6007
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.000150040296758847,
+      "loss": 0.9231,
+      "step": 6008
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.0001500251867437906,
+      "loss": 0.9575,
+      "step": 6009
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.00015001007520521847,
+      "loss": 0.8437,
+      "step": 6010
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.00014999496214359082,
+      "loss": 0.8053,
+      "step": 6011
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.00014997984755936789,
+      "loss": 0.926,
+      "step": 6012
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.00014996473145301002,
+      "loss": 1.092,
+      "step": 6013
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.00014994961382497758,
+      "loss": 0.8749,
+      "step": 6014
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.00014993449467573094,
+      "loss": 1.0126,
+      "step": 6015
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.0001499193740057306,
+      "loss": 0.821,
+      "step": 6016
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.00014990425181543699,
+      "loss": 0.9046,
+      "step": 6017
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.00014988912810531076,
+      "loss": 0.9892,
+      "step": 6018
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.00014987400287581242,
+      "loss": 0.8976,
+      "step": 6019
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.00014985887612740262,
+      "loss": 0.9673,
+      "step": 6020
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.00014984374786054206,
+      "loss": 0.8387,
+      "step": 6021
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.00014982861807569147,
+      "loss": 1.0174,
+      "step": 6022
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.0001498134867733116,
+      "loss": 0.8993,
+      "step": 6023
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.00014979835395386335,
+      "loss": 0.9946,
+      "step": 6024
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.0001497832196178075,
+      "loss": 0.9624,
+      "step": 6025
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.000149768083765605,
+      "loss": 1.0629,
+      "step": 6026
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.00014975294639771685,
+      "loss": 0.9894,
+      "step": 6027
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.00014973780751460398,
+      "loss": 1.0198,
+      "step": 6028
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.0001497226671167275,
+      "loss": 1.0221,
+      "step": 6029
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.0001497075252045485,
+      "loss": 0.87,
+      "step": 6030
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.00014969238177852813,
+      "loss": 0.8599,
+      "step": 6031
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.0001496772368391276,
+      "loss": 0.9268,
+      "step": 6032
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.0001496620903868081,
+      "loss": 0.8216,
+      "step": 6033
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.00014964694242203093,
+      "loss": 0.9056,
+      "step": 6034
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.00014963179294525747,
+      "loss": 0.8617,
+      "step": 6035
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.000149616641956949,
+      "loss": 0.9465,
+      "step": 6036
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.00014960148945756705,
+      "loss": 0.8618,
+      "step": 6037
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.00014958633544757304,
+      "loss": 0.8719,
+      "step": 6038
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.00014957117992742846,
+      "loss": 0.8822,
+      "step": 6039
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.00014955602289759491,
+      "loss": 0.9378,
+      "step": 6040
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.00014954086435853397,
+      "loss": 0.9839,
+      "step": 6041
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.00014952570431070732,
+      "loss": 0.8608,
+      "step": 6042
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.00014951054275457666,
+      "loss": 0.9657,
+      "step": 6043
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.0001494953796906037,
+      "loss": 1.0927,
+      "step": 6044
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.00014948021511925022,
+      "loss": 0.9571,
+      "step": 6045
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.00014946504904097814,
+      "loss": 0.9815,
+      "step": 6046
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.00014944988145624928,
+      "loss": 0.9426,
+      "step": 6047
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.00014943471236552556,
+      "loss": 1.0687,
+      "step": 6048
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.000149419541769269,
+      "loss": 0.9821,
+      "step": 6049
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.00014940436966794156,
+      "loss": 0.8027,
+      "step": 6050
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.00014938919606200535,
+      "loss": 0.9957,
+      "step": 6051
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.00014937402095192249,
+      "loss": 0.9947,
+      "step": 6052
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.00014935884433815507,
+      "loss": 0.8694,
+      "step": 6053
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.0001493436662211654,
+      "loss": 0.9303,
+      "step": 6054
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.00014932848660141563,
+      "loss": 0.9663,
+      "step": 6055
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.00014931330547936808,
+      "loss": 0.8781,
+      "step": 6056
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.00014929812285548514,
+      "loss": 0.9897,
+      "step": 6057
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.00014928293873022917,
+      "loss": 0.9394,
+      "step": 6058
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.00014926775310406258,
+      "loss": 0.8912,
+      "step": 6059
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.00014925256597744785,
+      "loss": 0.8949,
+      "step": 6060
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.00014923737735084752,
+      "loss": 1.0378,
+      "step": 6061
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.00014922218722472418,
+      "loss": 0.8543,
+      "step": 6062
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.0001492069955995404,
+      "loss": 0.9254,
+      "step": 6063
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.00014919180247575885,
+      "loss": 0.8144,
+      "step": 6064
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.0001491766078538423,
+      "loss": 0.9414,
+      "step": 6065
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.00014916141173425338,
+      "loss": 0.8716,
+      "step": 6066
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.00014914621411745498,
+      "loss": 1.0014,
+      "step": 6067
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.00014913101500390994,
+      "loss": 0.763,
+      "step": 6068
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.0001491158143940811,
+      "loss": 0.9909,
+      "step": 6069
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.00014910061228843143,
+      "loss": 0.9459,
+      "step": 6070
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.0001490854086874239,
+      "loss": 1.0234,
+      "step": 6071
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.00014907020359152155,
+      "loss": 1.063,
+      "step": 6072
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.00014905499700118742,
+      "loss": 0.902,
+      "step": 6073
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.00014903978891688463,
+      "loss": 0.9453,
+      "step": 6074
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.0001490245793390764,
+      "loss": 0.9645,
+      "step": 6075
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.0001490093682682259,
+      "loss": 0.9507,
+      "step": 6076
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.0001489941557047963,
+      "loss": 0.942,
+      "step": 6077
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.00014897894164925103,
+      "loss": 0.81,
+      "step": 6078
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.00014896372610205338,
+      "loss": 0.9674,
+      "step": 6079
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.00014894850906366672,
+      "loss": 0.8648,
+      "step": 6080
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.00014893329053455455,
+      "loss": 0.9379,
+      "step": 6081
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.0001489180705151802,
+      "loss": 0.945,
+      "step": 6082
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.00014890284900600736,
+      "loss": 1.0459,
+      "step": 6083
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.00014888762600749955,
+      "loss": 1.0523,
+      "step": 6084
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.00014887240152012035,
+      "loss": 0.9358,
+      "step": 6085
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.00014885717554433345,
+      "loss": 0.8641,
+      "step": 6086
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.00014884194808060256,
+      "loss": 1.0336,
+      "step": 6087
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.00014882671912939139,
+      "loss": 0.9005,
+      "step": 6088
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.00014881148869116378,
+      "loss": 0.8705,
+      "step": 6089
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.00014879625676638358,
+      "loss": 0.8436,
+      "step": 6090
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.00014878102335551464,
+      "loss": 0.9749,
+      "step": 6091
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.00014876578845902095,
+      "loss": 1.0166,
+      "step": 6092
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.00014875055207736644,
+      "loss": 0.8692,
+      "step": 6093
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.00014873531421101514,
+      "loss": 0.8731,
+      "step": 6094
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.0001487200748604311,
+      "loss": 0.8397,
+      "step": 6095
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.0001487048340260785,
+      "loss": 0.8424,
+      "step": 6096
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.00014868959170842146,
+      "loss": 1.0035,
+      "step": 6097
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.00014867434790792416,
+      "loss": 0.8146,
+      "step": 6098
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.00014865910262505086,
+      "loss": 0.9029,
+      "step": 6099
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.0001486438558602659,
+      "loss": 0.9416,
+      "step": 6100
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.00014862860761403356,
+      "loss": 0.8697,
+      "step": 6101
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.00014861335788681824,
+      "loss": 0.9783,
+      "step": 6102
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.00014859810667908443,
+      "loss": 0.844,
+      "step": 6103
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.0001485828539912965,
+      "loss": 0.9432,
+      "step": 6104
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.00014856759982391905,
+      "loss": 0.9832,
+      "step": 6105
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.00014855234417741667,
+      "loss": 0.9585,
+      "step": 6106
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.00014853708705225385,
+      "loss": 0.9123,
+      "step": 6107
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.00014852182844889535,
+      "loss": 0.9106,
+      "step": 6108
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.00014850656836780583,
+      "loss": 0.9536,
+      "step": 6109
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.00014849130680945005,
+      "loss": 0.9155,
+      "step": 6110
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.00014847604377429282,
+      "loss": 0.7985,
+      "step": 6111
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.0001484607792627989,
+      "loss": 0.9357,
+      "step": 6112
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.00014844551327543323,
+      "loss": 0.7658,
+      "step": 6113
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.00014843024581266074,
+      "loss": 0.9618,
+      "step": 6114
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.00014841497687494636,
+      "loss": 0.8913,
+      "step": 6115
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.00014839970646275517,
+      "loss": 0.8776,
+      "step": 6116
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.00014838443457655217,
+      "loss": 0.8407,
+      "step": 6117
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.0001483691612168025,
+      "loss": 1.0418,
+      "step": 6118
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.00014835388638397126,
+      "loss": 0.9571,
+      "step": 6119
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.0001483386100785237,
+      "loss": 0.8994,
+      "step": 6120
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.00014832333230092507,
+      "loss": 0.8663,
+      "step": 6121
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.0001483080530516406,
+      "loss": 0.9107,
+      "step": 6122
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.0001482927723311356,
+      "loss": 0.8531,
+      "step": 6123
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.00014827749013987554,
+      "loss": 0.8436,
+      "step": 6124
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.00014826220647832575,
+      "loss": 0.9718,
+      "step": 6125
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.00014824692134695174,
+      "loss": 0.9527,
+      "step": 6126
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.00014823163474621904,
+      "loss": 0.972,
+      "step": 6127
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.00014821634667659312,
+      "loss": 0.8306,
+      "step": 6128
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.00014820105713853965,
+      "loss": 0.8719,
+      "step": 6129
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.00014818576613252423,
+      "loss": 0.8834,
+      "step": 6130
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.0001481704736590126,
+      "loss": 0.8778,
+      "step": 6131
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.00014815517971847043,
+      "loss": 1.0034,
+      "step": 6132
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.00014813988431136357,
+      "loss": 0.8443,
+      "step": 6133
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.00014812458743815774,
+      "loss": 0.9156,
+      "step": 6134
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.0001481092890993189,
+      "loss": 0.9143,
+      "step": 6135
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.0001480939892953129,
+      "loss": 0.9637,
+      "step": 6136
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.00014807868802660571,
+      "loss": 0.9385,
+      "step": 6137
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.00014806338529366338,
+      "loss": 1.0002,
+      "step": 6138
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.0001480480810969519,
+      "loss": 1.013,
+      "step": 6139
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.00014803277543693735,
+      "loss": 0.9747,
+      "step": 6140
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.0001480174683140859,
+      "loss": 1.0095,
+      "step": 6141
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.0001480021597288637,
+      "loss": 0.9079,
+      "step": 6142
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.00014798684968173701,
+      "loss": 0.956,
+      "step": 6143
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.00014797153817317207,
+      "loss": 0.9869,
+      "step": 6144
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.00014795622520363518,
+      "loss": 0.7589,
+      "step": 6145
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.0001479409107735927,
+      "loss": 0.8946,
+      "step": 6146
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.00014792559488351103,
+      "loss": 0.9096,
+      "step": 6147
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.00014791027753385664,
+      "loss": 0.9758,
+      "step": 6148
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.00014789495872509603,
+      "loss": 0.9109,
+      "step": 6149
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.0001478796384576957,
+      "loss": 0.9778,
+      "step": 6150
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.0001478643167321222,
+      "loss": 0.9112,
+      "step": 6151
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.00014784899354884225,
+      "loss": 0.9722,
+      "step": 6152
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.00014783366890832241,
+      "loss": 0.8894,
+      "step": 6153
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.00014781834281102945,
+      "loss": 0.8272,
+      "step": 6154
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.00014780301525743012,
+      "loss": 0.7718,
+      "step": 6155
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.00014778768624799124,
+      "loss": 0.8323,
+      "step": 6156
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.00014777235578317962,
+      "loss": 0.9116,
+      "step": 6157
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.00014775702386346214,
+      "loss": 0.8779,
+      "step": 6158
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.00014774169048930574,
+      "loss": 0.912,
+      "step": 6159
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.00014772635566117742,
+      "loss": 0.8777,
+      "step": 6160
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.0001477110193795442,
+      "loss": 1.0507,
+      "step": 6161
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.00014769568164487313,
+      "loss": 0.9619,
+      "step": 6162
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.00014768034245763134,
+      "loss": 0.8664,
+      "step": 6163
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.00014766500181828595,
+      "loss": 1.0131,
+      "step": 6164
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.0001476496597273042,
+      "loss": 0.9052,
+      "step": 6165
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.00014763431618515329,
+      "loss": 0.908,
+      "step": 6166
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.00014761897119230057,
+      "loss": 0.882,
+      "step": 6167
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.0001476036247492133,
+      "loss": 0.8881,
+      "step": 6168
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.0001475882768563589,
+      "loss": 1.0075,
+      "step": 6169
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.00014757292751420474,
+      "loss": 1.0616,
+      "step": 6170
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.00014755757672321837,
+      "loss": 1.0058,
+      "step": 6171
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.00014754222448386722,
+      "loss": 0.9863,
+      "step": 6172
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.0001475268707966189,
+      "loss": 1.0424,
+      "step": 6173
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.00014751151566194098,
+      "loss": 0.8784,
+      "step": 6174
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.00014749615908030107,
+      "loss": 0.9075,
+      "step": 6175
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.00014748080105216692,
+      "loss": 0.8238,
+      "step": 6176
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.00014746544157800617,
+      "loss": 0.9757,
+      "step": 6177
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.00014745008065828667,
+      "loss": 1.0306,
+      "step": 6178
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.0001474347182934762,
+      "loss": 0.9728,
+      "step": 6179
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.00014741935448404263,
+      "loss": 0.8003,
+      "step": 6180
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.00014740398923045387,
+      "loss": 0.949,
+      "step": 6181
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.00014738862253317785,
+      "loss": 0.8755,
+      "step": 6182
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.00014737325439268258,
+      "loss": 0.9393,
+      "step": 6183
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.00014735788480943612,
+      "loss": 0.9456,
+      "step": 6184
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.0001473425137839065,
+      "loss": 0.89,
+      "step": 6185
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.00014732714131656183,
+      "loss": 1.0556,
+      "step": 6186
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.00014731176740787036,
+      "loss": 0.9061,
+      "step": 6187
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.00014729639205830023,
+      "loss": 0.8433,
+      "step": 6188
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.00014728101526831973,
+      "loss": 0.8622,
+      "step": 6189
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.00014726563703839718,
+      "loss": 0.8749,
+      "step": 6190
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.00014725025736900084,
+      "loss": 0.9899,
+      "step": 6191
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.0001472348762605992,
+      "loss": 0.8843,
+      "step": 6192
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.00014721949371366059,
+      "loss": 1.1226,
+      "step": 6193
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.00014720410972865358,
+      "loss": 1.0383,
+      "step": 6194
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.00014718872430604665,
+      "loss": 0.9744,
+      "step": 6195
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.00014717333744630835,
+      "loss": 0.8933,
+      "step": 6196
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.00014715794914990727,
+      "loss": 0.9101,
+      "step": 6197
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.00014714255941731212,
+      "loss": 0.8809,
+      "step": 6198
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.00014712716824899155,
+      "loss": 0.9555,
+      "step": 6199
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.0001471117756454143,
+      "loss": 1.0296,
+      "step": 6200
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.00014709638160704917,
+      "loss": 0.9959,
+      "step": 6201
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.00014708098613436498,
+      "loss": 0.9452,
+      "step": 6202
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.00014706558922783058,
+      "loss": 0.9266,
+      "step": 6203
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.0001470501908879149,
+      "loss": 0.9094,
+      "step": 6204
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.0001470347911150869,
+      "loss": 0.9233,
+      "step": 6205
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.00014701938990981556,
+      "loss": 0.8426,
+      "step": 6206
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.00014700398727256992,
+      "loss": 0.9834,
+      "step": 6207
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.0001469885832038191,
+      "loss": 0.8112,
+      "step": 6208
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.00014697317770403224,
+      "loss": 0.9431,
+      "step": 6209
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.00014695777077367844,
+      "loss": 0.959,
+      "step": 6210
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.00014694236241322698,
+      "loss": 0.8945,
+      "step": 6211
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.00014692695262314708,
+      "loss": 0.8907,
+      "step": 6212
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.0001469115414039081,
+      "loss": 0.9716,
+      "step": 6213
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.00014689612875597936,
+      "loss": 1.0648,
+      "step": 6214
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.00014688071467983024,
+      "loss": 0.9281,
+      "step": 6215
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.00014686529917593016,
+      "loss": 0.9232,
+      "step": 6216
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.00014684988224474867,
+      "loss": 0.8372,
+      "step": 6217
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.00014683446388675522,
+      "loss": 0.9148,
+      "step": 6218
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.0001468190441024194,
+      "loss": 0.9675,
+      "step": 6219
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.00014680362289221086,
+      "loss": 0.9169,
+      "step": 6220
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.00014678820025659916,
+      "loss": 0.9356,
+      "step": 6221
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.00014677277619605408,
+      "loss": 0.9688,
+      "step": 6222
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.0001467573507110453,
+      "loss": 0.8071,
+      "step": 6223
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.00014674192380204266,
+      "loss": 0.9399,
+      "step": 6224
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.00014672649546951596,
+      "loss": 0.9356,
+      "step": 6225
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.00014671106571393505,
+      "loss": 0.9389,
+      "step": 6226
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.00014669563453576984,
+      "loss": 0.9056,
+      "step": 6227
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.00014668020193549035,
+      "loss": 0.9764,
+      "step": 6228
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.00014666476791356652,
+      "loss": 0.9407,
+      "step": 6229
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.0001466493324704684,
+      "loss": 0.9848,
+      "step": 6230
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.0001466338956066661,
+      "loss": 0.8739,
+      "step": 6231
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.0001466184573226297,
+      "loss": 0.832,
+      "step": 6232
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.00014660301761882942,
+      "loss": 0.9274,
+      "step": 6233
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.00014658757649573548,
+      "loss": 1.0381,
+      "step": 6234
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.0001465721339538181,
+      "loss": 0.9387,
+      "step": 6235
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.00014655668999354765,
+      "loss": 0.8734,
+      "step": 6236
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.00014654124461539435,
+      "loss": 0.9741,
+      "step": 6237
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.00014652579781982873,
+      "loss": 1.0214,
+      "step": 6238
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.0001465103496073211,
+      "loss": 0.9085,
+      "step": 6239
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.00014649489997834204,
+      "loss": 0.9876,
+      "step": 6240
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.000146479448933362,
+      "loss": 0.8039,
+      "step": 6241
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.00014646399647285158,
+      "loss": 0.9368,
+      "step": 6242
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.00014644854259728135,
+      "loss": 0.8123,
+      "step": 6243
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.000146433087307122,
+      "loss": 0.9282,
+      "step": 6244
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.00014641763060284417,
+      "loss": 1.0541,
+      "step": 6245
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.00014640217248491864,
+      "loss": 1.0969,
+      "step": 6246
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.00014638671295381613,
+      "loss": 0.9006,
+      "step": 6247
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.00014637125201000752,
+      "loss": 1.0174,
+      "step": 6248
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.00014635578965396367,
+      "loss": 0.8658,
+      "step": 6249
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.0001463403258861554,
+      "loss": 0.9849,
+      "step": 6250
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.00014632486070705376,
+      "loss": 0.8382,
+      "step": 6251
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.0001463093941171297,
+      "loss": 0.8932,
+      "step": 6252
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.00014629392611685427,
+      "loss": 0.9452,
+      "step": 6253
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.0001462784567066985,
+      "loss": 0.9031,
+      "step": 6254
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.0001462629858871336,
+      "loss": 0.8035,
+      "step": 6255
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.00014624751365863064,
+      "loss": 0.8734,
+      "step": 6256
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.0001462320400216609,
+      "loss": 0.8394,
+      "step": 6257
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.00014621656497669554,
+      "loss": 0.8958,
+      "step": 6258
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.00014620108852420594,
+      "loss": 0.8089,
+      "step": 6259
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.0001461856106646634,
+      "loss": 0.9426,
+      "step": 6260
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.00014617013139853928,
+      "loss": 0.9136,
+      "step": 6261
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.00014615465072630504,
+      "loss": 0.8039,
+      "step": 6262
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.00014613916864843212,
+      "loss": 0.8843,
+      "step": 6263
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.00014612368516539202,
+      "loss": 0.9258,
+      "step": 6264
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.0001461082002776563,
+      "loss": 0.8492,
+      "step": 6265
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.00014609271398569658,
+      "loss": 0.9128,
+      "step": 6266
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.00014607722628998444,
+      "loss": 0.8403,
+      "step": 6267
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.00014606173719099158,
+      "loss": 0.8253,
+      "step": 6268
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.00014604624668918974,
+      "loss": 0.8962,
+      "step": 6269
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.0001460307547850506,
+      "loss": 0.6943,
+      "step": 6270
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.00014601526147904614,
+      "loss": 0.8734,
+      "step": 6271
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.00014599976677164803,
+      "loss": 0.7244,
+      "step": 6272
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.00014598427066332825,
+      "loss": 0.8256,
+      "step": 6273
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.0001459687731545587,
+      "loss": 0.9603,
+      "step": 6274
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.00014595327424581138,
+      "loss": 0.9721,
+      "step": 6275
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.0001459377739375583,
+      "loss": 0.7923,
+      "step": 6276
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.00014592227223027152,
+      "loss": 0.9521,
+      "step": 6277
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.00014590676912442315,
+      "loss": 0.8606,
+      "step": 6278
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.00014589126462048534,
+      "loss": 0.9594,
+      "step": 6279
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.00014587575871893025,
+      "loss": 0.8861,
+      "step": 6280
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.0001458602514202301,
+      "loss": 0.8743,
+      "step": 6281
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.00014584474272485727,
+      "loss": 0.8695,
+      "step": 6282
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.00014582923263328396,
+      "loss": 0.9393,
+      "step": 6283
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.0001458137211459826,
+      "loss": 0.833,
+      "step": 6284
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.00014579820826342554,
+      "loss": 0.863,
+      "step": 6285
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.00014578269398608524,
+      "loss": 0.9422,
+      "step": 6286
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.0001457671783144342,
+      "loss": 1.0334,
+      "step": 6287
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.00014575166124894498,
+      "loss": 0.9291,
+      "step": 6288
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.0001457361427900901,
+      "loss": 0.9475,
+      "step": 6289
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.0001457206229383422,
+      "loss": 0.8481,
+      "step": 6290
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.00014570510169417388,
+      "loss": 0.984,
+      "step": 6291
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.00014568957905805794,
+      "loss": 1.033,
+      "step": 6292
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.00014567405503046701,
+      "loss": 0.9597,
+      "step": 6293
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.000145658529611874,
+      "loss": 0.9262,
+      "step": 6294
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.00014564300280275164,
+      "loss": 0.9746,
+      "step": 6295
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.00014562747460357284,
+      "loss": 0.8776,
+      "step": 6296
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.0001456119450148105,
+      "loss": 0.8892,
+      "step": 6297
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.00014559641403693755,
+      "loss": 0.9198,
+      "step": 6298
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.000145580881670427,
+      "loss": 0.9569,
+      "step": 6299
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.00014556534791575193,
+      "loss": 0.9745,
+      "step": 6300
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.00014554981277338538,
+      "loss": 0.783,
+      "step": 6301
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.00014553427624380044,
+      "loss": 1.0423,
+      "step": 6302
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.00014551873832747037,
+      "loss": 0.7795,
+      "step": 6303
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.00014550319902486826,
+      "loss": 0.8986,
+      "step": 6304
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.00014548765833646745,
+      "loss": 0.8597,
+      "step": 6305
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.00014547211626274122,
+      "loss": 0.9653,
+      "step": 6306
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.00014545657280416285,
+      "loss": 0.9317,
+      "step": 6307
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.00014544102796120573,
+      "loss": 0.8975,
+      "step": 6308
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.00014542548173434333,
+      "loss": 0.8531,
+      "step": 6309
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.00014540993412404907,
+      "loss": 0.9882,
+      "step": 6310
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.00014539438513079645,
+      "loss": 0.9937,
+      "step": 6311
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.00014537883475505904,
+      "loss": 0.9539,
+      "step": 6312
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.0001453632829973104,
+      "loss": 0.8779,
+      "step": 6313
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.00014534772985802416,
+      "loss": 0.8706,
+      "step": 6314
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.00014533217533767398,
+      "loss": 0.7772,
+      "step": 6315
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.0001453166194367336,
+      "loss": 0.9573,
+      "step": 6316
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.00014530106215567682,
+      "loss": 0.9246,
+      "step": 6317
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.0001452855034949773,
+      "loss": 0.8422,
+      "step": 6318
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.000145269943455109,
+      "loss": 0.9743,
+      "step": 6319
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.00014525438203654578,
+      "loss": 0.9494,
+      "step": 6320
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.00014523881923976149,
+      "loss": 0.92,
+      "step": 6321
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.0001452232550652302,
+      "loss": 0.8497,
+      "step": 6322
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.00014520768951342583,
+      "loss": 1.0575,
+      "step": 6323
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.00014519212258482246,
+      "loss": 0.945,
+      "step": 6324
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.00014517655427989423,
+      "loss": 1.0385,
+      "step": 6325
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.00014516098459911516,
+      "loss": 0.8774,
+      "step": 6326
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.0001451454135429595,
+      "loss": 0.9629,
+      "step": 6327
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.00014512984111190154,
+      "loss": 0.7557,
+      "step": 6328
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.00014511426730641534,
+      "loss": 0.9893,
+      "step": 6329
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.0001450986921269754,
+      "loss": 0.858,
+      "step": 6330
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.00014508311557405592,
+      "loss": 0.9264,
+      "step": 6331
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.00014506753764813134,
+      "loss": 0.9321,
+      "step": 6332
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.0001450519583496761,
+      "loss": 0.8286,
+      "step": 6333
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.00014503637767916466,
+      "loss": 0.8411,
+      "step": 6334
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.0001450207956370715,
+      "loss": 0.8407,
+      "step": 6335
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.00014500521222387123,
+      "loss": 0.9191,
+      "step": 6336
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.00014498962744003834,
+      "loss": 0.8655,
+      "step": 6337
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.0001449740412860476,
+      "loss": 0.9254,
+      "step": 6338
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.00014495845376237353,
+      "loss": 0.9492,
+      "step": 6339
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.000144942864869491,
+      "loss": 0.879,
+      "step": 6340
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.00014492727460787468,
+      "loss": 0.8867,
+      "step": 6341
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.0001449116829779994,
+      "loss": 0.9034,
+      "step": 6342
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.00014489608998033995,
+      "loss": 0.9131,
+      "step": 6343
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.0001448804956153713,
+      "loss": 0.8458,
+      "step": 6344
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.0001448648998835683,
+      "loss": 0.8328,
+      "step": 6345
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.000144849302785406,
+      "loss": 0.951,
+      "step": 6346
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.00014483370432135932,
+      "loss": 0.9295,
+      "step": 6347
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.00014481810449190337,
+      "loss": 0.9488,
+      "step": 6348
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.0001448025032975132,
+      "loss": 0.8689,
+      "step": 6349
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.000144786900738664,
+      "loss": 1.0281,
+      "step": 6350
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.0001447712968158309,
+      "loss": 0.9835,
+      "step": 6351
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.00014475569152948913,
+      "loss": 1.0125,
+      "step": 6352
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.00014474008488011395,
+      "loss": 0.9756,
+      "step": 6353
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.00014472447686818065,
+      "loss": 0.9131,
+      "step": 6354
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.00014470886749416462,
+      "loss": 0.8908,
+      "step": 6355
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.00014469325675854115,
+      "loss": 0.8776,
+      "step": 6356
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.00014467764466178574,
+      "loss": 0.9154,
+      "step": 6357
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.00014466203120437385,
+      "loss": 1.0599,
+      "step": 6358
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.00014464641638678097,
+      "loss": 0.8925,
+      "step": 6359
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.00014463080020948264,
+      "loss": 0.8454,
+      "step": 6360
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.00014461518267295444,
+      "loss": 0.8445,
+      "step": 6361
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.00014459956377767204,
+      "loss": 1.0137,
+      "step": 6362
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.0001445839435241111,
+      "loss": 0.9246,
+      "step": 6363
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.00014456832191274733,
+      "loss": 0.9843,
+      "step": 6364
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.00014455269894405646,
+      "loss": 0.9768,
+      "step": 6365
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.00014453707461851439,
+      "loss": 0.8552,
+      "step": 6366
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.0001445214489365968,
+      "loss": 0.8512,
+      "step": 6367
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.00014450582189877968,
+      "loss": 0.9936,
+      "step": 6368
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.00014449019350553895,
+      "loss": 0.8863,
+      "step": 6369
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.0001444745637573505,
+      "loss": 0.8588,
+      "step": 6370
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.00014445893265469038,
+      "loss": 0.8562,
+      "step": 6371
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.00014444330019803468,
+      "loss": 0.8468,
+      "step": 6372
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.0001444276663878594,
+      "loss": 1.0618,
+      "step": 6373
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.00014441203122464072,
+      "loss": 0.9338,
+      "step": 6374
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.00014439639470885477,
+      "loss": 0.9038,
+      "step": 6375
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.0001443807568409778,
+      "loss": 0.9476,
+      "step": 6376
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.00014436511762148607,
+      "loss": 0.8513,
+      "step": 6377
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.00014434947705085582,
+      "loss": 0.8957,
+      "step": 6378
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.00014433383512956342,
+      "loss": 0.8584,
+      "step": 6379
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.00014431819185808524,
+      "loss": 0.8945,
+      "step": 6380
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.00014430254723689766,
+      "loss": 0.8193,
+      "step": 6381
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.0001442869012664772,
+      "loss": 0.9528,
+      "step": 6382
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.0001442712539473003,
+      "loss": 0.9554,
+      "step": 6383
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.00014425560527984353,
+      "loss": 0.8578,
+      "step": 6384
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.00014423995526458347,
+      "loss": 0.8302,
+      "step": 6385
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.00014422430390199675,
+      "loss": 0.8763,
+      "step": 6386
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.00014420865119256,
+      "loss": 0.8714,
+      "step": 6387
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.00014419299713674992,
+      "loss": 0.9148,
+      "step": 6388
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.00014417734173504326,
+      "loss": 0.92,
+      "step": 6389
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.00014416168498791686,
+      "loss": 0.7994,
+      "step": 6390
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.00014414602689584747,
+      "loss": 0.9848,
+      "step": 6391
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.00014413036745931202,
+      "loss": 0.8393,
+      "step": 6392
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.00014411470667878738,
+      "loss": 1.0037,
+      "step": 6393
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.0001440990445547505,
+      "loss": 0.9834,
+      "step": 6394
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.00014408338108767838,
+      "loss": 0.8234,
+      "step": 6395
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.00014406771627804802,
+      "loss": 0.9777,
+      "step": 6396
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.00014405205012633658,
+      "loss": 0.9222,
+      "step": 6397
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.00014403638263302105,
+      "loss": 0.9527,
+      "step": 6398
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.00014402071379857867,
+      "loss": 0.9689,
+      "step": 6399
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.00014400504362348658,
+      "loss": 0.7688,
+      "step": 6400
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.00014398937210822208,
+      "loss": 0.8772,
+      "step": 6401
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.00014397369925326237,
+      "loss": 1.0023,
+      "step": 6402
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.00014395802505908482,
+      "loss": 0.8715,
+      "step": 6403
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.00014394234952616678,
+      "loss": 0.9588,
+      "step": 6404
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.00014392667265498562,
+      "loss": 0.9366,
+      "step": 6405
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.0001439109944460188,
+      "loss": 0.9564,
+      "step": 6406
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.00014389531489974382,
+      "loss": 0.8828,
+      "step": 6407
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.00014387963401663812,
+      "loss": 0.7561,
+      "step": 6408
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.0001438639517971794,
+      "loss": 1.0128,
+      "step": 6409
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.00014384826824184513,
+      "loss": 0.848,
+      "step": 6410
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.00014383258335111298,
+      "loss": 0.6799,
+      "step": 6411
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.0001438168971254607,
+      "loss": 0.7969,
+      "step": 6412
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.00014380120956536594,
+      "loss": 0.9888,
+      "step": 6413
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.0001437855206713065,
+      "loss": 0.9633,
+      "step": 6414
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.0001437698304437602,
+      "loss": 0.8486,
+      "step": 6415
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.00014375413888320483,
+      "loss": 0.8683,
+      "step": 6416
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.00014373844599011834,
+      "loss": 0.892,
+      "step": 6417
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.00014372275176497857,
+      "loss": 0.86,
+      "step": 6418
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.0001437070562082636,
+      "loss": 0.946,
+      "step": 6419
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.00014369135932045137,
+      "loss": 1.0149,
+      "step": 6420
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.00014367566110201993,
+      "loss": 0.9989,
+      "step": 6421
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.00014365996155344735,
+      "loss": 0.8647,
+      "step": 6422
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.00014364426067521184,
+      "loss": 0.9481,
+      "step": 6423
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.00014362855846779146,
+      "loss": 0.9703,
+      "step": 6424
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.00014361285493166452,
+      "loss": 1.0008,
+      "step": 6425
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.00014359715006730922,
+      "loss": 0.9917,
+      "step": 6426
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.00014358144387520384,
+      "loss": 0.9247,
+      "step": 6427
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.00014356573635582675,
+      "loss": 0.8826,
+      "step": 6428
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.0001435500275096563,
+      "loss": 0.9879,
+      "step": 6429
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.00014353431733717088,
+      "loss": 0.9337,
+      "step": 6430
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.000143518605838849,
+      "loss": 0.9409,
+      "step": 6431
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.00014350289301516913,
+      "loss": 1.0036,
+      "step": 6432
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.00014348717886660976,
+      "loss": 0.9053,
+      "step": 6433
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.0001434714633936495,
+      "loss": 0.9795,
+      "step": 6434
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.00014345574659676698,
+      "loss": 0.9524,
+      "step": 6435
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.00014344002847644087,
+      "loss": 0.8228,
+      "step": 6436
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.00014342430903314978,
+      "loss": 0.8538,
+      "step": 6437
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.00014340858826737254,
+      "loss": 1.002,
+      "step": 6438
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.00014339286617958785,
+      "loss": 0.9018,
+      "step": 6439
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.00014337714277027456,
+      "loss": 0.8966,
+      "step": 6440
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.00014336141803991155,
+      "loss": 0.8517,
+      "step": 6441
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.00014334569198897767,
+      "loss": 0.9359,
+      "step": 6442
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.0001433299646179519,
+      "loss": 0.97,
+      "step": 6443
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.00014331423592731318,
+      "loss": 0.987,
+      "step": 6444
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.00014329850591754054,
+      "loss": 0.8586,
+      "step": 6445
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.000143282774589113,
+      "loss": 0.9788,
+      "step": 6446
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.00014326704194250978,
+      "loss": 0.914,
+      "step": 6447
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.00014325130797820986,
+      "loss": 0.8995,
+      "step": 6448
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.00014323557269669253,
+      "loss": 0.9262,
+      "step": 6449
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.00014321983609843697,
+      "loss": 0.8702,
+      "step": 6450
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.0001432040981839224,
+      "loss": 0.8393,
+      "step": 6451
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.0001431883589536282,
+      "loss": 0.8965,
+      "step": 6452
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.00014317261840803363,
+      "loss": 0.9335,
+      "step": 6453
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.00014315687654761812,
+      "loss": 0.9054,
+      "step": 6454
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.00014314113337286107,
+      "loss": 0.9329,
+      "step": 6455
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.00014312538888424192,
+      "loss": 1.0141,
+      "step": 6456
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.00014310964308224014,
+      "loss": 0.8949,
+      "step": 6457
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.0001430938959673354,
+      "loss": 0.8299,
+      "step": 6458
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.00014307814754000714,
+      "loss": 0.9852,
+      "step": 6459
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.000143062397800735,
+      "loss": 0.7955,
+      "step": 6460
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.00014304664674999874,
+      "loss": 0.9559,
+      "step": 6461
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.00014303089438827792,
+      "loss": 0.9687,
+      "step": 6462
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.0001430151407160524,
+      "loss": 0.9143,
+      "step": 6463
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.00014299938573380182,
+      "loss": 0.9612,
+      "step": 6464
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.00014298362944200612,
+      "loss": 0.907,
+      "step": 6465
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.00014296787184114512,
+      "loss": 0.9869,
+      "step": 6466
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.00014295211293169872,
+      "loss": 0.8155,
+      "step": 6467
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.00014293635271414682,
+      "loss": 0.7263,
+      "step": 6468
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.00014292059118896944,
+      "loss": 0.7972,
+      "step": 6469
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.00014290482835664655,
+      "loss": 0.8468,
+      "step": 6470
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.00014288906421765828,
+      "loss": 1.0157,
+      "step": 6471
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.0001428732987724847,
+      "loss": 0.9035,
+      "step": 6472
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.00014285753202160587,
+      "loss": 1.0377,
+      "step": 6473
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.00014284176396550208,
+      "loss": 0.9534,
+      "step": 6474
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.00014282599460465348,
+      "loss": 1.0292,
+      "step": 6475
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.0001428102239395403,
+      "loss": 0.9032,
+      "step": 6476
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.00014279445197064292,
+      "loss": 0.8565,
+      "step": 6477
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.0001427786786984416,
+      "loss": 1.0243,
+      "step": 6478
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.00014276290412341676,
+      "loss": 0.8688,
+      "step": 6479
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.00014274712824604882,
+      "loss": 0.8329,
+      "step": 6480
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.00014273135106681814,
+      "loss": 0.989,
+      "step": 6481
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.00014271557258620536,
+      "loss": 0.9984,
+      "step": 6482
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.00014269979280469089,
+      "loss": 0.8928,
+      "step": 6483
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.00014268401172275535,
+      "loss": 0.9937,
+      "step": 6484
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.0001426682293408794,
+      "loss": 1.015,
+      "step": 6485
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.0001426524456595436,
+      "loss": 0.8941,
+      "step": 6486
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.0001426366606792287,
+      "loss": 1.0131,
+      "step": 6487
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.00014262087440041542,
+      "loss": 0.9346,
+      "step": 6488
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.00014260508682358456,
+      "loss": 0.9258,
+      "step": 6489
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.00014258929794921688,
+      "loss": 0.9213,
+      "step": 6490
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.0001425735077777932,
+      "loss": 0.9034,
+      "step": 6491
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.00014255771630979452,
+      "loss": 0.7939,
+      "step": 6492
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.00014254192354570165,
+      "loss": 0.8797,
+      "step": 6493
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.00014252612948599565,
+      "loss": 0.9573,
+      "step": 6494
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.00014251033413115752,
+      "loss": 0.8396,
+      "step": 6495
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.00014249453748166823,
+      "loss": 0.9778,
+      "step": 6496
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.00014247873953800893,
+      "loss": 0.9961,
+      "step": 6497
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.00014246294030066076,
+      "loss": 0.9197,
+      "step": 6498
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.00014244713977010483,
+      "loss": 0.9488,
+      "step": 6499
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.00014243133794682235,
+      "loss": 0.8923,
+      "step": 6500
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.00014241553483129463,
+      "loss": 0.9969,
+      "step": 6501
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.00014239973042400291,
+      "loss": 0.8824,
+      "step": 6502
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.00014238392472542847,
+      "loss": 0.8657,
+      "step": 6503
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.00014236811773605276,
+      "loss": 0.9619,
+      "step": 6504
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.0001423523094563571,
+      "loss": 0.9368,
+      "step": 6505
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.000142336499886823,
+      "loss": 0.8581,
+      "step": 6506
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.0001423206890279319,
+      "loss": 0.8265,
+      "step": 6507
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.00014230487688016532,
+      "loss": 0.8823,
+      "step": 6508
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.00014228906344400483,
+      "loss": 0.9406,
+      "step": 6509
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.00014227324871993202,
+      "loss": 0.8942,
+      "step": 6510
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.0001422574327084285,
+      "loss": 0.9921,
+      "step": 6511
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.00014224161540997603,
+      "loss": 1.0007,
+      "step": 6512
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.00014222579682505624,
+      "loss": 0.8741,
+      "step": 6513
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.0001422099769541509,
+      "loss": 0.9544,
+      "step": 6514
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.00014219415579774187,
+      "loss": 0.9423,
+      "step": 6515
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.00014217833335631086,
+      "loss": 0.8973,
+      "step": 6516
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.00014216250963033987,
+      "loss": 0.9345,
+      "step": 6517
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.00014214668462031078,
+      "loss": 0.9441,
+      "step": 6518
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.00014213085832670544,
+      "loss": 0.9085,
+      "step": 6519
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.00014211503075000598,
+      "loss": 0.9307,
+      "step": 6520
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.00014209920189069432,
+      "loss": 0.8601,
+      "step": 6521
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.0001420833717492526,
+      "loss": 0.8875,
+      "step": 6522
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.0001420675403261629,
+      "loss": 0.9999,
+      "step": 6523
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.00014205170762190736,
+      "loss": 0.8165,
+      "step": 6524
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.00014203587363696815,
+      "loss": 0.8793,
+      "step": 6525
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.00014202003837182752,
+      "loss": 0.9595,
+      "step": 6526
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.00014200420182696773,
+      "loss": 0.9708,
+      "step": 6527
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.00014198836400287107,
+      "loss": 0.9038,
+      "step": 6528
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.00014197252490001985,
+      "loss": 1.0109,
+      "step": 6529
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.00014195668451889656,
+      "loss": 0.941,
+      "step": 6530
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.0001419408428599835,
+      "loss": 0.8162,
+      "step": 6531
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.00014192499992376318,
+      "loss": 0.9639,
+      "step": 6532
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.00014190915571071807,
+      "loss": 0.8843,
+      "step": 6533
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.00014189331022133078,
+      "loss": 0.8999,
+      "step": 6534
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.00014187746345608376,
+      "loss": 0.8719,
+      "step": 6535
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.00014186161541545974,
+      "loss": 0.8666,
+      "step": 6536
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.00014184576609994128,
+      "loss": 0.8585,
+      "step": 6537
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.00014182991551001112,
+      "loss": 0.9112,
+      "step": 6538
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.000141814063646152,
+      "loss": 0.8332,
+      "step": 6539
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.00014179821050884666,
+      "loss": 0.8812,
+      "step": 6540
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.00014178235609857792,
+      "loss": 1.1193,
+      "step": 6541
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.0001417665004158286,
+      "loss": 1.0176,
+      "step": 6542
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.00014175064346108162,
+      "loss": 0.7793,
+      "step": 6543
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.00014173478523481992,
+      "loss": 0.8621,
+      "step": 6544
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.0001417189257375264,
+      "loss": 0.9487,
+      "step": 6545
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.0001417030649696841,
+      "loss": 0.9308,
+      "step": 6546
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.00014168720293177605,
+      "loss": 0.9476,
+      "step": 6547
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.00014167133962428533,
+      "loss": 0.9063,
+      "step": 6548
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.000141655475047695,
+      "loss": 0.8397,
+      "step": 6549
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.00014163960920248836,
+      "loss": 0.9299,
+      "step": 6550
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.00014162374208914847,
+      "loss": 0.8508,
+      "step": 6551
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.0001416078737081586,
+      "loss": 0.9194,
+      "step": 6552
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.00014159200406000207,
+      "loss": 0.804,
+      "step": 6553
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.0001415761331451621,
+      "loss": 0.9612,
+      "step": 6554
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.0001415602609641221,
+      "loss": 0.8043,
+      "step": 6555
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.00014154438751736546,
+      "loss": 0.972,
+      "step": 6556
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.00014152851280537556,
+      "loss": 0.9207,
+      "step": 6557
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.0001415126368286359,
+      "loss": 0.9062,
+      "step": 6558
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.00014149675958763,
+      "loss": 0.9226,
+      "step": 6559
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.00014148088108284138,
+      "loss": 1.0479,
+      "step": 6560
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.00014146500131475358,
+      "loss": 0.9719,
+      "step": 6561
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.00014144912028385027,
+      "loss": 1.016,
+      "step": 6562
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.0001414332379906151,
+      "loss": 1.0389,
+      "step": 6563
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.00014141735443553175,
+      "loss": 0.8552,
+      "step": 6564
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.00014140146961908395,
+      "loss": 0.9955,
+      "step": 6565
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.00014138558354175547,
+      "loss": 0.8203,
+      "step": 6566
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.00014136969620403013,
+      "loss": 0.9391,
+      "step": 6567
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.0001413538076063918,
+      "loss": 0.8734,
+      "step": 6568
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.00014133791774932432,
+      "loss": 0.9962,
+      "step": 6569
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.00014132202663331167,
+      "loss": 0.9864,
+      "step": 6570
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.00014130613425883775,
+      "loss": 1.0693,
+      "step": 6571
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.0001412902406263866,
+      "loss": 0.9865,
+      "step": 6572
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.00014127434573644226,
+      "loss": 0.8599,
+      "step": 6573
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.00014125844958948883,
+      "loss": 0.8503,
+      "step": 6574
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.00014124255218601038,
+      "loss": 1.0015,
+      "step": 6575
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.00014122665352649108,
+      "loss": 0.9807,
+      "step": 6576
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.00014121075361141515,
+      "loss": 0.9241,
+      "step": 6577
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.00014119485244126678,
+      "loss": 0.8968,
+      "step": 6578
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.00014117895001653026,
+      "loss": 0.9669,
+      "step": 6579
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.0001411630463376899,
+      "loss": 0.9143,
+      "step": 6580
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.00014114714140523005,
+      "loss": 0.9524,
+      "step": 6581
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.0001411312352196351,
+      "loss": 0.8811,
+      "step": 6582
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.00014111532778138944,
+      "loss": 0.8902,
+      "step": 6583
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.00014109941909097755,
+      "loss": 1.0258,
+      "step": 6584
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.00014108350914888396,
+      "loss": 0.8937,
+      "step": 6585
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.00014106759795559314,
+      "loss": 0.8887,
+      "step": 6586
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.00014105168551158972,
+      "loss": 1.008,
+      "step": 6587
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.0001410357718173583,
+      "loss": 0.919,
+      "step": 6588
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.00014101985687338354,
+      "loss": 0.9426,
+      "step": 6589
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.00014100394068015007,
+      "loss": 0.9461,
+      "step": 6590
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.00014098802323814268,
+      "loss": 0.9039,
+      "step": 6591
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.00014097210454784613,
+      "loss": 0.8544,
+      "step": 6592
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.0001409561846097452,
+      "loss": 0.9125,
+      "step": 6593
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.00014094026342432477,
+      "loss": 1.0427,
+      "step": 6594
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.00014092434099206966,
+      "loss": 0.9768,
+      "step": 6595
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.00014090841731346484,
+      "loss": 0.8274,
+      "step": 6596
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.00014089249238899524,
+      "loss": 0.959,
+      "step": 6597
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.00014087656621914585,
+      "loss": 0.9913,
+      "step": 6598
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.0001408606388044017,
+      "loss": 0.7943,
+      "step": 6599
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.00014084471014524789,
+      "loss": 1.0308,
+      "step": 6600
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.00014082878024216953,
+      "loss": 1.049,
+      "step": 6601
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.00014081284909565168,
+      "loss": 0.8167,
+      "step": 6602
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.0001407969167061796,
+      "loss": 0.975,
+      "step": 6603
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.00014078098307423851,
+      "loss": 1.0838,
+      "step": 6604
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.00014076504820031364,
+      "loss": 0.9765,
+      "step": 6605
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.00014074911208489027,
+      "loss": 0.8713,
+      "step": 6606
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.00014073317472845383,
+      "loss": 0.8965,
+      "step": 6607
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.00014071723613148955,
+      "loss": 0.9269,
+      "step": 6608
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.00014070129629448297,
+      "loss": 0.9354,
+      "step": 6609
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.00014068535521791944,
+      "loss": 0.9279,
+      "step": 6610
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.00014066941290228447,
+      "loss": 0.8483,
+      "step": 6611
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.00014065346934806364,
+      "loss": 0.8475,
+      "step": 6612
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.00014063752455574247,
+      "loss": 0.8956,
+      "step": 6613
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.0001406215785258065,
+      "loss": 0.96,
+      "step": 6614
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.0001406056312587415,
+      "loss": 0.8904,
+      "step": 6615
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.000140589682755033,
+      "loss": 0.91,
+      "step": 6616
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.0001405737330151668,
+      "loss": 1.0182,
+      "step": 6617
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.00014055778203962867,
+      "loss": 1.095,
+      "step": 6618
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.0001405418298289043,
+      "loss": 0.7807,
+      "step": 6619
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.00014052587638347958,
+      "loss": 0.8089,
+      "step": 6620
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.00014050992170384035,
+      "loss": 0.8496,
+      "step": 6621
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.00014049396579047252,
+      "loss": 0.8683,
+      "step": 6622
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.00014047800864386205,
+      "loss": 0.9072,
+      "step": 6623
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.00014046205026449485,
+      "loss": 0.8622,
+      "step": 6624
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.00014044609065285698,
+      "loss": 0.8264,
+      "step": 6625
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.0001404301298094345,
+      "loss": 0.9852,
+      "step": 6626
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.00014041416773471346,
+      "loss": 0.982,
+      "step": 6627
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.00014039820442918,
+      "loss": 1.0645,
+      "step": 6628
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.00014038223989332034,
+      "loss": 0.8548,
+      "step": 6629
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.00014036627412762056,
+      "loss": 0.9626,
+      "step": 6630
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.000140350307132567,
+      "loss": 0.9381,
+      "step": 6631
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.00014033433890864585,
+      "loss": 0.8538,
+      "step": 6632
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.00014031836945634348,
+      "loss": 1.05,
+      "step": 6633
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.00014030239877614627,
+      "loss": 0.8162,
+      "step": 6634
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.00014028642686854053,
+      "loss": 0.8743,
+      "step": 6635
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.00014027045373401272,
+      "loss": 0.6993,
+      "step": 6636
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.0001402544793730493,
+      "loss": 0.8403,
+      "step": 6637
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.00014023850378613674,
+      "loss": 0.8313,
+      "step": 6638
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.00014022252697376162,
+      "loss": 0.7997,
+      "step": 6639
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.00014020654893641052,
+      "loss": 0.9423,
+      "step": 6640
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.00014019056967457,
+      "loss": 0.9697,
+      "step": 6641
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.00014017458918872677,
+      "loss": 1.0426,
+      "step": 6642
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.00014015860747936744,
+      "loss": 0.8821,
+      "step": 6643
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.0001401426245469788,
+      "loss": 0.9608,
+      "step": 6644
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.0001401266403920476,
+      "loss": 0.8432,
+      "step": 6645
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.00014011065501506057,
+      "loss": 0.9237,
+      "step": 6646
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.00014009466841650463,
+      "loss": 0.9809,
+      "step": 6647
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.0001400786805968666,
+      "loss": 0.8797,
+      "step": 6648
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.0001400626915566334,
+      "loss": 0.8493,
+      "step": 6649
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.00014004670129629201,
+      "loss": 1.1723,
+      "step": 6650
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.00014003070981632936,
+      "loss": 0.9058,
+      "step": 6651
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.00014001471711723248,
+      "loss": 0.9385,
+      "step": 6652
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.00013999872319948845,
+      "loss": 0.9224,
+      "step": 6653
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.00013998272806358436,
+      "loss": 0.8571,
+      "step": 6654
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.00013996673171000736,
+      "loss": 0.9078,
+      "step": 6655
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.00013995073413924458,
+      "loss": 0.8899,
+      "step": 6656
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.00013993473535178321,
+      "loss": 0.8983,
+      "step": 6657
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.00013991873534811058,
+      "loss": 0.8818,
+      "step": 6658
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.00013990273412871384,
+      "loss": 0.9947,
+      "step": 6659
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.00013988673169408041,
+      "loss": 0.9554,
+      "step": 6660
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.00013987072804469764,
+      "loss": 0.9491,
+      "step": 6661
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.00013985472318105288,
+      "loss": 0.9275,
+      "step": 6662
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.00013983871710363354,
+      "loss": 0.8874,
+      "step": 6663
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.0001398227098129272,
+      "loss": 0.935,
+      "step": 6664
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.00013980670130942118,
+      "loss": 0.9046,
+      "step": 6665
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.00013979069159360317,
+      "loss": 0.8272,
+      "step": 6666
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.00013977468066596067,
+      "loss": 0.996,
+      "step": 6667
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.00013975866852698133,
+      "loss": 0.8436,
+      "step": 6668
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.00013974265517715278,
+      "loss": 0.9798,
+      "step": 6669
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.0001397266406169627,
+      "loss": 0.872,
+      "step": 6670
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.0001397106248468988,
+      "loss": 0.9399,
+      "step": 6671
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.0001396946078674489,
+      "loss": 0.7784,
+      "step": 6672
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.00013967858967910076,
+      "loss": 0.8494,
+      "step": 6673
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.0001396625702823422,
+      "loss": 0.9406,
+      "step": 6674
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.0001396465496776611,
+      "loss": 0.8736,
+      "step": 6675
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.00013963052786554537,
+      "loss": 0.9316,
+      "step": 6676
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.00013961450484648299,
+      "loss": 0.8389,
+      "step": 6677
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.00013959848062096187,
+      "loss": 0.9306,
+      "step": 6678
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.00013958245518947007,
+      "loss": 0.9492,
+      "step": 6679
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.00013956642855249564,
+      "loss": 0.8513,
+      "step": 6680
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.00013955040071052664,
+      "loss": 0.8244,
+      "step": 6681
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.00013953437166405123,
+      "loss": 0.8707,
+      "step": 6682
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.0001395183414135576,
+      "loss": 0.9056,
+      "step": 6683
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.00013950230995953388,
+      "loss": 0.8743,
+      "step": 6684
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.0001394862773024684,
+      "loss": 0.8818,
+      "step": 6685
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.00013947024344284937,
+      "loss": 0.9615,
+      "step": 6686
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.00013945420838116506,
+      "loss": 0.9191,
+      "step": 6687
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.00013943817211790392,
+      "loss": 0.9247,
+      "step": 6688
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.00013942213465355425,
+      "loss": 0.9289,
+      "step": 6689
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.0001394060959886045,
+      "loss": 0.9727,
+      "step": 6690
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.00013939005612354318,
+      "loss": 1.0353,
+      "step": 6691
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.0001393740150588587,
+      "loss": 0.906,
+      "step": 6692
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.00013935797279503962,
+      "loss": 1.0515,
+      "step": 6693
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.00013934192933257454,
+      "loss": 0.8493,
+      "step": 6694
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.000139325884671952,
+      "loss": 0.8638,
+      "step": 6695
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.0001393098388136607,
+      "loss": 0.9702,
+      "step": 6696
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.00013929379175818928,
+      "loss": 0.8457,
+      "step": 6697
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.00013927774350602646,
+      "loss": 0.9326,
+      "step": 6698
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.00013926169405766101,
+      "loss": 0.8707,
+      "step": 6699
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.00013924564341358166,
+      "loss": 0.945,
+      "step": 6700
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.0001392295915742773,
+      "loss": 0.9731,
+      "step": 6701
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.00013921353854023678,
+      "loss": 1.0431,
+      "step": 6702
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.00013919748431194892,
+      "loss": 0.9077,
+      "step": 6703
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.00013918142888990275,
+      "loss": 0.8757,
+      "step": 6704
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.00013916537227458715,
+      "loss": 0.9017,
+      "step": 6705
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.00013914931446649118,
+      "loss": 0.9036,
+      "step": 6706
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.00013913325546610386,
+      "loss": 0.8624,
+      "step": 6707
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.00013911719527391432,
+      "loss": 0.9203,
+      "step": 6708
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.00013910113389041157,
+      "loss": 0.8841,
+      "step": 6709
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.00013908507131608485,
+      "loss": 0.9303,
+      "step": 6710
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.0001390690075514233,
+      "loss": 0.7995,
+      "step": 6711
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.00013905294259691613,
+      "loss": 0.8294,
+      "step": 6712
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.00013903687645305262,
+      "loss": 0.9477,
+      "step": 6713
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.0001390208091203221,
+      "loss": 0.9048,
+      "step": 6714
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.00013900474059921382,
+      "loss": 0.9317,
+      "step": 6715
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.00013898867089021718,
+      "loss": 0.807,
+      "step": 6716
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.00013897259999382162,
+      "loss": 0.8827,
+      "step": 6717
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.00013895652791051655,
+      "loss": 0.9226,
+      "step": 6718
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.00013894045464079145,
+      "loss": 0.8621,
+      "step": 6719
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.0001389243801851358,
+      "loss": 0.9532,
+      "step": 6720
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.0001389083045440392,
+      "loss": 0.8746,
+      "step": 6721
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.0001388922277179912,
+      "loss": 0.9206,
+      "step": 6722
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.00013887614970748143,
+      "loss": 0.7996,
+      "step": 6723
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.00013886007051299953,
+      "loss": 1.0645,
+      "step": 6724
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.0001388439901350352,
+      "loss": 0.8237,
+      "step": 6725
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.0001388279085740782,
+      "loss": 0.7972,
+      "step": 6726
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.00013881182583061821,
+      "loss": 0.8909,
+      "step": 6727
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.0001387957419051451,
+      "loss": 0.87,
+      "step": 6728
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.0001387796567981487,
+      "loss": 0.9674,
+      "step": 6729
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.00013876357051011886,
+      "loss": 0.7733,
+      "step": 6730
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.00013874748304154553,
+      "loss": 0.9822,
+      "step": 6731
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.0001387313943929186,
+      "loss": 1.0318,
+      "step": 6732
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.00013871530456472807,
+      "loss": 0.9485,
+      "step": 6733
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.00013869921355746394,
+      "loss": 0.8626,
+      "step": 6734
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.0001386831213716163,
+      "loss": 0.8957,
+      "step": 6735
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.00013866702800767518,
+      "loss": 0.885,
+      "step": 6736
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.0001386509334661308,
+      "loss": 0.7858,
+      "step": 6737
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.0001386348377474732,
+      "loss": 0.8556,
+      "step": 6738
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.00013861874085219263,
+      "loss": 0.9273,
+      "step": 6739
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.00013860264278077935,
+      "loss": 0.7801,
+      "step": 6740
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.00013858654353372357,
+      "loss": 0.9613,
+      "step": 6741
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.00013857044311151566,
+      "loss": 0.8951,
+      "step": 6742
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.00013855434151464587,
+      "loss": 0.9612,
+      "step": 6743
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.00013853823874360465,
+      "loss": 1.0257,
+      "step": 6744
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.0001385221347988824,
+      "loss": 0.8897,
+      "step": 6745
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.00013850602968096948,
+      "loss": 0.9595,
+      "step": 6746
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.0001384899233903565,
+      "loss": 1.0185,
+      "step": 6747
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.0001384738159275339,
+      "loss": 0.9489,
+      "step": 6748
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.00013845770729299225,
+      "loss": 0.924,
+      "step": 6749
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.0001384415974872221,
+      "loss": 0.8043,
+      "step": 6750
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.0001384254865107141,
+      "loss": 0.9834,
+      "step": 6751
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.00013840937436395895,
+      "loss": 1.0354,
+      "step": 6752
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.00013839326104744733,
+      "loss": 1.0065,
+      "step": 6753
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.00013837714656166994,
+      "loss": 0.8641,
+      "step": 6754
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.0001383610309071175,
+      "loss": 0.9181,
+      "step": 6755
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.00013834491408428097,
+      "loss": 0.9669,
+      "step": 6756
+    },
+    {
+      "epoch": 1.48,
+      "eval_loss": 0.5813515186309814,
+      "eval_runtime": 323.4477,
+      "eval_samples_per_second": 56.08,
+      "eval_steps_per_second": 28.042,
+      "step": 6756
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.00013832879609365102,
+      "loss": 1.003,
+      "step": 6757
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.0001383126769357186,
+      "loss": 0.9497,
+      "step": 6758
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.00013829655661097462,
+      "loss": 0.9183,
+      "step": 6759
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.00013828043511991,
+      "loss": 1.0175,
+      "step": 6760
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.00013826431246301578,
+      "loss": 0.8512,
+      "step": 6761
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.00013824818864078287,
+      "loss": 0.9177,
+      "step": 6762
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.00013823206365370239,
+      "loss": 0.9708,
+      "step": 6763
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.00013821593750226543,
+      "loss": 0.8938,
+      "step": 6764
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.00013819981018696307,
+      "loss": 0.9185,
+      "step": 6765
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.00013818368170828653,
+      "loss": 0.8515,
+      "step": 6766
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.00013816755206672696,
+      "loss": 0.8007,
+      "step": 6767
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.00013815142126277557,
+      "loss": 0.8844,
+      "step": 6768
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.0001381352892969237,
+      "loss": 1.0909,
+      "step": 6769
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.0001381191561696625,
+      "loss": 1.0035,
+      "step": 6770
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.00013810302188148348,
+      "loss": 0.7174,
+      "step": 6771
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.0001380868864328779,
+      "loss": 0.8169,
+      "step": 6772
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.0001380707498243372,
+      "loss": 0.9545,
+      "step": 6773
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.0001380546120563528,
+      "loss": 0.9109,
+      "step": 6774
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.0001380384731294162,
+      "loss": 0.9466,
+      "step": 6775
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.0001380223330440189,
+      "loss": 0.8275,
+      "step": 6776
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.00013800619180065243,
+      "loss": 0.9042,
+      "step": 6777
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.0001379900493998084,
+      "loss": 0.8505,
+      "step": 6778
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.00013797390584197841,
+      "loss": 0.9285,
+      "step": 6779
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.00013795776112765414,
+      "loss": 0.8859,
+      "step": 6780
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.00013794161525732722,
+      "loss": 0.9023,
+      "step": 6781
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.0001379254682314894,
+      "loss": 1.0259,
+      "step": 6782
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.00013790932005063248,
+      "loss": 0.9202,
+      "step": 6783
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.00013789317071524816,
+      "loss": 0.8923,
+      "step": 6784
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.00013787702022582834,
+      "loss": 0.9495,
+      "step": 6785
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.00013786086858286488,
+      "loss": 0.9322,
+      "step": 6786
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.00013784471578684966,
+      "loss": 0.8264,
+      "step": 6787
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.0001378285618382746,
+      "loss": 0.9116,
+      "step": 6788
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.00013781240673763168,
+      "loss": 0.9356,
+      "step": 6789
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.0001377962504854129,
+      "loss": 0.7758,
+      "step": 6790
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.00013778009308211032,
+      "loss": 0.8595,
+      "step": 6791
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.000137763934528216,
+      "loss": 0.9531,
+      "step": 6792
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.000137747774824222,
+      "loss": 0.8686,
+      "step": 6793
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.0001377316139706206,
+      "loss": 1.0573,
+      "step": 6794
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.0001377154519679038,
+      "loss": 0.9438,
+      "step": 6795
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.00013769928881656392,
+      "loss": 0.8695,
+      "step": 6796
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.0001376831245170932,
+      "loss": 0.9756,
+      "step": 6797
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.00013766695906998393,
+      "loss": 1.0019,
+      "step": 6798
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.0001376507924757284,
+      "loss": 0.8922,
+      "step": 6799
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.00013763462473481898,
+      "loss": 0.9548,
+      "step": 6800
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.00013761845584774803,
+      "loss": 0.8333,
+      "step": 6801
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.00013760228581500802,
+      "loss": 0.9498,
+      "step": 6802
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.00013758611463709133,
+      "loss": 1.0282,
+      "step": 6803
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.00013756994231449055,
+      "loss": 0.8441,
+      "step": 6804
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.00013755376884769813,
+      "loss": 0.9417,
+      "step": 6805
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.00013753759423720672,
+      "loss": 0.8017,
+      "step": 6806
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.00013752141848350885,
+      "loss": 0.9399,
+      "step": 6807
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.0001375052415870971,
+      "loss": 1.0084,
+      "step": 6808
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.00013748906354846426,
+      "loss": 1.0951,
+      "step": 6809
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.00013747288436810295,
+      "loss": 0.8805,
+      "step": 6810
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.00013745670404650593,
+      "loss": 0.9252,
+      "step": 6811
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.000137440522584166,
+      "loss": 0.954,
+      "step": 6812
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.0001374243399815759,
+      "loss": 0.9076,
+      "step": 6813
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.00013740815623922853,
+      "loss": 0.8918,
+      "step": 6814
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.00013739197135761674,
+      "loss": 0.8736,
+      "step": 6815
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.00013737578533723344,
+      "loss": 0.8329,
+      "step": 6816
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.0001373595981785716,
+      "loss": 0.7904,
+      "step": 6817
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.00013734340988212418,
+      "loss": 0.9551,
+      "step": 6818
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.00013732722044838415,
+      "loss": 0.9438,
+      "step": 6819
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.0001373110298778446,
+      "loss": 0.8398,
+      "step": 6820
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.00013729483817099866,
+      "loss": 0.8919,
+      "step": 6821
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.00013727864532833939,
+      "loss": 0.8512,
+      "step": 6822
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.00013726245135035992,
+      "loss": 0.9223,
+      "step": 6823
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.00013724625623755355,
+      "loss": 0.9344,
+      "step": 6824
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.00013723005999041334,
+      "loss": 0.8416,
+      "step": 6825
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.00013721386260943268,
+      "loss": 0.8563,
+      "step": 6826
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.00013719766409510476,
+      "loss": 0.8147,
+      "step": 6827
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.000137181464447923,
+      "loss": 0.8904,
+      "step": 6828
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.00013716526366838071,
+      "loss": 0.9609,
+      "step": 6829
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.0001371490617569713,
+      "loss": 0.8142,
+      "step": 6830
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.00013713285871418815,
+      "loss": 0.8082,
+      "step": 6831
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.00013711665454052481,
+      "loss": 0.9409,
+      "step": 6832
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.0001371004492364747,
+      "loss": 0.9654,
+      "step": 6833
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.0001370842428025314,
+      "loss": 0.7849,
+      "step": 6834
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.00013706803523918845,
+      "loss": 0.9473,
+      "step": 6835
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.00013705182654693944,
+      "loss": 0.8989,
+      "step": 6836
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.00013703561672627807,
+      "loss": 0.849,
+      "step": 6837
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.00013701940577769793,
+      "loss": 0.8338,
+      "step": 6838
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.00013700319370169275,
+      "loss": 0.9657,
+      "step": 6839
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.00013698698049875632,
+      "loss": 0.8997,
+      "step": 6840
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.00013697076616938234,
+      "loss": 0.8287,
+      "step": 6841
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.00013695455071406463,
+      "loss": 1.0055,
+      "step": 6842
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.0001369383341332971,
+      "loss": 0.9579,
+      "step": 6843
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.00013692211642757352,
+      "loss": 0.9753,
+      "step": 6844
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.00013690589759738787,
+      "loss": 0.8904,
+      "step": 6845
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.0001368896776432341,
+      "loss": 0.7948,
+      "step": 6846
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.00013687345656560612,
+      "loss": 0.8713,
+      "step": 6847
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.00013685723436499805,
+      "loss": 0.8089,
+      "step": 6848
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.00013684101104190381,
+      "loss": 0.9882,
+      "step": 6849
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.00013682478659681756,
+      "loss": 0.9147,
+      "step": 6850
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.00013680856103023343,
+      "loss": 0.9046,
+      "step": 6851
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.00013679233434264556,
+      "loss": 0.9053,
+      "step": 6852
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.00013677610653454808,
+      "loss": 0.8416,
+      "step": 6853
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.00013675987760643522,
+      "loss": 0.9046,
+      "step": 6854
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.00013674364755880127,
+      "loss": 0.8521,
+      "step": 6855
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.0001367274163921405,
+      "loss": 0.9753,
+      "step": 6856
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.00013671118410694724,
+      "loss": 1.0064,
+      "step": 6857
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.00013669495070371581,
+      "loss": 0.87,
+      "step": 6858
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.00013667871618294064,
+      "loss": 0.9234,
+      "step": 6859
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.0001366624805451161,
+      "loss": 0.908,
+      "step": 6860
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.0001366462437907367,
+      "loss": 0.9048,
+      "step": 6861
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.0001366300059202969,
+      "loss": 0.7732,
+      "step": 6862
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.00013661376693429128,
+      "loss": 0.9099,
+      "step": 6863
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.0001365975268332143,
+      "loss": 0.8616,
+      "step": 6864
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.0001365812856175606,
+      "loss": 0.8605,
+      "step": 6865
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.0001365650432878248,
+      "loss": 0.9193,
+      "step": 6866
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.0001365487998445016,
+      "loss": 0.9379,
+      "step": 6867
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.00013653255528808561,
+      "loss": 0.8703,
+      "step": 6868
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.00013651630961907164,
+      "loss": 0.984,
+      "step": 6869
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.00013650006283795444,
+      "loss": 0.896,
+      "step": 6870
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.00013648381494522875,
+      "loss": 0.8773,
+      "step": 6871
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.00013646756594138946,
+      "loss": 0.9696,
+      "step": 6872
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.0001364513158269314,
+      "loss": 0.9583,
+      "step": 6873
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.00013643506460234942,
+      "loss": 0.9074,
+      "step": 6874
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.00013641881226813857,
+      "loss": 0.9576,
+      "step": 6875
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.00013640255882479374,
+      "loss": 0.869,
+      "step": 6876
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.0001363863042728099,
+      "loss": 0.8666,
+      "step": 6877
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.00013637004861268215,
+      "loss": 0.9533,
+      "step": 6878
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.0001363537918449055,
+      "loss": 0.8687,
+      "step": 6879
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.0001363375339699751,
+      "loss": 0.9456,
+      "step": 6880
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.00013632127498838605,
+      "loss": 0.9013,
+      "step": 6881
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.00013630501490063352,
+      "loss": 0.92,
+      "step": 6882
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.0001362887537072127,
+      "loss": 0.9243,
+      "step": 6883
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.00013627249140861887,
+      "loss": 0.7989,
+      "step": 6884
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.0001362562280053472,
+      "loss": 0.9815,
+      "step": 6885
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.00013623996349789317,
+      "loss": 0.756,
+      "step": 6886
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.0001362236978867519,
+      "loss": 0.8537,
+      "step": 6887
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.0001362074311724189,
+      "loss": 0.9446,
+      "step": 6888
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.00013619116335538952,
+      "loss": 1.0054,
+      "step": 6889
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.00013617489443615923,
+      "loss": 0.9014,
+      "step": 6890
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.00013615862441522349,
+      "loss": 0.8204,
+      "step": 6891
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.00013614235329307777,
+      "loss": 0.9393,
+      "step": 6892
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.00013612608107021763,
+      "loss": 0.8691,
+      "step": 6893
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.00013610980774713863,
+      "loss": 0.9321,
+      "step": 6894
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.0001360935333243364,
+      "loss": 0.9568,
+      "step": 6895
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.00013607725780230657,
+      "loss": 0.8597,
+      "step": 6896
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.00013606098118154478,
+      "loss": 1.0556,
+      "step": 6897
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.00013604470346254677,
+      "loss": 0.863,
+      "step": 6898
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.00013602842464580824,
+      "loss": 0.8812,
+      "step": 6899
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.000136012144731825,
+      "loss": 0.8795,
+      "step": 6900
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.00013599586372109283,
+      "loss": 1.066,
+      "step": 6901
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.00013597958161410756,
+      "loss": 0.9384,
+      "step": 6902
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.0001359632984113651,
+      "loss": 0.8729,
+      "step": 6903
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.00013594701411336134,
+      "loss": 0.9522,
+      "step": 6904
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.0001359307287205922,
+      "loss": 0.9678,
+      "step": 6905
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.00013591444223355365,
+      "loss": 0.8054,
+      "step": 6906
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.00013589815465274172,
+      "loss": 0.7799,
+      "step": 6907
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.0001358818659786524,
+      "loss": 0.8933,
+      "step": 6908
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.00013586557621178184,
+      "loss": 1.0058,
+      "step": 6909
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.00013584928535262608,
+      "loss": 0.7018,
+      "step": 6910
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.00013583299340168126,
+      "loss": 0.9374,
+      "step": 6911
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.00013581670035944356,
+      "loss": 1.0444,
+      "step": 6912
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.00013580040622640922,
+      "loss": 0.8845,
+      "step": 6913
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.00013578411100307442,
+      "loss": 0.9592,
+      "step": 6914
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.00013576781468993546,
+      "loss": 0.8914,
+      "step": 6915
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.00013575151728748868,
+      "loss": 0.817,
+      "step": 6916
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.00013573521879623035,
+      "loss": 0.8753,
+      "step": 6917
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.00013571891921665687,
+      "loss": 0.9368,
+      "step": 6918
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.00013570261854926465,
+      "loss": 0.8986,
+      "step": 6919
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.00013568631679455008,
+      "loss": 0.9959,
+      "step": 6920
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.00013567001395300973,
+      "loss": 0.9486,
+      "step": 6921
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.00013565371002514,
+      "loss": 0.9292,
+      "step": 6922
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.0001356374050114375,
+      "loss": 0.9318,
+      "step": 6923
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.00013562109891239875,
+      "loss": 0.846,
+      "step": 6924
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.00013560479172852033,
+      "loss": 0.9608,
+      "step": 6925
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.00013558848346029897,
+      "loss": 0.8848,
+      "step": 6926
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.00013557217410823126,
+      "loss": 0.7882,
+      "step": 6927
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.00013555586367281392,
+      "loss": 1.0629,
+      "step": 6928
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.0001355395521545437,
+      "loss": 0.8496,
+      "step": 6929
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.00013552323955391734,
+      "loss": 1.0507,
+      "step": 6930
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.0001355069258714316,
+      "loss": 0.7903,
+      "step": 6931
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.00013549061110758345,
+      "loss": 0.9556,
+      "step": 6932
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.00013547429526286964,
+      "loss": 0.9934,
+      "step": 6933
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.00013545797833778707,
+      "loss": 0.9589,
+      "step": 6934
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.00013544166033283272,
+      "loss": 0.9011,
+      "step": 6935
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.00013542534124850355,
+      "loss": 0.8097,
+      "step": 6936
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.0001354090210852965,
+      "loss": 0.9168,
+      "step": 6937
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.0001353926998437087,
+      "loss": 0.8446,
+      "step": 6938
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.00013537637752423712,
+      "loss": 0.8702,
+      "step": 6939
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.0001353600541273789,
+      "loss": 0.8653,
+      "step": 6940
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.00013534372965363117,
+      "loss": 1.0191,
+      "step": 6941
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.00013532740410349104,
+      "loss": 0.8981,
+      "step": 6942
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.00013531107747745576,
+      "loss": 0.9184,
+      "step": 6943
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.0001352947497760226,
+      "loss": 0.8609,
+      "step": 6944
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.00013527842099968868,
+      "loss": 0.8517,
+      "step": 6945
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.00013526209114895143,
+      "loss": 0.9707,
+      "step": 6946
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.00013524576022430808,
+      "loss": 0.8184,
+      "step": 6947
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.00013522942822625605,
+      "loss": 0.9553,
+      "step": 6948
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.0001352130951552927,
+      "loss": 1.0028,
+      "step": 6949
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.00013519676101191548,
+      "loss": 0.944,
+      "step": 6950
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.00013518042579662183,
+      "loss": 0.7947,
+      "step": 6951
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.0001351640895099092,
+      "loss": 0.9258,
+      "step": 6952
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.00013514775215227518,
+      "loss": 1.04,
+      "step": 6953
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.00013513141372421727,
+      "loss": 0.7911,
+      "step": 6954
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.0001351150742262331,
+      "loss": 0.9702,
+      "step": 6955
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.00013509873365882025,
+      "loss": 0.8992,
+      "step": 6956
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.00013508239202247638,
+      "loss": 0.8852,
+      "step": 6957
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.00013506604931769916,
+      "loss": 0.9454,
+      "step": 6958
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.00013504970554498638,
+      "loss": 0.8359,
+      "step": 6959
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.00013503336070483572,
+      "loss": 0.8567,
+      "step": 6960
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.00013501701479774498,
+      "loss": 0.8213,
+      "step": 6961
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.000135000667824212,
+      "loss": 0.9609,
+      "step": 6962
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.00013498431978473455,
+      "loss": 0.8403,
+      "step": 6963
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.00013496797067981056,
+      "loss": 1.0112,
+      "step": 6964
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.00013495162050993796,
+      "loss": 0.932,
+      "step": 6965
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.00013493526927561467,
+      "loss": 0.9797,
+      "step": 6966
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.0001349189169773387,
+      "loss": 0.9071,
+      "step": 6967
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.000134902563615608,
+      "loss": 0.9607,
+      "step": 6968
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.00013488620919092063,
+      "loss": 1.0147,
+      "step": 6969
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.0001348698537037747,
+      "loss": 0.8959,
+      "step": 6970
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.00013485349715466825,
+      "loss": 0.9007,
+      "step": 6971
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.00013483713954409954,
+      "loss": 0.8975,
+      "step": 6972
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.0001348207808725666,
+      "loss": 0.8835,
+      "step": 6973
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.00013480442114056768,
+      "loss": 0.9536,
+      "step": 6974
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.0001347880603486011,
+      "loss": 1.0024,
+      "step": 6975
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.00013477169849716498,
+      "loss": 0.7954,
+      "step": 6976
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.0001347553355867577,
+      "loss": 0.8618,
+      "step": 6977
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.00013473897161787762,
+      "loss": 1.0549,
+      "step": 6978
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.00013472260659102306,
+      "loss": 0.8638,
+      "step": 6979
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.00013470624050669242,
+      "loss": 1.0027,
+      "step": 6980
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.00013468987336538415,
+      "loss": 0.9683,
+      "step": 6981
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.00013467350516759667,
+      "loss": 0.9558,
+      "step": 6982
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.00013465713591382853,
+      "loss": 0.9849,
+      "step": 6983
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.00013464076560457824,
+      "loss": 0.8277,
+      "step": 6984
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.0001346243942403443,
+      "loss": 0.903,
+      "step": 6985
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.00013460802182162538,
+      "loss": 0.9913,
+      "step": 6986
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.00013459164834892003,
+      "loss": 0.8778,
+      "step": 6987
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.00013457527382272692,
+      "loss": 0.84,
+      "step": 6988
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.0001345588982435448,
+      "loss": 0.9112,
+      "step": 6989
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.00013454252161187232,
+      "loss": 0.8622,
+      "step": 6990
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.00013452614392820827,
+      "loss": 0.8927,
+      "step": 6991
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.00013450976519305141,
+      "loss": 0.9455,
+      "step": 6992
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.00013449338540690053,
+      "loss": 0.9335,
+      "step": 6993
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.00013447700457025453,
+      "loss": 0.7967,
+      "step": 6994
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.00013446062268361223,
+      "loss": 0.8392,
+      "step": 6995
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.0001344442397474726,
+      "loss": 0.7994,
+      "step": 6996
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.00013442785576233454,
+      "loss": 0.9245,
+      "step": 6997
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.00013441147072869704,
+      "loss": 0.959,
+      "step": 6998
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.0001343950846470591,
+      "loss": 0.8491,
+      "step": 6999
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.00013437869751791977,
+      "loss": 0.8545,
+      "step": 7000
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.0001343623093417781,
+      "loss": 0.8758,
+      "step": 7001
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.0001343459201191332,
+      "loss": 0.9108,
+      "step": 7002
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.00013432952985048421,
+      "loss": 0.8479,
+      "step": 7003
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.00013431313853633027,
+      "loss": 0.8765,
+      "step": 7004
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.00013429674617717062,
+      "loss": 1.119,
+      "step": 7005
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.00013428035277350444,
+      "loss": 0.9576,
+      "step": 7006
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.00013426395832583104,
+      "loss": 1.0267,
+      "step": 7007
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.0001342475628346497,
+      "loss": 0.9702,
+      "step": 7008
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.00013423116630045971,
+      "loss": 1.0765,
+      "step": 7009
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.00013421476872376044,
+      "loss": 0.8774,
+      "step": 7010
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.0001341983701050513,
+      "loss": 0.8725,
+      "step": 7011
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.00013418197044483168,
+      "loss": 0.8707,
+      "step": 7012
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.0001341655697436011,
+      "loss": 0.8817,
+      "step": 7013
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.00013414916800185894,
+      "loss": 1.0369,
+      "step": 7014
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.00013413276522010475,
+      "loss": 0.9339,
+      "step": 7015
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.00013411636139883814,
+      "loss": 0.9063,
+      "step": 7016
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.0001340999565385586,
+      "loss": 0.9526,
+      "step": 7017
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.0001340835506397658,
+      "loss": 0.8635,
+      "step": 7018
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.0001340671437029594,
+      "loss": 0.9515,
+      "step": 7019
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.00013405073572863896,
+      "loss": 0.9122,
+      "step": 7020
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.0001340343267173043,
+      "loss": 0.7963,
+      "step": 7021
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.00013401791666945508,
+      "loss": 1.0382,
+      "step": 7022
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.0001340015055855911,
+      "loss": 0.9478,
+      "step": 7023
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.0001339850934662122,
+      "loss": 0.8143,
+      "step": 7024
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.00013396868031181815,
+      "loss": 0.9843,
+      "step": 7025
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.00013395226612290882,
+      "loss": 0.9886,
+      "step": 7026
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.00013393585089998414,
+      "loss": 0.9354,
+      "step": 7027
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.000133919434643544,
+      "loss": 0.8295,
+      "step": 7028
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.00013390301735408838,
+      "loss": 0.9445,
+      "step": 7029
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.00013388659903211726,
+      "loss": 0.9083,
+      "step": 7030
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.00013387017967813063,
+      "loss": 1.045,
+      "step": 7031
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.00013385375929262858,
+      "loss": 0.8111,
+      "step": 7032
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.00013383733787611116,
+      "loss": 0.885,
+      "step": 7033
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.0001338209154290785,
+      "loss": 0.8496,
+      "step": 7034
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.0001338044919520308,
+      "loss": 0.9083,
+      "step": 7035
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.00013378806744546816,
+      "loss": 0.9076,
+      "step": 7036
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.0001337716419098908,
+      "loss": 0.8519,
+      "step": 7037
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.00013375521534579897,
+      "loss": 0.8716,
+      "step": 7038
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.00013373878775369295,
+      "loss": 0.8288,
+      "step": 7039
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.00013372235913407306,
+      "loss": 0.8876,
+      "step": 7040
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.00013370592948743957,
+      "loss": 0.9662,
+      "step": 7041
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.0001336894988142929,
+      "loss": 0.8233,
+      "step": 7042
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.00013367306711513344,
+      "loss": 0.8986,
+      "step": 7043
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.00013365663439046153,
+      "loss": 0.9433,
+      "step": 7044
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.00013364020064077778,
+      "loss": 1.0393,
+      "step": 7045
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.0001336237658665826,
+      "loss": 0.8549,
+      "step": 7046
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.0001336073300683765,
+      "loss": 0.9714,
+      "step": 7047
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.00013359089324666004,
+      "loss": 1.0295,
+      "step": 7048
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.00013357445540193382,
+      "loss": 0.7622,
+      "step": 7049
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.0001335580165346984,
+      "loss": 0.9748,
+      "step": 7050
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.0001335415766454545,
+      "loss": 0.7691,
+      "step": 7051
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.00013352513573470275,
+      "loss": 0.8416,
+      "step": 7052
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.0001335086938029439,
+      "loss": 0.9035,
+      "step": 7053
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.00013349225085067864,
+      "loss": 0.8062,
+      "step": 7054
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.00013347580687840776,
+      "loss": 0.9658,
+      "step": 7055
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.00013345936188663207,
+      "loss": 0.8791,
+      "step": 7056
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.00013344291587585237,
+      "loss": 0.8396,
+      "step": 7057
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.00013342646884656955,
+      "loss": 0.9439,
+      "step": 7058
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.00013341002079928453,
+      "loss": 0.8005,
+      "step": 7059
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.00013339357173449817,
+      "loss": 0.9831,
+      "step": 7060
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.00013337712165271148,
+      "loss": 0.7646,
+      "step": 7061
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.00013336067055442542,
+      "loss": 0.9419,
+      "step": 7062
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.000133344218440141,
+      "loss": 0.9868,
+      "step": 7063
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.00013332776531035928,
+      "loss": 0.832,
+      "step": 7064
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.00013331131116558136,
+      "loss": 0.8991,
+      "step": 7065
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.00013329485600630835,
+      "loss": 0.8772,
+      "step": 7066
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.00013327839983304134,
+      "loss": 0.9423,
+      "step": 7067
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.00013326194264628156,
+      "loss": 0.867,
+      "step": 7068
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.00013324548444653016,
+      "loss": 1.0366,
+      "step": 7069
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.00013322902523428844,
+      "loss": 0.866,
+      "step": 7070
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.0001332125650100576,
+      "loss": 0.86,
+      "step": 7071
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.00013319610377433898,
+      "loss": 0.9374,
+      "step": 7072
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.0001331796415276339,
+      "loss": 0.8775,
+      "step": 7073
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.00013316317827044368,
+      "loss": 0.7995,
+      "step": 7074
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.00013314671400326976,
+      "loss": 0.8834,
+      "step": 7075
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.00013313024872661352,
+      "loss": 0.9064,
+      "step": 7076
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.0001331137824409764,
+      "loss": 0.8905,
+      "step": 7077
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.00013309731514685996,
+      "loss": 0.9636,
+      "step": 7078
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.00013308084684476564,
+      "loss": 0.9381,
+      "step": 7079
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.00013306437753519496,
+      "loss": 0.7942,
+      "step": 7080
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.00013304790721864956,
+      "loss": 0.9638,
+      "step": 7081
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.00013303143589563104,
+      "loss": 0.8704,
+      "step": 7082
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.00013301496356664096,
+      "loss": 0.9118,
+      "step": 7083
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.00013299849023218105,
+      "loss": 0.9958,
+      "step": 7084
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.00013298201589275296,
+      "loss": 0.8872,
+      "step": 7085
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.0001329655405488585,
+      "loss": 0.867,
+      "step": 7086
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.0001329490642009993,
+      "loss": 0.8608,
+      "step": 7087
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.00013293258684967727,
+      "loss": 0.8854,
+      "step": 7088
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.0001329161084953942,
+      "loss": 0.8411,
+      "step": 7089
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.0001328996291386518,
+      "loss": 0.9192,
+      "step": 7090
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.00013288314877995213,
+      "loss": 0.9859,
+      "step": 7091
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.000132866667419797,
+      "loss": 0.8586,
+      "step": 7092
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.00013285018505868843,
+      "loss": 0.9315,
+      "step": 7093
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.00013283370169712828,
+      "loss": 0.9855,
+      "step": 7094
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.00013281721733561863,
+      "loss": 0.959,
+      "step": 7095
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.00013280073197466147,
+      "loss": 0.9402,
+      "step": 7096
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.0001327842456147589,
+      "loss": 0.9257,
+      "step": 7097
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.00013276775825641298,
+      "loss": 0.9806,
+      "step": 7098
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.00013275126990012584,
+      "loss": 0.8605,
+      "step": 7099
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.00013273478054639964,
+      "loss": 0.8485,
+      "step": 7100
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.00013271829019573657,
+      "loss": 0.954,
+      "step": 7101
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.00013270179884863878,
+      "loss": 0.9455,
+      "step": 7102
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.0001326853065056086,
+      "loss": 0.9221,
+      "step": 7103
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.00013266881316714827,
+      "loss": 0.8851,
+      "step": 7104
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.0001326523188337601,
+      "loss": 0.8492,
+      "step": 7105
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.00013263582350594644,
+      "loss": 0.893,
+      "step": 7106
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.00013261932718420957,
+      "loss": 0.9166,
+      "step": 7107
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.000132602829869052,
+      "loss": 0.9044,
+      "step": 7108
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.00013258633156097607,
+      "loss": 0.9071,
+      "step": 7109
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.00013256983226048424,
+      "loss": 0.8889,
+      "step": 7110
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.0001325533319680791,
+      "loss": 1.1134,
+      "step": 7111
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.00013253683068426307,
+      "loss": 1.0037,
+      "step": 7112
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.00013252032840953867,
+      "loss": 1.0489,
+      "step": 7113
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.00013250382514440855,
+      "loss": 0.7987,
+      "step": 7114
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.00013248732088937527,
+      "loss": 0.9245,
+      "step": 7115
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.00013247081564494152,
+      "loss": 0.8801,
+      "step": 7116
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.00013245430941160992,
+      "loss": 1.011,
+      "step": 7117
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.00013243780218988316,
+      "loss": 0.8569,
+      "step": 7118
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.00013242129398026399,
+      "loss": 0.8086,
+      "step": 7119
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.00013240478478325515,
+      "loss": 0.926,
+      "step": 7120
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.00013238827459935947,
+      "loss": 0.954,
+      "step": 7121
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.00013237176342907974,
+      "loss": 0.9769,
+      "step": 7122
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.0001323552512729188,
+      "loss": 0.9602,
+      "step": 7123
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.00013233873813137952,
+      "loss": 0.8889,
+      "step": 7124
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.00013232222400496482,
+      "loss": 0.8092,
+      "step": 7125
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.0001323057088941776,
+      "loss": 0.9111,
+      "step": 7126
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.00013228919279952093,
+      "loss": 0.9705,
+      "step": 7127
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.0001322726757214977,
+      "loss": 0.8055,
+      "step": 7128
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.000132256157660611,
+      "loss": 1.0209,
+      "step": 7129
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.00013223963861736386,
+      "loss": 0.7838,
+      "step": 7130
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.00013222311859225936,
+      "loss": 0.8739,
+      "step": 7131
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.00013220659758580064,
+      "loss": 0.9864,
+      "step": 7132
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.0001321900755984908,
+      "loss": 0.901,
+      "step": 7133
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.0001321735526308331,
+      "loss": 0.8954,
+      "step": 7134
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.0001321570286833307,
+      "loss": 0.7605,
+      "step": 7135
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.00013214050375648682,
+      "loss": 0.9507,
+      "step": 7136
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.00013212397785080472,
+      "loss": 0.8064,
+      "step": 7137
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.00013210745096678777,
+      "loss": 0.7112,
+      "step": 7138
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.0001320909231049392,
+      "loss": 1.0926,
+      "step": 7139
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.00013207439426576244,
+      "loss": 0.9628,
+      "step": 7140
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.0001320578644497608,
+      "loss": 0.9083,
+      "step": 7141
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.00013204133365743776,
+      "loss": 0.9804,
+      "step": 7142
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.00013202480188929675,
+      "loss": 0.9118,
+      "step": 7143
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.00013200826914584122,
+      "loss": 0.8473,
+      "step": 7144
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.00013199173542757472,
+      "loss": 0.8545,
+      "step": 7145
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.00013197520073500077,
+      "loss": 1.0412,
+      "step": 7146
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.0001319586650686229,
+      "loss": 0.8978,
+      "step": 7147
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.00013194212842894472,
+      "loss": 0.8468,
+      "step": 7148
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.00013192559081646988,
+      "loss": 0.8184,
+      "step": 7149
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.00013190905223170197,
+      "loss": 0.9216,
+      "step": 7150
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.00013189251267514473,
+      "loss": 1.0163,
+      "step": 7151
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.00013187597214730186,
+      "loss": 0.8834,
+      "step": 7152
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.00013185943064867708,
+      "loss": 0.9367,
+      "step": 7153
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.00013184288817977418,
+      "loss": 0.9629,
+      "step": 7154
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.00013182634474109695,
+      "loss": 0.8143,
+      "step": 7155
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.00013180980033314924,
+      "loss": 0.897,
+      "step": 7156
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.00013179325495643493,
+      "loss": 0.9634,
+      "step": 7157
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.00013177670861145782,
+      "loss": 0.8248,
+      "step": 7158
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.0001317601612987219,
+      "loss": 0.9306,
+      "step": 7159
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.0001317436130187311,
+      "loss": 0.7926,
+      "step": 7160
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.0001317270637719894,
+      "loss": 0.8756,
+      "step": 7161
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.00013171051355900082,
+      "loss": 0.9019,
+      "step": 7162
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.0001316939623802694,
+      "loss": 0.7903,
+      "step": 7163
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.00013167741023629913,
+      "loss": 0.9059,
+      "step": 7164
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.00013166085712759418,
+      "loss": 0.9866,
+      "step": 7165
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.00013164430305465867,
+      "loss": 0.926,
+      "step": 7166
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.00013162774801799676,
+      "loss": 0.9112,
+      "step": 7167
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.0001316111920181126,
+      "loss": 0.8871,
+      "step": 7168
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.0001315946350555104,
+      "loss": 0.9423,
+      "step": 7169
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.00013157807713069444,
+      "loss": 0.9127,
+      "step": 7170
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.00013156151824416894,
+      "loss": 0.848,
+      "step": 7171
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.00013154495839643823,
+      "loss": 0.8088,
+      "step": 7172
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.00013152839758800669,
+      "loss": 0.767,
+      "step": 7173
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.0001315118358193786,
+      "loss": 0.9875,
+      "step": 7174
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.00013149527309105837,
+      "loss": 0.9624,
+      "step": 7175
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.00013147870940355042,
+      "loss": 0.8885,
+      "step": 7176
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.0001314621447573592,
+      "loss": 0.8472,
+      "step": 7177
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.0001314455791529892,
+      "loss": 0.783,
+      "step": 7178
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.0001314290125909449,
+      "loss": 0.8765,
+      "step": 7179
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.00013141244507173088,
+      "loss": 1.0383,
+      "step": 7180
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.00013139587659585165,
+      "loss": 0.8417,
+      "step": 7181
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.00013137930716381182,
+      "loss": 1.0033,
+      "step": 7182
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.000131362736776116,
+      "loss": 0.8695,
+      "step": 7183
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.00013134616543326888,
+      "loss": 0.8184,
+      "step": 7184
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.0001313295931357751,
+      "loss": 0.9026,
+      "step": 7185
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.00013131301988413937,
+      "loss": 0.8131,
+      "step": 7186
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.00013129644567886646,
+      "loss": 0.9615,
+      "step": 7187
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.00013127987052046112,
+      "loss": 0.8823,
+      "step": 7188
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.00013126329440942813,
+      "loss": 0.8355,
+      "step": 7189
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.00013124671734627232,
+      "loss": 0.9043,
+      "step": 7190
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.0001312301393314986,
+      "loss": 0.9676,
+      "step": 7191
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.00013121356036561177,
+      "loss": 0.9269,
+      "step": 7192
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.0001311969804491168,
+      "loss": 0.8297,
+      "step": 7193
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.00013118039958251858,
+      "loss": 0.9619,
+      "step": 7194
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.0001311638177663221,
+      "loss": 0.9699,
+      "step": 7195
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.0001311472350010324,
+      "loss": 0.9389,
+      "step": 7196
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.0001311306512871545,
+      "loss": 0.9005,
+      "step": 7197
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.0001311140666251934,
+      "loss": 0.8559,
+      "step": 7198
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.0001310974810156542,
+      "loss": 0.9492,
+      "step": 7199
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.00013108089445904207,
+      "loss": 0.8865,
+      "step": 7200
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.00013106430695586206,
+      "loss": 0.9223,
+      "step": 7201
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.00013104771850661943,
+      "loss": 0.9202,
+      "step": 7202
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.0001310311291118194,
+      "loss": 0.923,
+      "step": 7203
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.00013101453877196705,
+      "loss": 0.9013,
+      "step": 7204
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.00013099794748756778,
+      "loss": 1.0935,
+      "step": 7205
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.00013098135525912684,
+      "loss": 0.8133,
+      "step": 7206
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.0001309647620871495,
+      "loss": 0.8308,
+      "step": 7207
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.00013094816797214122,
+      "loss": 0.8971,
+      "step": 7208
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.00013093157291460726,
+      "loss": 0.9393,
+      "step": 7209
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.00013091497691505304,
+      "loss": 1.0258,
+      "step": 7210
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.00013089837997398403,
+      "loss": 0.9057,
+      "step": 7211
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.00013088178209190565,
+      "loss": 0.8578,
+      "step": 7212
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.0001308651832693234,
+      "loss": 0.8662,
+      "step": 7213
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.00013084858350674287,
+      "loss": 1.0149,
+      "step": 7214
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.00013083198280466947,
+      "loss": 1.0057,
+      "step": 7215
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.00013081538116360884,
+      "loss": 0.9015,
+      "step": 7216
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.0001307987785840666,
+      "loss": 0.7913,
+      "step": 7217
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.00013078217506654838,
+      "loss": 0.9604,
+      "step": 7218
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.00013076557061155985,
+      "loss": 0.8672,
+      "step": 7219
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.00013074896521960664,
+      "loss": 0.9349,
+      "step": 7220
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.0001307323588911945,
+      "loss": 0.9561,
+      "step": 7221
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.00013071575162682921,
+      "loss": 0.9781,
+      "step": 7222
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.00013069914342701646,
+      "loss": 0.9761,
+      "step": 7223
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.00013068253429226215,
+      "loss": 0.8129,
+      "step": 7224
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.000130665924223072,
+      "loss": 0.9679,
+      "step": 7225
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.000130649313219952,
+      "loss": 0.8914,
+      "step": 7226
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.00013063270128340796,
+      "loss": 0.9214,
+      "step": 7227
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.00013061608841394577,
+      "loss": 0.8233,
+      "step": 7228
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.00013059947461207142,
+      "loss": 0.8897,
+      "step": 7229
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.0001305828598782909,
+      "loss": 0.9554,
+      "step": 7230
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.0001305662442131102,
+      "loss": 0.9824,
+      "step": 7231
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.00013054962761703526,
+      "loss": 0.9983,
+      "step": 7232
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.0001305330100905723,
+      "loss": 0.9519,
+      "step": 7233
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.00013051639163422725,
+      "loss": 0.9701,
+      "step": 7234
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.0001304997722485063,
+      "loss": 0.9668,
+      "step": 7235
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.00013048315193391562,
+      "loss": 0.9944,
+      "step": 7236
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.00013046653069096134,
+      "loss": 0.8442,
+      "step": 7237
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.00013044990852014967,
+      "loss": 0.9902,
+      "step": 7238
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.00013043328542198683,
+      "loss": 0.8626,
+      "step": 7239
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.00013041666139697905,
+      "loss": 0.8096,
+      "step": 7240
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.00013040003644563267,
+      "loss": 0.9117,
+      "step": 7241
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.000130383410568454,
+      "loss": 0.837,
+      "step": 7242
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.00013036678376594934,
+      "loss": 0.7817,
+      "step": 7243
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.00013035015603862508,
+      "loss": 0.9373,
+      "step": 7244
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.0001303335273869876,
+      "loss": 0.8969,
+      "step": 7245
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.00013031689781154338,
+      "loss": 0.9543,
+      "step": 7246
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.0001303002673127988,
+      "loss": 0.937,
+      "step": 7247
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.00013028363589126036,
+      "loss": 0.9365,
+      "step": 7248
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.0001302670035474346,
+      "loss": 0.9554,
+      "step": 7249
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.00013025037028182806,
+      "loss": 0.945,
+      "step": 7250
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.00013023373609494726,
+      "loss": 0.8874,
+      "step": 7251
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.00013021710098729884,
+      "loss": 0.9242,
+      "step": 7252
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.00013020046495938937,
+      "loss": 0.9352,
+      "step": 7253
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.00013018382801172556,
+      "loss": 0.8662,
+      "step": 7254
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.00013016719014481408,
+      "loss": 0.8994,
+      "step": 7255
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.00013015055135916158,
+      "loss": 0.8651,
+      "step": 7256
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.00013013391165527487,
+      "loss": 0.8412,
+      "step": 7257
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.00013011727103366062,
+      "loss": 0.8862,
+      "step": 7258
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.0001301006294948257,
+      "loss": 1.0114,
+      "step": 7259
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.0001300839870392769,
+      "loss": 0.9847,
+      "step": 7260
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.00013006734366752105,
+      "loss": 0.7893,
+      "step": 7261
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.00013005069938006505,
+      "loss": 0.942,
+      "step": 7262
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.0001300340541774158,
+      "loss": 0.7529,
+      "step": 7263
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.00013001740806008018,
+      "loss": 0.871,
+      "step": 7264
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.00013000076102856521,
+      "loss": 1.0948,
+      "step": 7265
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.00012998411308337787,
+      "loss": 0.8848,
+      "step": 7266
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.00012996746422502513,
+      "loss": 0.8309,
+      "step": 7267
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.00012995081445401404,
+      "loss": 0.8065,
+      "step": 7268
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.0001299341637708517,
+      "loss": 0.8963,
+      "step": 7269
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.0001299175121760452,
+      "loss": 0.8875,
+      "step": 7270
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.0001299008596701016,
+      "loss": 0.9044,
+      "step": 7271
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.00012988420625352818,
+      "loss": 0.8594,
+      "step": 7272
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.00012986755192683202,
+      "loss": 1.0988,
+      "step": 7273
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.0001298508966905203,
+      "loss": 0.9392,
+      "step": 7274
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.00012983424054510034,
+      "loss": 0.9543,
+      "step": 7275
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.00012981758349107938,
+      "loss": 0.9671,
+      "step": 7276
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.00012980092552896468,
+      "loss": 0.9495,
+      "step": 7277
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.00012978426665926358,
+      "loss": 0.8515,
+      "step": 7278
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.00012976760688248343,
+      "loss": 0.9719,
+      "step": 7279
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.00012975094619913156,
+      "loss": 0.9008,
+      "step": 7280
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.00012973428460971544,
+      "loss": 0.9114,
+      "step": 7281
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.00012971762211474243,
+      "loss": 0.8929,
+      "step": 7282
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.00012970095871472006,
+      "loss": 1.0465,
+      "step": 7283
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.00012968429441015576,
+      "loss": 1.0068,
+      "step": 7284
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.00012966762920155702,
+      "loss": 0.9229,
+      "step": 7285
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.00012965096308943143,
+      "loss": 0.9581,
+      "step": 7286
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.00012963429607428653,
+      "loss": 0.9632,
+      "step": 7287
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.00012961762815662993,
+      "loss": 1.0452,
+      "step": 7288
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.00012960095933696927,
+      "loss": 0.9974,
+      "step": 7289
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.00012958428961581215,
+      "loss": 0.8963,
+      "step": 7290
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.00012956761899366625,
+      "loss": 0.8574,
+      "step": 7291
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.00012955094747103932,
+      "loss": 0.9216,
+      "step": 7292
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.00012953427504843902,
+      "loss": 1.0305,
+      "step": 7293
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.00012951760172637318,
+      "loss": 0.7307,
+      "step": 7294
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.00012950092750534955,
+      "loss": 0.9347,
+      "step": 7295
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.00012948425238587595,
+      "loss": 0.9133,
+      "step": 7296
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.00012946757636846017,
+      "loss": 0.9113,
+      "step": 7297
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.0001294508994536102,
+      "loss": 0.8659,
+      "step": 7298
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.00012943422164183381,
+      "loss": 0.7666,
+      "step": 7299
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.000129417542933639,
+      "loss": 0.8108,
+      "step": 7300
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.00012940086332953368,
+      "loss": 0.9028,
+      "step": 7301
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.00012938418283002582,
+      "loss": 0.867,
+      "step": 7302
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.0001293675014356235,
+      "loss": 0.8505,
+      "step": 7303
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.00012935081914683462,
+      "loss": 1.1647,
+      "step": 7304
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.00012933413596416733,
+      "loss": 0.9371,
+      "step": 7305
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.00012931745188812975,
+      "loss": 0.9186,
+      "step": 7306
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.0001293007669192299,
+      "loss": 0.8462,
+      "step": 7307
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.00012928408105797597,
+      "loss": 0.9083,
+      "step": 7308
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.0001292673943048761,
+      "loss": 0.7932,
+      "step": 7309
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.00012925070666043852,
+      "loss": 0.8185,
+      "step": 7310
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.00012923401812517146,
+      "loss": 0.8638,
+      "step": 7311
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.00012921732869958314,
+      "loss": 0.8345,
+      "step": 7312
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.0001292006383841818,
+      "loss": 1.0842,
+      "step": 7313
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.00012918394717947583,
+      "loss": 0.8901,
+      "step": 7314
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.0001291672550859735,
+      "loss": 0.8262,
+      "step": 7315
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.00012915056210418318,
+      "loss": 0.9026,
+      "step": 7316
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.00012913386823461327,
+      "loss": 0.8892,
+      "step": 7317
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.00012911717347777218,
+      "loss": 0.8661,
+      "step": 7318
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.00012910047783416828,
+      "loss": 0.8697,
+      "step": 7319
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.00012908378130431012,
+      "loss": 0.8265,
+      "step": 7320
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.00012906708388870617,
+      "loss": 0.968,
+      "step": 7321
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.00012905038558786493,
+      "loss": 0.9142,
+      "step": 7322
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.000129033686402295,
+      "loss": 0.9173,
+      "step": 7323
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.00012901698633250485,
+      "loss": 0.8414,
+      "step": 7324
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.0001290002853790032,
+      "loss": 0.8489,
+      "step": 7325
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.00012898358354229857,
+      "loss": 0.931,
+      "step": 7326
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.0001289668808228997,
+      "loss": 0.9442,
+      "step": 7327
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.00012895017722131522,
+      "loss": 0.8189,
+      "step": 7328
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.00012893347273805386,
+      "loss": 0.8728,
+      "step": 7329
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.00012891676737362436,
+      "loss": 0.9822,
+      "step": 7330
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.0001289000611285354,
+      "loss": 0.9501,
+      "step": 7331
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.0001288833540032959,
+      "loss": 0.8539,
+      "step": 7332
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.00012886664599841461,
+      "loss": 0.7935,
+      "step": 7333
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.00012884993711440035,
+      "loss": 0.9115,
+      "step": 7334
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.00012883322735176203,
+      "loss": 0.9069,
+      "step": 7335
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.00012881651671100853,
+      "loss": 0.8926,
+      "step": 7336
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.00012879980519264875,
+      "loss": 0.8666,
+      "step": 7337
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.0001287830927971917,
+      "loss": 0.928,
+      "step": 7338
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.00012876637952514628,
+      "loss": 0.9169,
+      "step": 7339
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.00012874966537702152,
+      "loss": 0.9061,
+      "step": 7340
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.0001287329503533265,
+      "loss": 0.848,
+      "step": 7341
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.00012871623445457017,
+      "loss": 0.8065,
+      "step": 7342
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.0001286995176812617,
+      "loss": 0.9595,
+      "step": 7343
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.00012868280003391019,
+      "loss": 0.8522,
+      "step": 7344
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.00012866608151302473,
+      "loss": 0.9667,
+      "step": 7345
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.00012864936211911453,
+      "loss": 0.7579,
+      "step": 7346
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.00012863264185268875,
+      "loss": 0.8586,
+      "step": 7347
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.0001286159207142566,
+      "loss": 0.7962,
+      "step": 7348
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.00012859919870432737,
+      "loss": 1.0125,
+      "step": 7349
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.00012858247582341026,
+      "loss": 1.0061,
+      "step": 7350
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.0001285657520720146,
+      "loss": 1.0028,
+      "step": 7351
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.00012854902745064976,
+      "loss": 0.9273,
+      "step": 7352
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.000128532301959825,
+      "loss": 0.8335,
+      "step": 7353
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.00012851557560004972,
+      "loss": 0.9432,
+      "step": 7354
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.00012849884837183334,
+      "loss": 0.8247,
+      "step": 7355
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.0001284821202756853,
+      "loss": 0.7869,
+      "step": 7356
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.00012846539131211503,
+      "loss": 0.8943,
+      "step": 7357
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.000128448661481632,
+      "loss": 0.8665,
+      "step": 7358
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.00012843193078474572,
+      "loss": 0.9509,
+      "step": 7359
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.0001284151992219658,
+      "loss": 0.8873,
+      "step": 7360
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.00012839846679380166,
+      "loss": 1.0102,
+      "step": 7361
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.00012838173350076298,
+      "loss": 0.9564,
+      "step": 7362
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.00012836499934335937,
+      "loss": 0.8317,
+      "step": 7363
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.00012834826432210045,
+      "loss": 0.8523,
+      "step": 7364
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.00012833152843749586,
+      "loss": 0.9572,
+      "step": 7365
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.00012831479169005534,
+      "loss": 0.9855,
+      "step": 7366
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.00012829805408028853,
+      "loss": 0.8871,
+      "step": 7367
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.0001282813156087053,
+      "loss": 0.923,
+      "step": 7368
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.00012826457627581532,
+      "loss": 0.846,
+      "step": 7369
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.00012824783608212837,
+      "loss": 0.9792,
+      "step": 7370
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.00012823109502815434,
+      "loss": 0.8409,
+      "step": 7371
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.00012821435311440306,
+      "loss": 0.8945,
+      "step": 7372
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.00012819761034138443,
+      "loss": 1.0037,
+      "step": 7373
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.00012818086670960828,
+      "loss": 1.015,
+      "step": 7374
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.00012816412221958455,
+      "loss": 0.8497,
+      "step": 7375
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.00012814737687182328,
+      "loss": 0.9119,
+      "step": 7376
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.00012813063066683433,
+      "loss": 0.8484,
+      "step": 7377
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.00012811388360512774,
+      "loss": 0.9181,
+      "step": 7378
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.00012809713568721364,
+      "loss": 0.862,
+      "step": 7379
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.00012808038691360197,
+      "loss": 0.9057,
+      "step": 7380
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.00012806363728480285,
+      "loss": 0.839,
+      "step": 7381
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.0001280468868013264,
+      "loss": 0.8788,
+      "step": 7382
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.00012803013546368273,
+      "loss": 0.9485,
+      "step": 7383
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.00012801338327238205,
+      "loss": 0.7925,
+      "step": 7384
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.00012799663022793452,
+      "loss": 0.9469,
+      "step": 7385
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.00012797987633085032,
+      "loss": 0.9117,
+      "step": 7386
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.00012796312158163976,
+      "loss": 0.8964,
+      "step": 7387
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.00012794636598081302,
+      "loss": 0.8467,
+      "step": 7388
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.00012792960952888045,
+      "loss": 0.8752,
+      "step": 7389
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.00012791285222635239,
+      "loss": 0.9515,
+      "step": 7390
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.0001278960940737391,
+      "loss": 0.9536,
+      "step": 7391
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.00012787933507155103,
+      "loss": 0.9028,
+      "step": 7392
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.00012786257522029855,
+      "loss": 0.971,
+      "step": 7393
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.00012784581452049204,
+      "loss": 0.9795,
+      "step": 7394
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.000127829052972642,
+      "loss": 0.8818,
+      "step": 7395
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.00012781229057725888,
+      "loss": 0.9644,
+      "step": 7396
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.00012779552733485316,
+      "loss": 0.9521,
+      "step": 7397
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.00012777876324593542,
+      "loss": 0.8997,
+      "step": 7398
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.00012776199831101615,
+      "loss": 0.8446,
+      "step": 7399
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.00012774523253060595,
+      "loss": 0.9286,
+      "step": 7400
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.00012772846590521544,
+      "loss": 0.8119,
+      "step": 7401
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.0001277116984353552,
+      "loss": 0.8815,
+      "step": 7402
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.00012769493012153594,
+      "loss": 0.9352,
+      "step": 7403
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.00012767816096426833,
+      "loss": 0.91,
+      "step": 7404
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.000127661390964063,
+      "loss": 0.9579,
+      "step": 7405
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.00012764462012143077,
+      "loss": 0.97,
+      "step": 7406
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.00012762784843688233,
+      "loss": 0.8712,
+      "step": 7407
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.0001276110759109285,
+      "loss": 0.8849,
+      "step": 7408
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.0001275943025440801,
+      "loss": 0.9716,
+      "step": 7409
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.00012757752833684793,
+      "loss": 0.9981,
+      "step": 7410
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.00012756075328974284,
+      "loss": 0.8596,
+      "step": 7411
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.00012754397740327576,
+      "loss": 0.9143,
+      "step": 7412
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.00012752720067795754,
+      "loss": 0.7571,
+      "step": 7413
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.00012751042311429916,
+      "loss": 0.7999,
+      "step": 7414
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.0001274936447128116,
+      "loss": 0.8484,
+      "step": 7415
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.00012747686547400576,
+      "loss": 0.8744,
+      "step": 7416
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.00012746008539839273,
+      "loss": 1.0454,
+      "step": 7417
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.0001274433044864835,
+      "loss": 0.9376,
+      "step": 7418
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.00012742652273878917,
+      "loss": 1.0099,
+      "step": 7419
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.00012740974015582081,
+      "loss": 0.8811,
+      "step": 7420
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.00012739295673808954,
+      "loss": 0.8685,
+      "step": 7421
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.00012737617248610646,
+      "loss": 0.8033,
+      "step": 7422
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.0001273593874003828,
+      "loss": 0.9354,
+      "step": 7423
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.0001273426014814297,
+      "loss": 0.8195,
+      "step": 7424
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.0001273258147297584,
+      "loss": 0.9282,
+      "step": 7425
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.00012730902714588012,
+      "loss": 0.8958,
+      "step": 7426
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.00012729223873030613,
+      "loss": 0.8862,
+      "step": 7427
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.00012727544948354776,
+      "loss": 0.8093,
+      "step": 7428
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.00012725865940611624,
+      "loss": 0.9146,
+      "step": 7429
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.000127241868498523,
+      "loss": 0.8923,
+      "step": 7430
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.00012722507676127934,
+      "loss": 0.9797,
+      "step": 7431
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.00012720828419489667,
+      "loss": 1.0183,
+      "step": 7432
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.00012719149079988649,
+      "loss": 0.9338,
+      "step": 7433
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.00012717469657676014,
+      "loss": 0.773,
+      "step": 7434
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.0001271579015260291,
+      "loss": 0.9228,
+      "step": 7435
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.00012714110564820486,
+      "loss": 0.9791,
+      "step": 7436
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.00012712430894379898,
+      "loss": 0.95,
+      "step": 7437
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.00012710751141332301,
+      "loss": 0.8032,
+      "step": 7438
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.00012709071305728849,
+      "loss": 0.9429,
+      "step": 7439
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.00012707391387620698,
+      "loss": 0.9629,
+      "step": 7440
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.00012705711387059015,
+      "loss": 0.9015,
+      "step": 7441
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.00012704031304094964,
+      "loss": 0.9841,
+      "step": 7442
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.00012702351138779708,
+      "loss": 0.8118,
+      "step": 7443
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.00012700670891164424,
+      "loss": 0.8462,
+      "step": 7444
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.00012698990561300273,
+      "loss": 0.8971,
+      "step": 7445
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.00012697310149238438,
+      "loss": 0.9848,
+      "step": 7446
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.00012695629655030093,
+      "loss": 0.9293,
+      "step": 7447
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.00012693949078726417,
+      "loss": 0.8197,
+      "step": 7448
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.00012692268420378593,
+      "loss": 0.9821,
+      "step": 7449
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.00012690587680037806,
+      "loss": 0.8861,
+      "step": 7450
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.0001268890685775524,
+      "loss": 0.8289,
+      "step": 7451
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.00012687225953582088,
+      "loss": 1.0322,
+      "step": 7452
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.00012685544967569538,
+      "loss": 1.0348,
+      "step": 7453
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.00012683863899768787,
+      "loss": 0.8844,
+      "step": 7454
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.00012682182750231032,
+      "loss": 0.991,
+      "step": 7455
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.00012680501519007472,
+      "loss": 0.7389,
+      "step": 7456
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.0001267882020614931,
+      "loss": 0.9564,
+      "step": 7457
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.00012677138811707748,
+      "loss": 0.9382,
+      "step": 7458
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.0001267545733573399,
+      "loss": 1.0165,
+      "step": 7459
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.00012673775778279252,
+      "loss": 0.9159,
+      "step": 7460
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.00012672094139394745,
+      "loss": 1.0137,
+      "step": 7461
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.0001267041241913168,
+      "loss": 0.8569,
+      "step": 7462
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.00012668730617541274,
+      "loss": 0.9523,
+      "step": 7463
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.00012667048734674744,
+      "loss": 0.863,
+      "step": 7464
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.00012665366770583318,
+      "loss": 1.0647,
+      "step": 7465
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.00012663684725318218,
+      "loss": 0.976,
+      "step": 7466
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.00012662002598930669,
+      "loss": 1.0334,
+      "step": 7467
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.000126603203914719,
+      "loss": 1.0589,
+      "step": 7468
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.00012658638102993144,
+      "loss": 0.9069,
+      "step": 7469
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.0001265695573354563,
+      "loss": 0.9627,
+      "step": 7470
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.00012655273283180605,
+      "loss": 0.8269,
+      "step": 7471
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.00012653590751949298,
+      "loss": 0.997,
+      "step": 7472
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.00012651908139902957,
+      "loss": 0.9298,
+      "step": 7473
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.00012650225447092823,
+      "loss": 0.9126,
+      "step": 7474
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.0001264854267357014,
+      "loss": 0.887,
+      "step": 7475
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.00012646859819386163,
+      "loss": 0.8891,
+      "step": 7476
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.00012645176884592136,
+      "loss": 1.1118,
+      "step": 7477
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.00012643493869239316,
+      "loss": 0.9611,
+      "step": 7478
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.00012641810773378963,
+      "loss": 0.9542,
+      "step": 7479
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.00012640127597062333,
+      "loss": 0.9774,
+      "step": 7480
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.0001263844434034068,
+      "loss": 0.9629,
+      "step": 7481
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.00012636761003265282,
+      "loss": 0.8975,
+      "step": 7482
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.0001263507758588739,
+      "loss": 0.8086,
+      "step": 7483
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.00012633394088258285,
+      "loss": 0.85,
+      "step": 7484
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.0001263171051042923,
+      "loss": 0.7951,
+      "step": 7485
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.00012630026852451497,
+      "loss": 0.8785,
+      "step": 7486
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.00012628343114376369,
+      "loss": 0.7733,
+      "step": 7487
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.0001262665929625512,
+      "loss": 0.8687,
+      "step": 7488
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.00012624975398139028,
+      "loss": 0.9603,
+      "step": 7489
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.00012623291420079384,
+      "loss": 0.9351,
+      "step": 7490
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.00012621607362127464,
+      "loss": 0.895,
+      "step": 7491
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.0001261992322433456,
+      "loss": 0.8474,
+      "step": 7492
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.0001261823900675197,
+      "loss": 0.8549,
+      "step": 7493
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.00012616554709430974,
+      "loss": 1.0614,
+      "step": 7494
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.00012614870332422877,
+      "loss": 0.8714,
+      "step": 7495
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.0001261318587577897,
+      "loss": 0.864,
+      "step": 7496
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.00012611501339550557,
+      "loss": 0.9406,
+      "step": 7497
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.00012609816723788936,
+      "loss": 1.037,
+      "step": 7498
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.0001260813202854542,
+      "loss": 0.7845,
+      "step": 7499
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.00012606447253871305,
+      "loss": 0.8068,
+      "step": 7500
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.00012604762399817914,
+      "loss": 0.8548,
+      "step": 7501
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.0001260307746643655,
+      "loss": 0.9408,
+      "step": 7502
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.0001260139245377853,
+      "loss": 0.9623,
+      "step": 7503
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.0001259970736189517,
+      "loss": 0.9125,
+      "step": 7504
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.00012598022190837791,
+      "loss": 0.8709,
+      "step": 7505
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.00012596336940657715,
+      "loss": 0.869,
+      "step": 7506
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.00012594651611406266,
+      "loss": 0.8143,
+      "step": 7507
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.0001259296620313477,
+      "loss": 0.9097,
+      "step": 7508
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.00012591280715894555,
+      "loss": 0.8917,
+      "step": 7509
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.00012589595149736956,
+      "loss": 0.9208,
+      "step": 7510
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.00012587909504713303,
+      "loss": 0.8951,
+      "step": 7511
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.00012586223780874933,
+      "loss": 0.9982,
+      "step": 7512
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.0001258453797827319,
+      "loss": 0.9757,
+      "step": 7513
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.00012582852096959405,
+      "loss": 1.0466,
+      "step": 7514
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.00012581166136984933,
+      "loss": 1.0174,
+      "step": 7515
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.0001257948009840111,
+      "loss": 0.8437,
+      "step": 7516
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.0001257779398125929,
+      "loss": 0.9806,
+      "step": 7517
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.0001257610778561082,
+      "loss": 0.9481,
+      "step": 7518
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.00012574421511507054,
+      "loss": 0.9523,
+      "step": 7519
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.00012572735158999354,
+      "loss": 0.9509,
+      "step": 7520
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.00012571048728139068,
+      "loss": 0.7697,
+      "step": 7521
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.0001256936221897756,
+      "loss": 0.908,
+      "step": 7522
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.00012567675631566195,
+      "loss": 0.9215,
+      "step": 7523
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.00012565988965956336,
+      "loss": 0.8953,
+      "step": 7524
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.00012564302222199352,
+      "loss": 0.8085,
+      "step": 7525
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.0001256261540034661,
+      "loss": 0.8792,
+      "step": 7526
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.0001256092850044948,
+      "loss": 0.8512,
+      "step": 7527
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.00012559241522559344,
+      "loss": 0.9748,
+      "step": 7528
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.0001255755446672757,
+      "loss": 0.9555,
+      "step": 7529
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.00012555867333005546,
+      "loss": 0.8764,
+      "step": 7530
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.00012554180121444647,
+      "loss": 0.9103,
+      "step": 7531
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.00012552492832096262,
+      "loss": 1.0687,
+      "step": 7532
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.00012550805465011772,
+      "loss": 0.9594,
+      "step": 7533
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.0001254911802024257,
+      "loss": 0.9933,
+      "step": 7534
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.00012547430497840046,
+      "loss": 0.9998,
+      "step": 7535
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.00012545742897855595,
+      "loss": 0.9255,
+      "step": 7536
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.00012544055220340607,
+      "loss": 0.9754,
+      "step": 7537
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.00012542367465346482,
+      "loss": 0.9123,
+      "step": 7538
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.00012540679632924628,
+      "loss": 0.8755,
+      "step": 7539
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.00012538991723126439,
+      "loss": 1.0003,
+      "step": 7540
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.00012537303736003325,
+      "loss": 1.06,
+      "step": 7541
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.00012535615671606696,
+      "loss": 0.8295,
+      "step": 7542
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.00012533927529987952,
+      "loss": 0.9683,
+      "step": 7543
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.00012532239311198516,
+      "loss": 0.9164,
+      "step": 7544
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.00012530551015289793,
+      "loss": 0.9535,
+      "step": 7545
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.00012528862642313207,
+      "loss": 0.9219,
+      "step": 7546
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.0001252717419232018,
+      "loss": 0.879,
+      "step": 7547
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.00012525485665362127,
+      "loss": 0.957,
+      "step": 7548
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.00012523797061490471,
+      "loss": 0.9964,
+      "step": 7549
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.00012522108380756648,
+      "loss": 0.9273,
+      "step": 7550
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.00012520419623212077,
+      "loss": 0.9118,
+      "step": 7551
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.00012518730788908193,
+      "loss": 0.8719,
+      "step": 7552
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.00012517041877896427,
+      "loss": 0.8847,
+      "step": 7553
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.00012515352890228219,
+      "loss": 0.9423,
+      "step": 7554
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.00012513663825955003,
+      "loss": 0.9074,
+      "step": 7555
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.0001251197468512822,
+      "loss": 0.8689,
+      "step": 7556
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.00012510285467799318,
+      "loss": 1.0297,
+      "step": 7557
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.00012508596174019737,
+      "loss": 1.0518,
+      "step": 7558
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.00012506906803840925,
+      "loss": 1.009,
+      "step": 7559
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.0001250521735731433,
+      "loss": 0.9218,
+      "step": 7560
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.00012503527834491408,
+      "loss": 0.8956,
+      "step": 7561
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.0001250183823542361,
+      "loss": 0.8591,
+      "step": 7562
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.00012500148560162398,
+      "loss": 0.8863,
+      "step": 7563
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.00012498458808759224,
+      "loss": 0.942,
+      "step": 7564
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.00012496768981265554,
+      "loss": 0.9901,
+      "step": 7565
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.0001249507907773285,
+      "loss": 0.8092,
+      "step": 7566
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.00012493389098212576,
+      "loss": 0.8019,
+      "step": 7567
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.00012491699042756202,
+      "loss": 0.8952,
+      "step": 7568
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.00012490008911415204,
+      "loss": 0.8921,
+      "step": 7569
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.00012488318704241045,
+      "loss": 0.86,
+      "step": 7570
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.00012486628421285207,
+      "loss": 0.9299,
+      "step": 7571
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.00012484938062599167,
+      "loss": 0.776,
+      "step": 7572
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.00012483247628234404,
+      "loss": 0.9763,
+      "step": 7573
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.00012481557118242397,
+      "loss": 1.0415,
+      "step": 7574
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.00012479866532674633,
+      "loss": 0.9258,
+      "step": 7575
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.00012478175871582598,
+      "loss": 0.8607,
+      "step": 7576
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.00012476485135017788,
+      "loss": 0.8532,
+      "step": 7577
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.00012474794323031683,
+      "loss": 0.9039,
+      "step": 7578
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.0001247310343567578,
+      "loss": 1.0087,
+      "step": 7579
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.0001247141247300158,
+      "loss": 0.9894,
+      "step": 7580
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.00012469721435060576,
+      "loss": 0.8963,
+      "step": 7581
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.00012468030321904276,
+      "loss": 0.9489,
+      "step": 7582
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.00012466339133584172,
+      "loss": 0.973,
+      "step": 7583
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.00012464647870151776,
+      "loss": 0.8687,
+      "step": 7584
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.00012462956531658593,
+      "loss": 0.9055,
+      "step": 7585
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.00012461265118156133,
+      "loss": 0.9114,
+      "step": 7586
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.00012459573629695914,
+      "loss": 0.89,
+      "step": 7587
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.00012457882066329437,
+      "loss": 0.9437,
+      "step": 7588
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.0001245619042810823,
+      "loss": 0.8808,
+      "step": 7589
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.0001245449871508381,
+      "loss": 0.8894,
+      "step": 7590
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.0001245280692730769,
+      "loss": 0.9311,
+      "step": 7591
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.00012451115064831403,
+      "loss": 0.9599,
+      "step": 7592
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.00012449423127706472,
+      "loss": 0.8888,
+      "step": 7593
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.00012447731115984424,
+      "loss": 0.9823,
+      "step": 7594
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.00012446039029716786,
+      "loss": 0.8461,
+      "step": 7595
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.00012444346868955097,
+      "loss": 0.9026,
+      "step": 7596
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.00012442654633750886,
+      "loss": 0.8777,
+      "step": 7597
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.00012440962324155694,
+      "loss": 0.9111,
+      "step": 7598
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.00012439269940221057,
+      "loss": 0.8595,
+      "step": 7599
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.0001243757748199852,
+      "loss": 1.04,
+      "step": 7600
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.00012435884949539623,
+      "loss": 0.8856,
+      "step": 7601
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.00012434192342895912,
+      "loss": 0.8616,
+      "step": 7602
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.0001243249966211894,
+      "loss": 0.9264,
+      "step": 7603
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.00012430806907260254,
+      "loss": 0.7961,
+      "step": 7604
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.0001242911407837141,
+      "loss": 0.8479,
+      "step": 7605
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.00012427421175503954,
+      "loss": 1.0075,
+      "step": 7606
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.00012425728198709455,
+      "loss": 0.917,
+      "step": 7607
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.00012424035148039466,
+      "loss": 0.8361,
+      "step": 7608
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.0001242234202354555,
+      "loss": 0.836,
+      "step": 7609
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.00012420648825279268,
+      "loss": 0.9857,
+      "step": 7610
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.00012418955553292196,
+      "loss": 0.9523,
+      "step": 7611
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.00012417262207635893,
+      "loss": 0.8423,
+      "step": 7612
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.00012415568788361932,
+      "loss": 0.8555,
+      "step": 7613
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.00012413875295521885,
+      "loss": 0.8589,
+      "step": 7614
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.00012412181729167337,
+      "loss": 0.8756,
+      "step": 7615
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.0001241048808934985,
+      "loss": 0.9094,
+      "step": 7616
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.00012408794376121015,
+      "loss": 0.9141,
+      "step": 7617
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.0001240710058953241,
+      "loss": 0.9072,
+      "step": 7618
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.0001240540672963562,
+      "loss": 1.1013,
+      "step": 7619
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.00012403712796482232,
+      "loss": 0.9993,
+      "step": 7620
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.0001240201879012383,
+      "loss": 0.8152,
+      "step": 7621
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.0001240032471061201,
+      "loss": 0.7926,
+      "step": 7622
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.0001239863055799837,
+      "loss": 0.932,
+      "step": 7623
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.0001239693633233449,
+      "loss": 0.8301,
+      "step": 7624
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.00012395242033671984,
+      "loss": 0.9736,
+      "step": 7625
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.00012393547662062444,
+      "loss": 0.9555,
+      "step": 7626
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.0001239185321755747,
+      "loss": 0.8974,
+      "step": 7627
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.0001239015870020867,
+      "loss": 1.0114,
+      "step": 7628
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.0001238846411006765,
+      "loss": 0.8626,
+      "step": 7629
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.00012386769447186017,
+      "loss": 0.9045,
+      "step": 7630
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.00012385074711615386,
+      "loss": 0.8268,
+      "step": 7631
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.00012383379903407364,
+      "loss": 0.903,
+      "step": 7632
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.0001238168502261357,
+      "loss": 0.9246,
+      "step": 7633
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.00012379990069285625,
+      "loss": 0.8819,
+      "step": 7634
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.0001237829504347514,
+      "loss": 0.899,
+      "step": 7635
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.00012376599945233745,
+      "loss": 0.934,
+      "step": 7636
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.00012374904774613057,
+      "loss": 0.9134,
+      "step": 7637
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.00012373209531664709,
+      "loss": 0.9114,
+      "step": 7638
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.0001237151421644033,
+      "loss": 0.8711,
+      "step": 7639
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.00012369818828991545,
+      "loss": 0.9466,
+      "step": 7640
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.0001236812336936999,
+      "loss": 0.9416,
+      "step": 7641
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.000123664278376273,
+      "loss": 0.9713,
+      "step": 7642
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.0001236473223381511,
+      "loss": 0.8398,
+      "step": 7643
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.00012363036557985066,
+      "loss": 0.7204,
+      "step": 7644
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.00012361340810188801,
+      "loss": 0.9165,
+      "step": 7645
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.00012359644990477968,
+      "loss": 0.8732,
+      "step": 7646
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.00012357949098904206,
+      "loss": 0.8546,
+      "step": 7647
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.00012356253135519167,
+      "loss": 0.8863,
+      "step": 7648
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.000123545571003745,
+      "loss": 0.9901,
+      "step": 7649
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.0001235286099352186,
+      "loss": 0.8419,
+      "step": 7650
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.00012351164815012898,
+      "loss": 0.7924,
+      "step": 7651
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.00012349468564899275,
+      "loss": 0.8886,
+      "step": 7652
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.00012347772243232648,
+      "loss": 1.0088,
+      "step": 7653
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.00012346075850064677,
+      "loss": 0.88,
+      "step": 7654
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.00012344379385447032,
+      "loss": 0.836,
+      "step": 7655
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.0001234268284943137,
+      "loss": 0.8537,
+      "step": 7656
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.00012340986242069367,
+      "loss": 0.8429,
+      "step": 7657
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.0001233928956341269,
+      "loss": 0.9304,
+      "step": 7658
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.00012337592813513006,
+      "loss": 0.9709,
+      "step": 7659
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.00012335895992421995,
+      "loss": 0.9607,
+      "step": 7660
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.00012334199100191335,
+      "loss": 0.9722,
+      "step": 7661
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.00012332502136872702,
+      "loss": 0.9047,
+      "step": 7662
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.00012330805102517777,
+      "loss": 0.7747,
+      "step": 7663
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.00012329107997178246,
+      "loss": 0.9344,
+      "step": 7664
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.00012327410820905788,
+      "loss": 0.8418,
+      "step": 7665
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.000123257135737521,
+      "loss": 0.9606,
+      "step": 7666
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.00012324016255768863,
+      "loss": 0.7619,
+      "step": 7667
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.00012322318867007768,
+      "loss": 0.8359,
+      "step": 7668
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.0001232062140752052,
+      "loss": 0.8292,
+      "step": 7669
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.00012318923877358806,
+      "loss": 0.8993,
+      "step": 7670
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.00012317226276574322,
+      "loss": 0.9048,
+      "step": 7671
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.00012315528605218778,
+      "loss": 0.8731,
+      "step": 7672
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.00012313830863343867,
+      "loss": 0.9178,
+      "step": 7673
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.000123121330510013,
+      "loss": 0.9076,
+      "step": 7674
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.00012310435168242783,
+      "loss": 0.9531,
+      "step": 7675
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.0001230873721512002,
+      "loss": 0.9018,
+      "step": 7676
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.00012307039191684728,
+      "loss": 0.7959,
+      "step": 7677
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.00012305341097988617,
+      "loss": 0.8643,
+      "step": 7678
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.00012303642934083405,
+      "loss": 0.8874,
+      "step": 7679
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.0001230194470002081,
+      "loss": 0.9031,
+      "step": 7680
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.00012300246395852545,
+      "loss": 0.7917,
+      "step": 7681
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.00012298548021630334,
+      "loss": 0.9663,
+      "step": 7682
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.0001229684957740591,
+      "loss": 1.0311,
+      "step": 7683
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.00012295151063230987,
+      "loss": 0.8273,
+      "step": 7684
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.00012293452479157302,
+      "loss": 0.8907,
+      "step": 7685
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.00012291753825236583,
+      "loss": 0.8745,
+      "step": 7686
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.00012290055101520556,
+      "loss": 0.8948,
+      "step": 7687
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.00012288356308060965,
+      "loss": 1.0527,
+      "step": 7688
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.00012286657444909543,
+      "loss": 0.8714,
+      "step": 7689
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.00012284958512118028,
+      "loss": 0.9101,
+      "step": 7690
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.0001228325950973816,
+      "loss": 0.9575,
+      "step": 7691
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.00012281560437821688,
+      "loss": 0.9004,
+      "step": 7692
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.00012279861296420348,
+      "loss": 0.9092,
+      "step": 7693
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.00012278162085585891,
+      "loss": 0.9561,
+      "step": 7694
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.0001227646280537007,
+      "loss": 0.8875,
+      "step": 7695
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.00012274763455824634,
+      "loss": 0.9185,
+      "step": 7696
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.00012273064037001338,
+      "loss": 0.8779,
+      "step": 7697
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.00012271364548951933,
+      "loss": 0.8853,
+      "step": 7698
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.00012269664991728185,
+      "loss": 0.8901,
+      "step": 7699
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.00012267965365381843,
+      "loss": 0.9648,
+      "step": 7700
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.00012266265669964684,
+      "loss": 0.9815,
+      "step": 7701
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.00012264565905528455,
+      "loss": 0.9744,
+      "step": 7702
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.00012262866072124934,
+      "loss": 0.9816,
+      "step": 7703
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.00012261166169805888,
+      "loss": 0.8893,
+      "step": 7704
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.00012259466198623085,
+      "loss": 0.8267,
+      "step": 7705
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.00012257766158628295,
+      "loss": 0.8388,
+      "step": 7706
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.00012256066049873298,
+      "loss": 0.985,
+      "step": 7707
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.00012254365872409867,
+      "loss": 0.9435,
+      "step": 7708
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.00012252665626289786,
+      "loss": 0.8262,
+      "step": 7709
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.00012250965311564832,
+      "loss": 0.9426,
+      "step": 7710
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.00012249264928286785,
+      "loss": 0.996,
+      "step": 7711
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.00012247564476507438,
+      "loss": 0.9412,
+      "step": 7712
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.00012245863956278572,
+      "loss": 1.0535,
+      "step": 7713
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.00012244163367651975,
+      "loss": 0.9195,
+      "step": 7714
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.00012242462710679446,
+      "loss": 0.9342,
+      "step": 7715
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.00012240761985412774,
+      "loss": 0.7886,
+      "step": 7716
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.00012239061191903752,
+      "loss": 0.9387,
+      "step": 7717
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.00012237360330204182,
+      "loss": 1.005,
+      "step": 7718
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.0001223565940036586,
+      "loss": 0.9309,
+      "step": 7719
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.0001223395840244059,
+      "loss": 0.9017,
+      "step": 7720
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.00012232257336480178,
+      "loss": 0.8493,
+      "step": 7721
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.00012230556202536425,
+      "loss": 1.0115,
+      "step": 7722
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.0001222885500066114,
+      "loss": 0.9361,
+      "step": 7723
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.0001222715373090614,
+      "loss": 0.9893,
+      "step": 7724
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.00012225452393323225,
+      "loss": 0.8512,
+      "step": 7725
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.00012223750987964223,
+      "loss": 0.9442,
+      "step": 7726
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.00012222049514880936,
+      "loss": 0.8585,
+      "step": 7727
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.00012220347974125191,
+      "loss": 0.9188,
+      "step": 7728
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.0001221864636574881,
+      "loss": 0.8787,
+      "step": 7729
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.0001221694468980361,
+      "loss": 0.929,
+      "step": 7730
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.00012215242946341418,
+      "loss": 1.0158,
+      "step": 7731
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.0001221354113541406,
+      "loss": 0.9367,
+      "step": 7732
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.00012211839257073362,
+      "loss": 0.8082,
+      "step": 7733
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.00012210137311371161,
+      "loss": 0.8136,
+      "step": 7734
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.00012208435298359285,
+      "loss": 0.8561,
+      "step": 7735
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.00012206733218089573,
+      "loss": 0.9544,
+      "step": 7736
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.00012205031070613855,
+      "loss": 1.0157,
+      "step": 7737
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.00012203328855983975,
+      "loss": 0.973,
+      "step": 7738
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.00012201626574251772,
+      "loss": 0.8978,
+      "step": 7739
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.0001219992422546909,
+      "loss": 1.0237,
+      "step": 7740
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.00012198221809687771,
+      "loss": 1.0059,
+      "step": 7741
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.00012196519326959667,
+      "loss": 0.9571,
+      "step": 7742
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.00012194816777336626,
+      "loss": 0.8276,
+      "step": 7743
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.00012193114160870495,
+      "loss": 0.9532,
+      "step": 7744
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.0001219141147761313,
+      "loss": 1.002,
+      "step": 7745
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.00012189708727616386,
+      "loss": 1.0682,
+      "step": 7746
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.00012188005910932123,
+      "loss": 0.7901,
+      "step": 7747
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.00012186303027612194,
+      "loss": 0.8362,
+      "step": 7748
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.00012184600077708465,
+      "loss": 0.899,
+      "step": 7749
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.000121828970612728,
+      "loss": 0.7721,
+      "step": 7750
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.00012181193978357059,
+      "loss": 0.9912,
+      "step": 7751
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.00012179490829013112,
+      "loss": 0.9139,
+      "step": 7752
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.00012177787613292833,
+      "loss": 0.8027,
+      "step": 7753
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.00012176084331248085,
+      "loss": 0.8884,
+      "step": 7754
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.00012174380982930748,
+      "loss": 0.9771,
+      "step": 7755
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.00012172677568392697,
+      "loss": 0.8717,
+      "step": 7756
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.00012170974087685804,
+      "loss": 0.835,
+      "step": 7757
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.00012169270540861956,
+      "loss": 0.9313,
+      "step": 7758
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.00012167566927973029,
+      "loss": 0.9632,
+      "step": 7759
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.00012165863249070905,
+      "loss": 0.8751,
+      "step": 7760
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.0001216415950420748,
+      "loss": 0.8218,
+      "step": 7761
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.00012162455693434627,
+      "loss": 0.9825,
+      "step": 7762
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.00012160751816804244,
+      "loss": 0.924,
+      "step": 7763
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.00012159047874368224,
+      "loss": 0.9639,
+      "step": 7764
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.00012157343866178455,
+      "loss": 0.9292,
+      "step": 7765
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.00012155639792286835,
+      "loss": 0.8658,
+      "step": 7766
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.00012153935652745265,
+      "loss": 0.9437,
+      "step": 7767
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.00012152231447605638,
+      "loss": 0.9252,
+      "step": 7768
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.00012150527176919859,
+      "loss": 0.8666,
+      "step": 7769
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.00012148822840739833,
+      "loss": 0.9142,
+      "step": 7770
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.0001214711843911746,
+      "loss": 0.9462,
+      "step": 7771
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.00012145413972104657,
+      "loss": 0.8362,
+      "step": 7772
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.00012143709439753323,
+      "loss": 0.9112,
+      "step": 7773
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.00012142004842115375,
+      "loss": 0.9723,
+      "step": 7774
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.0001214030017924273,
+      "loss": 0.8871,
+      "step": 7775
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.00012138595451187293,
+      "loss": 0.9906,
+      "step": 7776
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.00012136890658000992,
+      "loss": 0.7857,
+      "step": 7777
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.00012135185799735742,
+      "loss": 0.7551,
+      "step": 7778
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.00012133480876443464,
+      "loss": 0.7403,
+      "step": 7779
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.00012131775888176082,
+      "loss": 0.8686,
+      "step": 7780
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.00012130070834985518,
+      "loss": 0.9612,
+      "step": 7781
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.00012128365716923704,
+      "loss": 0.926,
+      "step": 7782
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.00012126660534042571,
+      "loss": 0.8828,
+      "step": 7783
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.00012124955286394046,
+      "loss": 0.9661,
+      "step": 7784
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.00012123249974030062,
+      "loss": 0.8452,
+      "step": 7785
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.00012121544597002558,
+      "loss": 0.7641,
+      "step": 7786
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.00012119839155363466,
+      "loss": 1.0208,
+      "step": 7787
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.00012118133649164731,
+      "loss": 0.9956,
+      "step": 7788
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.0001211642807845829,
+      "loss": 0.9391,
+      "step": 7789
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.00012114722443296087,
+      "loss": 0.8641,
+      "step": 7790
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.0001211301674373007,
+      "loss": 0.9355,
+      "step": 7791
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.0001211131097981218,
+      "loss": 0.8553,
+      "step": 7792
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.00012109605151594373,
+      "loss": 0.742,
+      "step": 7793
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.00012107899259128592,
+      "loss": 0.8651,
+      "step": 7794
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.000121061933024668,
+      "loss": 0.993,
+      "step": 7795
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.00012104487281660944,
+      "loss": 0.9465,
+      "step": 7796
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.0001210278119676298,
+      "loss": 0.9743,
+      "step": 7797
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.0001210107504782487,
+      "loss": 0.977,
+      "step": 7798
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.00012099368834898579,
+      "loss": 0.9089,
+      "step": 7799
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.00012097662558036063,
+      "loss": 0.8754,
+      "step": 7800
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.00012095956217289288,
+      "loss": 0.827,
+      "step": 7801
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.00012094249812710224,
+      "loss": 0.7913,
+      "step": 7802
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.00012092543344350832,
+      "loss": 0.9703,
+      "step": 7803
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.00012090836812263093,
+      "loss": 0.9356,
+      "step": 7804
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.00012089130216498969,
+      "loss": 0.8952,
+      "step": 7805
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.00012087423557110442,
+      "loss": 0.8838,
+      "step": 7806
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.00012085716834149488,
+      "loss": 0.8661,
+      "step": 7807
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.00012084010047668077,
+      "loss": 0.9651,
+      "step": 7808
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.00012082303197718196,
+      "loss": 0.9232,
+      "step": 7809
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.00012080596284351829,
+      "loss": 0.9426,
+      "step": 7810
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.00012078889307620953,
+      "loss": 0.9187,
+      "step": 7811
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.0001207718226757756,
+      "loss": 0.9149,
+      "step": 7812
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.00012075475164273635,
+      "loss": 0.934,
+      "step": 7813
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.00012073767997761167,
+      "loss": 0.8939,
+      "step": 7814
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.0001207206076809215,
+      "loss": 0.802,
+      "step": 7815
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.00012070353475318577,
+      "loss": 0.9881,
+      "step": 7816
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.00012068646119492442,
+      "loss": 0.868,
+      "step": 7817
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.00012066938700665748,
+      "loss": 0.8627,
+      "step": 7818
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.00012065231218890485,
+      "loss": 0.9473,
+      "step": 7819
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.00012063523674218662,
+      "loss": 0.9064,
+      "step": 7820
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.00012061816066702278,
+      "loss": 0.9751,
+      "step": 7821
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.00012060108396393341,
+      "loss": 0.8815,
+      "step": 7822
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.00012058400663343859,
+      "loss": 0.8855,
+      "step": 7823
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.00012056692867605836,
+      "loss": 0.8203,
+      "step": 7824
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.00012054985009231285,
+      "loss": 0.8655,
+      "step": 7825
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.00012053277088272222,
+      "loss": 0.9142,
+      "step": 7826
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.00012051569104780655,
+      "loss": 0.8625,
+      "step": 7827
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.00012049861058808607,
+      "loss": 0.9882,
+      "step": 7828
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.00012048152950408096,
+      "loss": 0.9807,
+      "step": 7829
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.0001204644477963114,
+      "loss": 1.0,
+      "step": 7830
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.00012044736546529763,
+      "loss": 0.851,
+      "step": 7831
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.00012043028251155985,
+      "loss": 0.8755,
+      "step": 7832
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.00012041319893561837,
+      "loss": 0.9232,
+      "step": 7833
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.00012039611473799346,
+      "loss": 0.9723,
+      "step": 7834
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.0001203790299192054,
+      "loss": 0.9979,
+      "step": 7835
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.00012036194447977451,
+      "loss": 0.874,
+      "step": 7836
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.00012034485842022116,
+      "loss": 0.9564,
+      "step": 7837
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.00012032777174106567,
+      "loss": 0.9298,
+      "step": 7838
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.0001203106844428284,
+      "loss": 0.8812,
+      "step": 7839
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.00012029359652602981,
+      "loss": 0.7855,
+      "step": 7840
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.00012027650799119026,
+      "loss": 0.8237,
+      "step": 7841
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.00012025941883883018,
+      "loss": 0.8734,
+      "step": 7842
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.00012024232906947002,
+      "loss": 0.977,
+      "step": 7843
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.00012022523868363027,
+      "loss": 0.9378,
+      "step": 7844
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.00012020814768183142,
+      "loss": 1.0281,
+      "step": 7845
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.00012019105606459393,
+      "loss": 0.8761,
+      "step": 7846
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.0001201739638324384,
+      "loss": 0.9492,
+      "step": 7847
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.00012015687098588531,
+      "loss": 0.9665,
+      "step": 7848
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.00012013977752545524,
+      "loss": 0.8623,
+      "step": 7849
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.0001201226834516688,
+      "loss": 0.9036,
+      "step": 7850
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.00012010558876504651,
+      "loss": 0.8834,
+      "step": 7851
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.00012008849346610906,
+      "loss": 0.8137,
+      "step": 7852
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.00012007139755537711,
+      "loss": 0.9743,
+      "step": 7853
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.00012005430103337123,
+      "loss": 0.8941,
+      "step": 7854
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.00012003720390061215,
+      "loss": 0.7496,
+      "step": 7855
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.00012002010615762058,
+      "loss": 0.854,
+      "step": 7856
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.00012000300780491715,
+      "loss": 0.8671,
+      "step": 7857
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.00011998590884302267,
+      "loss": 0.8013,
+      "step": 7858
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.00011996880927245789,
+      "loss": 0.8878,
+      "step": 7859
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.00011995170909374351,
+      "loss": 0.9475,
+      "step": 7860
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.00011993460830740038,
+      "loss": 0.7592,
+      "step": 7861
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.00011991750691394922,
+      "loss": 0.9493,
+      "step": 7862
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.00011990040491391094,
+      "loss": 0.9494,
+      "step": 7863
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.0001198833023078064,
+      "loss": 0.826,
+      "step": 7864
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.00011986619909615636,
+      "loss": 0.8604,
+      "step": 7865
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.00011984909527948174,
+      "loss": 0.889,
+      "step": 7866
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.00011983199085830347,
+      "loss": 0.9865,
+      "step": 7867
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.00011981488583314242,
+      "loss": 0.9902,
+      "step": 7868
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.00011979778020451957,
+      "loss": 0.9169,
+      "step": 7869
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.00011978067397295585,
+      "loss": 0.9182,
+      "step": 7870
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.00011976356713897216,
+      "loss": 0.9508,
+      "step": 7871
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.00011974645970308962,
+      "loss": 0.9001,
+      "step": 7872
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.00011972935166582913,
+      "loss": 0.8622,
+      "step": 7873
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.00011971224302771173,
+      "loss": 0.9768,
+      "step": 7874
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.00011969513378925855,
+      "loss": 1.0483,
+      "step": 7875
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.00011967802395099057,
+      "loss": 1.0049,
+      "step": 7876
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.00011966091351342885,
+      "loss": 0.986,
+      "step": 7877
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.00011964380247709455,
+      "loss": 0.9338,
+      "step": 7878
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.00011962669084250875,
+      "loss": 0.7623,
+      "step": 7879
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.00011960957861019262,
+      "loss": 0.9244,
+      "step": 7880
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.00011959246578066727,
+      "loss": 0.9031,
+      "step": 7881
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.0001195753523544539,
+      "loss": 1.019,
+      "step": 7882
+    },
+    {
+      "epoch": 1.73,
+      "eval_loss": 0.5742496252059937,
+      "eval_runtime": 321.6444,
+      "eval_samples_per_second": 56.395,
+      "eval_steps_per_second": 28.199,
+      "step": 7882
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.00011955823833207366,
+      "loss": 0.7924,
+      "step": 7883
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.0001195411237140478,
+      "loss": 0.8691,
+      "step": 7884
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.00011952400850089754,
+      "loss": 0.8417,
+      "step": 7885
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.0001195068926931441,
+      "loss": 0.9065,
+      "step": 7886
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.00011948977629130878,
+      "loss": 1.0113,
+      "step": 7887
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.00011947265929591281,
+      "loss": 0.9099,
+      "step": 7888
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.00011945554170747753,
+      "loss": 0.9321,
+      "step": 7889
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.00011943842352652421,
+      "loss": 0.8157,
+      "step": 7890
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.00011942130475357424,
+      "loss": 0.9494,
+      "step": 7891
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.00011940418538914892,
+      "loss": 0.9395,
+      "step": 7892
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.00011938706543376968,
+      "loss": 1.0193,
+      "step": 7893
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.00011936994488795785,
+      "loss": 0.9066,
+      "step": 7894
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.00011935282375223485,
+      "loss": 1.0257,
+      "step": 7895
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.00011933570202712213,
+      "loss": 0.9211,
+      "step": 7896
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.00011931857971314107,
+      "loss": 0.9031,
+      "step": 7897
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.00011930145681081319,
+      "loss": 0.931,
+      "step": 7898
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.00011928433332065998,
+      "loss": 0.988,
+      "step": 7899
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.00011926720924320286,
+      "loss": 0.8,
+      "step": 7900
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.0001192500845789634,
+      "loss": 0.9374,
+      "step": 7901
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.00011923295932846314,
+      "loss": 0.9467,
+      "step": 7902
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.00011921583349222361,
+      "loss": 0.9403,
+      "step": 7903
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.00011919870707076636,
+      "loss": 0.9455,
+      "step": 7904
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.000119181580064613,
+      "loss": 0.7953,
+      "step": 7905
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.00011916445247428512,
+      "loss": 0.9359,
+      "step": 7906
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.00011914732430030435,
+      "loss": 0.8972,
+      "step": 7907
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.0001191301955431923,
+      "loss": 1.0081,
+      "step": 7908
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.00011911306620347065,
+      "loss": 0.7985,
+      "step": 7909
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.00011909593628166109,
+      "loss": 0.9083,
+      "step": 7910
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.00011907880577828527,
+      "loss": 1.0043,
+      "step": 7911
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.00011906167469386494,
+      "loss": 0.9371,
+      "step": 7912
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.00011904454302892179,
+      "loss": 0.8706,
+      "step": 7913
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.00011902741078397759,
+      "loss": 0.9126,
+      "step": 7914
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.00011901027795955411,
+      "loss": 0.9532,
+      "step": 7915
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.00011899314455617309,
+      "loss": 1.0702,
+      "step": 7916
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.00011897601057435634,
+      "loss": 0.869,
+      "step": 7917
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.00011895887601462572,
+      "loss": 0.8826,
+      "step": 7918
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.00011894174087750301,
+      "loss": 0.7941,
+      "step": 7919
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.00011892460516351007,
+      "loss": 0.8945,
+      "step": 7920
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.0001189074688731688,
+      "loss": 0.8926,
+      "step": 7921
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.00011889033200700106,
+      "loss": 0.9645,
+      "step": 7922
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.00011887319456552873,
+      "loss": 0.8584,
+      "step": 7923
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.00011885605654927378,
+      "loss": 1.0676,
+      "step": 7924
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.0001188389179587581,
+      "loss": 0.8567,
+      "step": 7925
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.00011882177879450369,
+      "loss": 0.9519,
+      "step": 7926
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.0001188046390570325,
+      "loss": 0.8088,
+      "step": 7927
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.0001187874987468665,
+      "loss": 0.9232,
+      "step": 7928
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.00011877035786452774,
+      "loss": 0.8255,
+      "step": 7929
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.0001187532164105382,
+      "loss": 0.8519,
+      "step": 7930
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.00011873607438541997,
+      "loss": 0.9127,
+      "step": 7931
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.00011871893178969507,
+      "loss": 0.831,
+      "step": 7932
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.00011870178862388561,
+      "loss": 1.0357,
+      "step": 7933
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.00011868464488851365,
+      "loss": 0.9637,
+      "step": 7934
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.00011866750058410135,
+      "loss": 0.9192,
+      "step": 7935
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.00011865035571117076,
+      "loss": 0.8045,
+      "step": 7936
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.00011863321027024412,
+      "loss": 0.8975,
+      "step": 7937
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.00011861606426184353,
+      "loss": 0.8071,
+      "step": 7938
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.0001185989176864912,
+      "loss": 0.9193,
+      "step": 7939
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.00011858177054470933,
+      "loss": 0.9061,
+      "step": 7940
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.0001185646228370201,
+      "loss": 0.8932,
+      "step": 7941
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.00011854747456394577,
+      "loss": 0.9967,
+      "step": 7942
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.00011853032572600862,
+      "loss": 0.9478,
+      "step": 7943
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.00011851317632373084,
+      "loss": 0.8577,
+      "step": 7944
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.00011849602635763481,
+      "loss": 0.9998,
+      "step": 7945
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.00011847887582824277,
+      "loss": 0.9031,
+      "step": 7946
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.00011846172473607704,
+      "loss": 0.8438,
+      "step": 7947
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.00011844457308165997,
+      "loss": 0.8798,
+      "step": 7948
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.0001184274208655139,
+      "loss": 0.7674,
+      "step": 7949
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.00011841026808816122,
+      "loss": 0.8535,
+      "step": 7950
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.00011839311475012434,
+      "loss": 0.9209,
+      "step": 7951
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.00011837596085192559,
+      "loss": 0.8227,
+      "step": 7952
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.00011835880639408746,
+      "loss": 0.8627,
+      "step": 7953
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.00011834165137713234,
+      "loss": 0.7385,
+      "step": 7954
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.00011832449580158269,
+      "loss": 0.8084,
+      "step": 7955
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.00011830733966796108,
+      "loss": 0.9103,
+      "step": 7956
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.00011829018297678987,
+      "loss": 0.9324,
+      "step": 7957
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.0001182730257285916,
+      "loss": 0.9774,
+      "step": 7958
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.00011825586792388883,
+      "loss": 0.9049,
+      "step": 7959
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.00011823870956320408,
+      "loss": 0.8894,
+      "step": 7960
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.00011822155064705991,
+      "loss": 0.9696,
+      "step": 7961
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.00011820439117597892,
+      "loss": 0.8616,
+      "step": 7962
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.00011818723115048363,
+      "loss": 0.8399,
+      "step": 7963
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.00011817007057109672,
+      "loss": 0.7957,
+      "step": 7964
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.00011815290943834076,
+      "loss": 0.9217,
+      "step": 7965
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.00011813574775273844,
+      "loss": 0.8248,
+      "step": 7966
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.00011811858551481241,
+      "loss": 0.9378,
+      "step": 7967
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.00011810142272508534,
+      "loss": 0.7495,
+      "step": 7968
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.00011808425938407988,
+      "loss": 1.0091,
+      "step": 7969
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.0001180670954923188,
+      "loss": 0.8294,
+      "step": 7970
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.00011804993105032478,
+      "loss": 0.8925,
+      "step": 7971
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.00011803276605862062,
+      "loss": 0.8845,
+      "step": 7972
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.00011801560051772902,
+      "loss": 0.7954,
+      "step": 7973
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.00011799843442817277,
+      "loss": 0.769,
+      "step": 7974
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.0001179812677904747,
+      "loss": 0.934,
+      "step": 7975
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.00011796410060515754,
+      "loss": 0.9096,
+      "step": 7976
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.0001179469328727442,
+      "loss": 0.7914,
+      "step": 7977
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.0001179297645937575,
+      "loss": 0.9267,
+      "step": 7978
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.00011791259576872029,
+      "loss": 0.9038,
+      "step": 7979
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.00011789542639815543,
+      "loss": 0.9102,
+      "step": 7980
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.00011787825648258583,
+      "loss": 0.9655,
+      "step": 7981
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.00011786108602253441,
+      "loss": 0.8669,
+      "step": 7982
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.00011784391501852407,
+      "loss": 0.8496,
+      "step": 7983
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.00011782674347107776,
+      "loss": 0.9112,
+      "step": 7984
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.00011780957138071849,
+      "loss": 1.0485,
+      "step": 7985
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.00011779239874796917,
+      "loss": 0.7947,
+      "step": 7986
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.00011777522557335277,
+      "loss": 0.8095,
+      "step": 7987
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.00011775805185739237,
+      "loss": 0.7724,
+      "step": 7988
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.00011774087760061099,
+      "loss": 0.8597,
+      "step": 7989
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.00011772370280353162,
+      "loss": 0.8837,
+      "step": 7990
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.00011770652746667737,
+      "loss": 0.9484,
+      "step": 7991
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.00011768935159057129,
+      "loss": 0.8934,
+      "step": 7992
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.00011767217517573643,
+      "loss": 0.9615,
+      "step": 7993
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.00011765499822269599,
+      "loss": 0.81,
+      "step": 7994
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.000117637820731973,
+      "loss": 0.9958,
+      "step": 7995
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.00011762064270409067,
+      "loss": 0.9425,
+      "step": 7996
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.00011760346413957213,
+      "loss": 0.9809,
+      "step": 7997
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.00011758628503894054,
+      "loss": 0.9108,
+      "step": 7998
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.0001175691054027191,
+      "loss": 0.8922,
+      "step": 7999
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.00011755192523143102,
+      "loss": 0.9724,
+      "step": 8000
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.0001175347445255995,
+      "loss": 0.8872,
+      "step": 8001
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.00011751756328574783,
+      "loss": 0.9564,
+      "step": 8002
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.00011750038151239921,
+      "loss": 0.8604,
+      "step": 8003
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.00011748319920607693,
+      "loss": 1.0429,
+      "step": 8004
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.0001174660163673043,
+      "loss": 0.8039,
+      "step": 8005
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.00011744883299660456,
+      "loss": 0.9626,
+      "step": 8006
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.00011743164909450111,
+      "loss": 0.9829,
+      "step": 8007
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.00011741446466151721,
+      "loss": 0.897,
+      "step": 8008
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.00011739727969817627,
+      "loss": 0.9415,
+      "step": 8009
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.00011738009420500162,
+      "loss": 0.8245,
+      "step": 8010
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.00011736290818251665,
+      "loss": 0.8902,
+      "step": 8011
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.00011734572163124477,
+      "loss": 0.9325,
+      "step": 8012
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.0001173285345517094,
+      "loss": 0.9349,
+      "step": 8013
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.00011731134694443397,
+      "loss": 0.8176,
+      "step": 8014
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.00011729415880994193,
+      "loss": 0.9373,
+      "step": 8015
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.00011727697014875673,
+      "loss": 0.8893,
+      "step": 8016
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.00011725978096140184,
+      "loss": 0.9369,
+      "step": 8017
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.0001172425912484008,
+      "loss": 0.8591,
+      "step": 8018
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.00011722540101027708,
+      "loss": 0.879,
+      "step": 8019
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.00011720821024755422,
+      "loss": 0.8249,
+      "step": 8020
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.0001171910189607558,
+      "loss": 0.8706,
+      "step": 8021
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.0001171738271504053,
+      "loss": 0.9111,
+      "step": 8022
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.00011715663481702637,
+      "loss": 0.9667,
+      "step": 8023
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.00011713944196114259,
+      "loss": 1.0082,
+      "step": 8024
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.00011712224858327753,
+      "loss": 0.8842,
+      "step": 8025
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.00011710505468395484,
+      "loss": 1.0382,
+      "step": 8026
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.00011708786026369821,
+      "loss": 0.9502,
+      "step": 8027
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.00011707066532303117,
+      "loss": 0.9387,
+      "step": 8028
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.00011705346986247752,
+      "loss": 0.946,
+      "step": 8029
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.00011703627388256086,
+      "loss": 0.8336,
+      "step": 8030
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.00011701907738380496,
+      "loss": 0.9791,
+      "step": 8031
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.00011700188036673348,
+      "loss": 0.8164,
+      "step": 8032
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.00011698468283187017,
+      "loss": 0.9966,
+      "step": 8033
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.00011696748477973878,
+      "loss": 0.8628,
+      "step": 8034
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.00011695028621086313,
+      "loss": 0.8911,
+      "step": 8035
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.00011693308712576691,
+      "loss": 0.9707,
+      "step": 8036
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.00011691588752497402,
+      "loss": 0.7795,
+      "step": 8037
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.0001168986874090082,
+      "loss": 0.7809,
+      "step": 8038
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.00011688148677839328,
+      "loss": 0.8987,
+      "step": 8039
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.00011686428563365315,
+      "loss": 0.944,
+      "step": 8040
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.00011684708397531161,
+      "loss": 0.914,
+      "step": 8041
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.0001168298818038926,
+      "loss": 1.0949,
+      "step": 8042
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.00011681267911991997,
+      "loss": 0.8639,
+      "step": 8043
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.00011679547592391763,
+      "loss": 1.0037,
+      "step": 8044
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.00011677827221640951,
+      "loss": 1.1124,
+      "step": 8045
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.00011676106799791957,
+      "loss": 0.819,
+      "step": 8046
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.00011674386326897173,
+      "loss": 0.9356,
+      "step": 8047
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.00011672665803008998,
+      "loss": 0.7943,
+      "step": 8048
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.0001167094522817983,
+      "loss": 0.8522,
+      "step": 8049
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.00011669224602462068,
+      "loss": 0.9904,
+      "step": 8050
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.00011667503925908116,
+      "loss": 0.9097,
+      "step": 8051
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.00011665783198570375,
+      "loss": 1.0053,
+      "step": 8052
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.00011664062420501249,
+      "loss": 0.7982,
+      "step": 8053
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.00011662341591753149,
+      "loss": 0.9521,
+      "step": 8054
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.00011660620712378476,
+      "loss": 0.919,
+      "step": 8055
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.00011658899782429644,
+      "loss": 0.9753,
+      "step": 8056
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.00011657178801959063,
+      "loss": 0.9802,
+      "step": 8057
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.00011655457771019142,
+      "loss": 0.9395,
+      "step": 8058
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.00011653736689662303,
+      "loss": 0.8829,
+      "step": 8059
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.00011652015557940956,
+      "loss": 0.9514,
+      "step": 8060
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.00011650294375907515,
+      "loss": 0.8719,
+      "step": 8061
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.00011648573143614405,
+      "loss": 0.9219,
+      "step": 8062
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.00011646851861114041,
+      "loss": 0.9314,
+      "step": 8063
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.0001164513052845885,
+      "loss": 0.9235,
+      "step": 8064
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.00011643409145701251,
+      "loss": 0.8215,
+      "step": 8065
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.00011641687712893668,
+      "loss": 1.0118,
+      "step": 8066
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.0001163996623008853,
+      "loss": 0.8991,
+      "step": 8067
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.00011638244697338261,
+      "loss": 0.9171,
+      "step": 8068
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.00011636523114695293,
+      "loss": 0.8727,
+      "step": 8069
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.00011634801482212059,
+      "loss": 0.9216,
+      "step": 8070
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.00011633079799940988,
+      "loss": 1.0425,
+      "step": 8071
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.00011631358067934515,
+      "loss": 0.8749,
+      "step": 8072
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.00011629636286245075,
+      "loss": 0.7736,
+      "step": 8073
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.00011627914454925102,
+      "loss": 0.9933,
+      "step": 8074
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.00011626192574027039,
+      "loss": 0.9442,
+      "step": 8075
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.00011624470643603324,
+      "loss": 0.8633,
+      "step": 8076
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.000116227486637064,
+      "loss": 1.0166,
+      "step": 8077
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.00011621026634388705,
+      "loss": 0.7576,
+      "step": 8078
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.00011619304555702688,
+      "loss": 0.8852,
+      "step": 8079
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.0001161758242770079,
+      "loss": 0.9279,
+      "step": 8080
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.00011615860250435466,
+      "loss": 0.8574,
+      "step": 8081
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.00011614138023959157,
+      "loss": 1.0614,
+      "step": 8082
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.00011612415748324322,
+      "loss": 0.8715,
+      "step": 8083
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.00011610693423583403,
+      "loss": 1.0116,
+      "step": 8084
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.0001160897104978886,
+      "loss": 0.8766,
+      "step": 8085
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.00011607248626993146,
+      "loss": 0.8135,
+      "step": 8086
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.00011605526155248713,
+      "loss": 0.8263,
+      "step": 8087
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.00011603803634608028,
+      "loss": 0.9378,
+      "step": 8088
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.00011602081065123547,
+      "loss": 0.8758,
+      "step": 8089
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.00011600358446847726,
+      "loss": 0.9351,
+      "step": 8090
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.00011598635779833027,
+      "loss": 1.0541,
+      "step": 8091
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.00011596913064131923,
+      "loss": 0.7816,
+      "step": 8092
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.00011595190299796869,
+      "loss": 1.0038,
+      "step": 8093
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.00011593467486880341,
+      "loss": 0.8592,
+      "step": 8094
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.00011591744625434802,
+      "loss": 0.874,
+      "step": 8095
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.0001159002171551272,
+      "loss": 0.8878,
+      "step": 8096
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.00011588298757166569,
+      "loss": 1.0672,
+      "step": 8097
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.0001158657575044882,
+      "loss": 0.9507,
+      "step": 8098
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.00011584852695411949,
+      "loss": 0.8954,
+      "step": 8099
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.00011583129592108433,
+      "loss": 0.9103,
+      "step": 8100
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.00011581406440590743,
+      "loss": 0.973,
+      "step": 8101
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.00011579683240911366,
+      "loss": 0.9449,
+      "step": 8102
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.0001157795999312277,
+      "loss": 0.9814,
+      "step": 8103
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.0001157623669727745,
+      "loss": 0.8614,
+      "step": 8104
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.0001157451335342788,
+      "loss": 0.9504,
+      "step": 8105
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.00011572789961626551,
+      "loss": 0.9317,
+      "step": 8106
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.0001157106652192594,
+      "loss": 0.9253,
+      "step": 8107
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.00011569343034378543,
+      "loss": 0.9155,
+      "step": 8108
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.00011567619499036841,
+      "loss": 0.9257,
+      "step": 8109
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.00011565895915953333,
+      "loss": 0.887,
+      "step": 8110
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.00011564172285180505,
+      "loss": 0.9073,
+      "step": 8111
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.00011562448606770848,
+      "loss": 0.8245,
+      "step": 8112
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.00011560724880776862,
+      "loss": 1.1005,
+      "step": 8113
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.00011559001107251042,
+      "loss": 0.9304,
+      "step": 8114
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.0001155727728624588,
+      "loss": 0.9263,
+      "step": 8115
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.00011555553417813883,
+      "loss": 0.9761,
+      "step": 8116
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.00011553829502007547,
+      "loss": 0.8789,
+      "step": 8117
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.00011552105538879372,
+      "loss": 0.8313,
+      "step": 8118
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.00011550381528481866,
+      "loss": 0.9146,
+      "step": 8119
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.0001154865747086753,
+      "loss": 0.802,
+      "step": 8120
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.00011546933366088873,
+      "loss": 0.9227,
+      "step": 8121
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.00011545209214198398,
+      "loss": 0.9719,
+      "step": 8122
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.00011543485015248619,
+      "loss": 0.9024,
+      "step": 8123
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.00011541760769292045,
+      "loss": 0.8105,
+      "step": 8124
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.00011540036476381185,
+      "loss": 0.9962,
+      "step": 8125
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.00011538312136568554,
+      "loss": 1.0269,
+      "step": 8126
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.00011536587749906671,
+      "loss": 0.901,
+      "step": 8127
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.00011534863316448043,
+      "loss": 0.9387,
+      "step": 8128
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.00011533138836245199,
+      "loss": 0.9932,
+      "step": 8129
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.0001153141430935065,
+      "loss": 0.9052,
+      "step": 8130
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.00011529689735816918,
+      "loss": 0.8382,
+      "step": 8131
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.00011527965115696524,
+      "loss": 0.9294,
+      "step": 8132
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.00011526240449041994,
+      "loss": 0.8478,
+      "step": 8133
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.0001152451573590585,
+      "loss": 0.8164,
+      "step": 8134
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.00011522790976340621,
+      "loss": 0.7406,
+      "step": 8135
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.0001152106617039883,
+      "loss": 0.9412,
+      "step": 8136
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.0001151934131813301,
+      "loss": 0.9424,
+      "step": 8137
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.00011517616419595692,
+      "loss": 0.9949,
+      "step": 8138
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.00011515891474839401,
+      "loss": 1.0104,
+      "step": 8139
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.0001151416648391668,
+      "loss": 0.8762,
+      "step": 8140
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.00011512441446880058,
+      "loss": 0.7898,
+      "step": 8141
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.00011510716363782068,
+      "loss": 0.8183,
+      "step": 8142
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.00011508991234675254,
+      "loss": 0.9759,
+      "step": 8143
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.00011507266059612149,
+      "loss": 0.8336,
+      "step": 8144
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.00011505540838645297,
+      "loss": 0.7878,
+      "step": 8145
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.00011503815571827242,
+      "loss": 0.9759,
+      "step": 8146
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.00011502090259210519,
+      "loss": 0.9414,
+      "step": 8147
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.00011500364900847676,
+      "loss": 0.8725,
+      "step": 8148
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.0001149863949679126,
+      "loss": 0.9535,
+      "step": 8149
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.00011496914047093816,
+      "loss": 0.9486,
+      "step": 8150
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.00011495188551807898,
+      "loss": 1.0108,
+      "step": 8151
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.00011493463010986052,
+      "loss": 0.9049,
+      "step": 8152
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.00011491737424680824,
+      "loss": 0.9066,
+      "step": 8153
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.00011490011792944777,
+      "loss": 0.92,
+      "step": 8154
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.00011488286115830455,
+      "loss": 0.8779,
+      "step": 8155
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.00011486560393390425,
+      "loss": 0.75,
+      "step": 8156
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.00011484834625677234,
+      "loss": 0.9957,
+      "step": 8157
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.00011483108812743441,
+      "loss": 0.8113,
+      "step": 8158
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.00011481382954641611,
+      "loss": 0.9063,
+      "step": 8159
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.000114796570514243,
+      "loss": 0.9783,
+      "step": 8160
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.00011477931103144072,
+      "loss": 0.958,
+      "step": 8161
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.00011476205109853493,
+      "loss": 0.8798,
+      "step": 8162
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.00011474479071605127,
+      "loss": 0.7823,
+      "step": 8163
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.00011472752988451536,
+      "loss": 0.9483,
+      "step": 8164
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.00011471026860445295,
+      "loss": 0.9439,
+      "step": 8165
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.00011469300687638967,
+      "loss": 0.7964,
+      "step": 8166
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.00011467574470085127,
+      "loss": 0.9138,
+      "step": 8167
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.00011465848207836343,
+      "loss": 0.8,
+      "step": 8168
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.00011464121900945193,
+      "loss": 1.023,
+      "step": 8169
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.00011462395549464249,
+      "loss": 0.9429,
+      "step": 8170
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.00011460669153446085,
+      "loss": 0.9466,
+      "step": 8171
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.00011458942712943277,
+      "loss": 0.8272,
+      "step": 8172
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.00011457216228008412,
+      "loss": 0.8753,
+      "step": 8173
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.00011455489698694064,
+      "loss": 0.9049,
+      "step": 8174
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.00011453763125052815,
+      "loss": 0.9452,
+      "step": 8175
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.00011452036507137249,
+      "loss": 0.7953,
+      "step": 8176
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.00011450309844999947,
+      "loss": 0.8489,
+      "step": 8177
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.000114485831386935,
+      "loss": 1.0188,
+      "step": 8178
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.00011446856388270488,
+      "loss": 0.9371,
+      "step": 8179
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.00011445129593783506,
+      "loss": 0.955,
+      "step": 8180
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.0001144340275528514,
+      "loss": 0.9321,
+      "step": 8181
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.00011441675872827977,
+      "loss": 1.0124,
+      "step": 8182
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.00011439948946464616,
+      "loss": 0.892,
+      "step": 8183
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.00011438221976247648,
+      "loss": 0.9017,
+      "step": 8184
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.00011436494962229664,
+      "loss": 0.9215,
+      "step": 8185
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.00011434767904463268,
+      "loss": 0.9284,
+      "step": 8186
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.00011433040803001053,
+      "loss": 0.8769,
+      "step": 8187
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.00011431313657895615,
+      "loss": 0.8183,
+      "step": 8188
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.00011429586469199561,
+      "loss": 0.9029,
+      "step": 8189
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.00011427859236965484,
+      "loss": 0.8908,
+      "step": 8190
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.00011426131961245992,
+      "loss": 0.9145,
+      "step": 8191
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.00011424404642093693,
+      "loss": 0.9835,
+      "step": 8192
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.00011422677279561185,
+      "loss": 0.8605,
+      "step": 8193
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.00011420949873701077,
+      "loss": 0.8059,
+      "step": 8194
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.00011419222424565981,
+      "loss": 0.84,
+      "step": 8195
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.000114174949322085,
+      "loss": 0.8474,
+      "step": 8196
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.00011415767396681249,
+      "loss": 0.8533,
+      "step": 8197
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.00011414039818036841,
+      "loss": 0.946,
+      "step": 8198
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.00011412312196327884,
+      "loss": 0.9351,
+      "step": 8199
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.00011410584531606998,
+      "loss": 0.9254,
+      "step": 8200
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.00011408856823926797,
+      "loss": 0.874,
+      "step": 8201
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.00011407129073339897,
+      "loss": 0.8703,
+      "step": 8202
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.00011405401279898922,
+      "loss": 0.864,
+      "step": 8203
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.00011403673443656485,
+      "loss": 0.9215,
+      "step": 8204
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.00011401945564665209,
+      "loss": 0.8237,
+      "step": 8205
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.0001140021764297772,
+      "loss": 0.9346,
+      "step": 8206
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.00011398489678646637,
+      "loss": 0.876,
+      "step": 8207
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.00011396761671724592,
+      "loss": 0.92,
+      "step": 8208
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.00011395033622264205,
+      "loss": 0.7337,
+      "step": 8209
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.00011393305530318103,
+      "loss": 0.928,
+      "step": 8210
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.0001139157739593892,
+      "loss": 0.7567,
+      "step": 8211
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.0001138984921917928,
+      "loss": 0.8658,
+      "step": 8212
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.00011388121000091824,
+      "loss": 0.8964,
+      "step": 8213
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.00011386392738729177,
+      "loss": 0.8474,
+      "step": 8214
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.00011384664435143977,
+      "loss": 0.8552,
+      "step": 8215
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.00011382936089388856,
+      "loss": 1.0869,
+      "step": 8216
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.00011381207701516452,
+      "loss": 0.885,
+      "step": 8217
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.00011379479271579405,
+      "loss": 0.9291,
+      "step": 8218
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.00011377750799630353,
+      "loss": 0.935,
+      "step": 8219
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.00011376022285721937,
+      "loss": 0.9214,
+      "step": 8220
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.00011374293729906797,
+      "loss": 1.088,
+      "step": 8221
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.00011372565132237581,
+      "loss": 0.9468,
+      "step": 8222
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.00011370836492766925,
+      "loss": 1.1106,
+      "step": 8223
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.00011369107811547484,
+      "loss": 0.726,
+      "step": 8224
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.00011367379088631898,
+      "loss": 0.8933,
+      "step": 8225
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.00011365650324072818,
+      "loss": 0.8403,
+      "step": 8226
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.00011363921517922896,
+      "loss": 1.035,
+      "step": 8227
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.00011362192670234774,
+      "loss": 0.9905,
+      "step": 8228
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.00011360463781061112,
+      "loss": 0.861,
+      "step": 8229
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.00011358734850454564,
+      "loss": 0.9326,
+      "step": 8230
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.00011357005878467781,
+      "loss": 0.8913,
+      "step": 8231
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.00011355276865153419,
+      "loss": 0.9067,
+      "step": 8232
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.00011353547810564138,
+      "loss": 1.0231,
+      "step": 8233
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.00011351818714752589,
+      "loss": 0.9752,
+      "step": 8234
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.00011350089577771441,
+      "loss": 1.0258,
+      "step": 8235
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.00011348360399673348,
+      "loss": 0.9067,
+      "step": 8236
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.00011346631180510974,
+      "loss": 0.9297,
+      "step": 8237
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.00011344901920336986,
+      "loss": 0.851,
+      "step": 8238
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.00011343172619204044,
+      "loss": 1.0946,
+      "step": 8239
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.00011341443277164812,
+      "loss": 1.0262,
+      "step": 8240
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.00011339713894271962,
+      "loss": 0.9018,
+      "step": 8241
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.0001133798447057816,
+      "loss": 0.9041,
+      "step": 8242
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.00011336255006136079,
+      "loss": 0.8919,
+      "step": 8243
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.00011334525500998385,
+      "loss": 0.882,
+      "step": 8244
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.0001133279595521775,
+      "loss": 0.931,
+      "step": 8245
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.00011331066368846852,
+      "loss": 0.8612,
+      "step": 8246
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.00011329336741938362,
+      "loss": 0.8635,
+      "step": 8247
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.00011327607074544953,
+      "loss": 0.9972,
+      "step": 8248
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.00011325877366719313,
+      "loss": 0.8815,
+      "step": 8249
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.00011324147618514108,
+      "loss": 0.9554,
+      "step": 8250
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.00011322417829982019,
+      "loss": 0.8214,
+      "step": 8251
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.00011320688001175734,
+      "loss": 0.876,
+      "step": 8252
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.00011318958132147928,
+      "loss": 0.96,
+      "step": 8253
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.00011317228222951288,
+      "loss": 0.8963,
+      "step": 8254
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.00011315498273638498,
+      "loss": 0.9604,
+      "step": 8255
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.00011313768284262241,
+      "loss": 0.9422,
+      "step": 8256
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.00011312038254875208,
+      "loss": 0.801,
+      "step": 8257
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.00011310308185530081,
+      "loss": 0.9912,
+      "step": 8258
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.00011308578076279554,
+      "loss": 0.8788,
+      "step": 8259
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.00011306847927176317,
+      "loss": 1.003,
+      "step": 8260
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.00011305117738273059,
+      "loss": 0.8532,
+      "step": 8261
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.00011303387509622476,
+      "loss": 1.002,
+      "step": 8262
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.0001130165724127726,
+      "loss": 0.8315,
+      "step": 8263
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.00011299926933290104,
+      "loss": 0.8629,
+      "step": 8264
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.00011298196585713712,
+      "loss": 0.9844,
+      "step": 8265
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.00011296466198600773,
+      "loss": 0.977,
+      "step": 8266
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.00011294735772003992,
+      "loss": 0.8613,
+      "step": 8267
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.0001129300530597607,
+      "loss": 0.9806,
+      "step": 8268
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.000112912748005697,
+      "loss": 0.9117,
+      "step": 8269
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.00011289544255837594,
+      "loss": 0.9077,
+      "step": 8270
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.00011287813671832447,
+      "loss": 0.8555,
+      "step": 8271
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.00011286083048606975,
+      "loss": 0.8643,
+      "step": 8272
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.00011284352386213874,
+      "loss": 0.8219,
+      "step": 8273
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.00011282621684705855,
+      "loss": 0.7097,
+      "step": 8274
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.00011280890944135623,
+      "loss": 0.9077,
+      "step": 8275
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.00011279160164555896,
+      "loss": 0.9196,
+      "step": 8276
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.00011277429346019374,
+      "loss": 0.9332,
+      "step": 8277
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.00011275698488578781,
+      "loss": 0.8767,
+      "step": 8278
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.00011273967592286823,
+      "loss": 0.8985,
+      "step": 8279
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.00011272236657196211,
+      "loss": 0.9502,
+      "step": 8280
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.00011270505683359669,
+      "loss": 0.8883,
+      "step": 8281
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.00011268774670829908,
+      "loss": 0.8954,
+      "step": 8282
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.00011267043619659645,
+      "loss": 1.0038,
+      "step": 8283
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.00011265312529901611,
+      "loss": 0.8534,
+      "step": 8284
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.00011263581401608507,
+      "loss": 0.8223,
+      "step": 8285
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.0001126185023483307,
+      "loss": 1.1704,
+      "step": 8286
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.00011260119029628016,
+      "loss": 0.8702,
+      "step": 8287
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.00011258387786046069,
+      "loss": 0.9789,
+      "step": 8288
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.00011256656504139958,
+      "loss": 0.9902,
+      "step": 8289
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.00011254925183962403,
+      "loss": 0.8923,
+      "step": 8290
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.00011253193825566135,
+      "loss": 0.9568,
+      "step": 8291
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.00011251462429003883,
+      "loss": 0.7793,
+      "step": 8292
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.00011249730994328373,
+      "loss": 0.8542,
+      "step": 8293
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.0001124799952159234,
+      "loss": 0.9029,
+      "step": 8294
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.00011246268010848519,
+      "loss": 0.9004,
+      "step": 8295
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.00011244536462149632,
+      "loss": 0.8128,
+      "step": 8296
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.0001124280487554842,
+      "loss": 0.958,
+      "step": 8297
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.00011241073251097622,
+      "loss": 0.9319,
+      "step": 8298
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.00011239341588849967,
+      "loss": 0.8409,
+      "step": 8299
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.000112376098888582,
+      "loss": 0.9076,
+      "step": 8300
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.00011235878151175056,
+      "loss": 0.8777,
+      "step": 8301
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.00011234146375853272,
+      "loss": 0.9101,
+      "step": 8302
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.00011232414562945597,
+      "loss": 0.8861,
+      "step": 8303
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.00011230682712504763,
+      "loss": 0.9672,
+      "step": 8304
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.00011228950824583522,
+      "loss": 0.9262,
+      "step": 8305
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.00011227218899234617,
+      "loss": 0.8616,
+      "step": 8306
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.00011225486936510794,
+      "loss": 0.8977,
+      "step": 8307
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.00011223754936464795,
+      "loss": 0.7386,
+      "step": 8308
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.00011222022899149374,
+      "loss": 0.8775,
+      "step": 8309
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.00011220290824617276,
+      "loss": 0.9003,
+      "step": 8310
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.00011218558712921254,
+      "loss": 0.8475,
+      "step": 8311
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.00011216826564114058,
+      "loss": 0.9181,
+      "step": 8312
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.0001121509437824844,
+      "loss": 0.9325,
+      "step": 8313
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.00011213362155377158,
+      "loss": 0.9242,
+      "step": 8314
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.0001121162989555296,
+      "loss": 1.0126,
+      "step": 8315
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.00011209897598828608,
+      "loss": 0.9263,
+      "step": 8316
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.00011208165265256853,
+      "loss": 0.7862,
+      "step": 8317
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.00011206432894890459,
+      "loss": 0.9229,
+      "step": 8318
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.00011204700487782183,
+      "loss": 0.8392,
+      "step": 8319
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.00011202968043984783,
+      "loss": 0.8962,
+      "step": 8320
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.00011201235563551023,
+      "loss": 0.9407,
+      "step": 8321
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.00011199503046533668,
+      "loss": 0.9936,
+      "step": 8322
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.00011197770492985478,
+      "loss": 0.9285,
+      "step": 8323
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.00011196037902959218,
+      "loss": 0.8912,
+      "step": 8324
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.00011194305276507658,
+      "loss": 0.9952,
+      "step": 8325
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.00011192572613683558,
+      "loss": 0.9074,
+      "step": 8326
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.00011190839914539695,
+      "loss": 0.884,
+      "step": 8327
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.00011189107179128828,
+      "loss": 0.8379,
+      "step": 8328
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.00011187374407503737,
+      "loss": 0.8382,
+      "step": 8329
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.00011185641599717191,
+      "loss": 0.9323,
+      "step": 8330
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.0001118390875582196,
+      "loss": 0.8083,
+      "step": 8331
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.00011182175875870815,
+      "loss": 0.9204,
+      "step": 8332
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.00011180442959916537,
+      "loss": 0.8148,
+      "step": 8333
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.00011178710008011902,
+      "loss": 0.8105,
+      "step": 8334
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.00011176977020209682,
+      "loss": 0.8575,
+      "step": 8335
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.0001117524399656266,
+      "loss": 0.882,
+      "step": 8336
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.00011173510937123611,
+      "loss": 0.9341,
+      "step": 8337
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.00011171777841945319,
+      "loss": 0.9161,
+      "step": 8338
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.0001117004471108056,
+      "loss": 0.9435,
+      "step": 8339
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.00011168311544582122,
+      "loss": 0.9882,
+      "step": 8340
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.00011166578342502789,
+      "loss": 0.9316,
+      "step": 8341
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.00011164845104895341,
+      "loss": 0.9207,
+      "step": 8342
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.00011163111831812564,
+      "loss": 0.953,
+      "step": 8343
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.00011161378523307251,
+      "loss": 1.0332,
+      "step": 8344
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.00011159645179432181,
+      "loss": 0.8985,
+      "step": 8345
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.00011157911800240151,
+      "loss": 1.0185,
+      "step": 8346
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.00011156178385783949,
+      "loss": 0.8353,
+      "step": 8347
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.0001115444493611636,
+      "loss": 0.9228,
+      "step": 8348
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.00011152711451290184,
+      "loss": 0.9873,
+      "step": 8349
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.00011150977931358208,
+      "loss": 0.8269,
+      "step": 8350
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.00011149244376373229,
+      "loss": 0.8956,
+      "step": 8351
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.00011147510786388046,
+      "loss": 0.945,
+      "step": 8352
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.00011145777161455453,
+      "loss": 0.8804,
+      "step": 8353
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.00011144043501628243,
+      "loss": 0.8359,
+      "step": 8354
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.0001114230980695922,
+      "loss": 0.8617,
+      "step": 8355
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.00011140576077501182,
+      "loss": 0.9468,
+      "step": 8356
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.00011138842313306927,
+      "loss": 0.837,
+      "step": 8357
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.00011137108514429262,
+      "loss": 0.91,
+      "step": 8358
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.00011135374680920987,
+      "loss": 0.8975,
+      "step": 8359
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.00011133640812834906,
+      "loss": 0.9132,
+      "step": 8360
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.00011131906910223821,
+      "loss": 0.9171,
+      "step": 8361
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.00011130172973140544,
+      "loss": 0.9332,
+      "step": 8362
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.00011128439001637876,
+      "loss": 0.9102,
+      "step": 8363
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.00011126704995768631,
+      "loss": 0.8334,
+      "step": 8364
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.00011124970955585615,
+      "loss": 0.8579,
+      "step": 8365
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.00011123236881141634,
+      "loss": 0.7951,
+      "step": 8366
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.00011121502772489505,
+      "loss": 0.8416,
+      "step": 8367
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.00011119768629682038,
+      "loss": 0.7592,
+      "step": 8368
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.00011118034452772048,
+      "loss": 0.8381,
+      "step": 8369
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.0001111630024181235,
+      "loss": 0.9221,
+      "step": 8370
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.00011114565996855755,
+      "loss": 0.9083,
+      "step": 8371
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.00011112831717955081,
+      "loss": 0.9398,
+      "step": 8372
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.00011111097405163149,
+      "loss": 0.9223,
+      "step": 8373
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.00011109363058532772,
+      "loss": 0.7388,
+      "step": 8374
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.0001110762867811677,
+      "loss": 0.8651,
+      "step": 8375
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.00011105894263967973,
+      "loss": 0.8709,
+      "step": 8376
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.0001110415981613919,
+      "loss": 0.9315,
+      "step": 8377
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.00011102425334683249,
+      "loss": 0.8914,
+      "step": 8378
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.00011100690819652974,
+      "loss": 1.0243,
+      "step": 8379
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.0001109895627110119,
+      "loss": 0.9995,
+      "step": 8380
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.00011097221689080722,
+      "loss": 0.8911,
+      "step": 8381
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.00011095487073644395,
+      "loss": 0.8897,
+      "step": 8382
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.00011093752424845037,
+      "loss": 0.904,
+      "step": 8383
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.0001109201774273548,
+      "loss": 0.8228,
+      "step": 8384
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.00011090283027368549,
+      "loss": 0.9558,
+      "step": 8385
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.00011088548278797077,
+      "loss": 1.0143,
+      "step": 8386
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.000110868134970739,
+      "loss": 0.9681,
+      "step": 8387
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.00011085078682251842,
+      "loss": 0.8317,
+      "step": 8388
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.00011083343834383741,
+      "loss": 0.8866,
+      "step": 8389
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.00011081608953522433,
+      "loss": 0.8396,
+      "step": 8390
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.00011079874039720752,
+      "loss": 0.8221,
+      "step": 8391
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.00011078139093031539,
+      "loss": 0.8178,
+      "step": 8392
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.00011076404113507626,
+      "loss": 0.9188,
+      "step": 8393
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.00011074669101201853,
+      "loss": 1.0735,
+      "step": 8394
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.00011072934056167064,
+      "loss": 0.9924,
+      "step": 8395
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.00011071198978456092,
+      "loss": 0.9579,
+      "step": 8396
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.00011069463868121784,
+      "loss": 0.9457,
+      "step": 8397
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.00011067728725216984,
+      "loss": 0.9057,
+      "step": 8398
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.00011065993549794533,
+      "loss": 0.9501,
+      "step": 8399
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.00011064258341907274,
+      "loss": 0.985,
+      "step": 8400
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.0001106252310160806,
+      "loss": 0.9628,
+      "step": 8401
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.00011060787828949728,
+      "loss": 0.7878,
+      "step": 8402
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.00011059052523985134,
+      "loss": 1.0037,
+      "step": 8403
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.00011057317186767122,
+      "loss": 0.9446,
+      "step": 8404
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.00011055581817348544,
+      "loss": 0.844,
+      "step": 8405
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.00011053846415782248,
+      "loss": 0.906,
+      "step": 8406
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.00011052110982121086,
+      "loss": 0.8172,
+      "step": 8407
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.00011050375516417912,
+      "loss": 0.9194,
+      "step": 8408
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.0001104864001872558,
+      "loss": 0.8577,
+      "step": 8409
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.00011046904489096946,
+      "loss": 0.8555,
+      "step": 8410
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.00011045168927584859,
+      "loss": 0.9861,
+      "step": 8411
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.00011043433334242184,
+      "loss": 0.9509,
+      "step": 8412
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.0001104169770912177,
+      "loss": 0.7932,
+      "step": 8413
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.00011039962052276483,
+      "loss": 0.9669,
+      "step": 8414
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.00011038226363759177,
+      "loss": 0.8865,
+      "step": 8415
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.00011036490643622717,
+      "loss": 0.9474,
+      "step": 8416
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.00011034754891919958,
+      "loss": 0.933,
+      "step": 8417
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.00011033019108703768,
+      "loss": 0.9025,
+      "step": 8418
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.00011031283294027008,
+      "loss": 0.9019,
+      "step": 8419
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.00011029547447942541,
+      "loss": 0.7996,
+      "step": 8420
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.00011027811570503235,
+      "loss": 0.8424,
+      "step": 8421
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.00011026075661761957,
+      "loss": 0.952,
+      "step": 8422
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.00011024339721771566,
+      "loss": 0.8824,
+      "step": 8423
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.00011022603750584939,
+      "loss": 0.9215,
+      "step": 8424
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.00011020867748254942,
+      "loss": 0.9424,
+      "step": 8425
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.00011019131714834441,
+      "loss": 0.8219,
+      "step": 8426
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.00011017395650376315,
+      "loss": 0.9685,
+      "step": 8427
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.0001101565955493343,
+      "loss": 0.8776,
+      "step": 8428
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.00011013923428558659,
+      "loss": 0.9225,
+      "step": 8429
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.00011012187271304877,
+      "loss": 1.0459,
+      "step": 8430
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.00011010451083224958,
+      "loss": 0.7993,
+      "step": 8431
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.00011008714864371774,
+      "loss": 0.8512,
+      "step": 8432
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.00011006978614798214,
+      "loss": 0.8497,
+      "step": 8433
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.0001100524233455714,
+      "loss": 0.9207,
+      "step": 8434
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.00011003506023701438,
+      "loss": 1.0068,
+      "step": 8435
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.00011001769682283987,
+      "loss": 0.8114,
+      "step": 8436
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.00011000033310357667,
+      "loss": 0.7751,
+      "step": 8437
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.00010998296907975358,
+      "loss": 0.9579,
+      "step": 8438
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.00010996560475189946,
+      "loss": 0.7548,
+      "step": 8439
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.00010994824012054305,
+      "loss": 1.0226,
+      "step": 8440
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.0001099308751862133,
+      "loss": 0.7843,
+      "step": 8441
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.00010991350994943899,
+      "loss": 1.0147,
+      "step": 8442
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.00010989614441074898,
+      "loss": 0.8895,
+      "step": 8443
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.00010987877857067217,
+      "loss": 0.9518,
+      "step": 8444
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.00010986141242973743,
+      "loss": 0.9601,
+      "step": 8445
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.0001098440459884736,
+      "loss": 0.914,
+      "step": 8446
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.00010982667924740964,
+      "loss": 0.9853,
+      "step": 8447
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.00010980931220707438,
+      "loss": 0.956,
+      "step": 8448
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.0001097919448679968,
+      "loss": 0.9002,
+      "step": 8449
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.00010977457723070579,
+      "loss": 0.9653,
+      "step": 8450
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.0001097572092957303,
+      "loss": 0.8657,
+      "step": 8451
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.00010973984106359927,
+      "loss": 0.8758,
+      "step": 8452
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.00010972247253484161,
+      "loss": 0.9033,
+      "step": 8453
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.00010970510370998629,
+      "loss": 0.8278,
+      "step": 8454
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.00010968773458956232,
+      "loss": 0.8923,
+      "step": 8455
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.00010967036517409863,
+      "loss": 0.7183,
+      "step": 8456
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.00010965299546412424,
+      "loss": 0.9368,
+      "step": 8457
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.00010963562546016812,
+      "loss": 0.8123,
+      "step": 8458
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.00010961825516275926,
+      "loss": 0.8367,
+      "step": 8459
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.00010960088457242672,
+      "loss": 0.9149,
+      "step": 8460
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.00010958351368969949,
+      "loss": 0.8496,
+      "step": 8461
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.00010956614251510657,
+      "loss": 0.8376,
+      "step": 8462
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.00010954877104917706,
+      "loss": 1.1034,
+      "step": 8463
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.00010953139929243998,
+      "loss": 0.9119,
+      "step": 8464
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.00010951402724542439,
+      "loss": 0.9753,
+      "step": 8465
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.00010949665490865932,
+      "loss": 1.0395,
+      "step": 8466
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.00010947928228267388,
+      "loss": 0.8709,
+      "step": 8467
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.00010946190936799718,
+      "loss": 0.8348,
+      "step": 8468
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.00010944453616515824,
+      "loss": 0.8376,
+      "step": 8469
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.0001094271626746862,
+      "loss": 0.9232,
+      "step": 8470
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.0001094097888971102,
+      "loss": 0.7876,
+      "step": 8471
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.0001093924148329593,
+      "loss": 0.9253,
+      "step": 8472
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.00010937504048276268,
+      "loss": 0.9701,
+      "step": 8473
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.00010935766584704943,
+      "loss": 0.9046,
+      "step": 8474
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.00010934029092634869,
+      "loss": 0.9415,
+      "step": 8475
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.0001093229157211897,
+      "loss": 0.8388,
+      "step": 8476
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.00010930554023210148,
+      "loss": 0.9697,
+      "step": 8477
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.0001092881644596133,
+      "loss": 0.8604,
+      "step": 8478
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.00010927078840425436,
+      "loss": 0.9669,
+      "step": 8479
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.00010925341206655375,
+      "loss": 0.9405,
+      "step": 8480
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.00010923603544704072,
+      "loss": 0.909,
+      "step": 8481
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.00010921865854624451,
+      "loss": 0.853,
+      "step": 8482
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.00010920128136469425,
+      "loss": 0.8384,
+      "step": 8483
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.00010918390390291926,
+      "loss": 0.828,
+      "step": 8484
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.00010916652616144868,
+      "loss": 0.8677,
+      "step": 8485
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.00010914914814081179,
+      "loss": 0.9007,
+      "step": 8486
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.00010913176984153786,
+      "loss": 0.8488,
+      "step": 8487
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.00010911439126415608,
+      "loss": 0.8765,
+      "step": 8488
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.00010909701240919577,
+      "loss": 0.875,
+      "step": 8489
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.00010907963327718617,
+      "loss": 0.8161,
+      "step": 8490
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.00010906225386865663,
+      "loss": 0.9564,
+      "step": 8491
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.00010904487418413634,
+      "loss": 0.9449,
+      "step": 8492
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.00010902749422415465,
+      "loss": 0.7876,
+      "step": 8493
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.00010901011398924086,
+      "loss": 1.0108,
+      "step": 8494
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.00010899273347992429,
+      "loss": 0.9123,
+      "step": 8495
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.00010897535269673426,
+      "loss": 0.931,
+      "step": 8496
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.0001089579716402001,
+      "loss": 0.8895,
+      "step": 8497
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.00010894059031085116,
+      "loss": 0.9943,
+      "step": 8498
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.00010892320870921675,
+      "loss": 0.8797,
+      "step": 8499
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.00010890582683582626,
+      "loss": 0.8585,
+      "step": 8500
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.00010888844469120907,
+      "loss": 0.8748,
+      "step": 8501
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.00010887106227589452,
+      "loss": 0.8937,
+      "step": 8502
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.00010885367959041198,
+      "loss": 0.8718,
+      "step": 8503
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.00010883629663529089,
+      "loss": 0.8458,
+      "step": 8504
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.00010881891341106059,
+      "loss": 1.0027,
+      "step": 8505
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.00010880152991825055,
+      "loss": 0.7976,
+      "step": 8506
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.0001087841461573901,
+      "loss": 0.9578,
+      "step": 8507
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.00010876676212900874,
+      "loss": 1.0047,
+      "step": 8508
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.00010874937783363588,
+      "loss": 1.1498,
+      "step": 8509
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.00010873199327180091,
+      "loss": 0.8882,
+      "step": 8510
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.00010871460844403332,
+      "loss": 0.9082,
+      "step": 8511
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.00010869722335086259,
+      "loss": 0.8755,
+      "step": 8512
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.00010867983799281813,
+      "loss": 0.9085,
+      "step": 8513
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.00010866245237042944,
+      "loss": 0.9729,
+      "step": 8514
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.00010864506648422599,
+      "loss": 0.9166,
+      "step": 8515
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.00010862768033473725,
+      "loss": 0.9235,
+      "step": 8516
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.00010861029392249275,
+      "loss": 0.8966,
+      "step": 8517
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.00010859290724802198,
+      "loss": 0.8979,
+      "step": 8518
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.00010857552031185443,
+      "loss": 0.8297,
+      "step": 8519
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.00010855813311451965,
+      "loss": 0.6946,
+      "step": 8520
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.00010854074565654713,
+      "loss": 0.9488,
+      "step": 8521
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.00010852335793846644,
+      "loss": 1.0097,
+      "step": 8522
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.00010850596996080711,
+      "loss": 0.8745,
+      "step": 8523
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.00010848858172409866,
+      "loss": 0.8462,
+      "step": 8524
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.00010847119322887071,
+      "loss": 0.8525,
+      "step": 8525
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.00010845380447565279,
+      "loss": 0.8619,
+      "step": 8526
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.00010843641546497445,
+      "loss": 0.942,
+      "step": 8527
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.00010841902619736533,
+      "loss": 0.7955,
+      "step": 8528
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.00010840163667335497,
+      "loss": 0.916,
+      "step": 8529
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.00010838424689347299,
+      "loss": 0.9422,
+      "step": 8530
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.00010836685685824901,
+      "loss": 0.8407,
+      "step": 8531
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.00010834946656821259,
+      "loss": 1.0344,
+      "step": 8532
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.00010833207602389342,
+      "loss": 0.8925,
+      "step": 8533
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.00010831468522582104,
+      "loss": 0.9596,
+      "step": 8534
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.00010829729417452516,
+      "loss": 1.042,
+      "step": 8535
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.00010827990287053543,
+      "loss": 1.0135,
+      "step": 8536
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.00010826251131438147,
+      "loss": 0.8999,
+      "step": 8537
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.00010824511950659293,
+      "loss": 0.9448,
+      "step": 8538
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.0001082277274476995,
+      "loss": 1.0228,
+      "step": 8539
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.00010821033513823083,
+      "loss": 0.957,
+      "step": 8540
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.00010819294257871664,
+      "loss": 0.8486,
+      "step": 8541
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.00010817554976968658,
+      "loss": 0.9283,
+      "step": 8542
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.0001081581567116704,
+      "loss": 0.9066,
+      "step": 8543
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.00010814076340519774,
+      "loss": 0.9447,
+      "step": 8544
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.00010812336985079837,
+      "loss": 0.7216,
+      "step": 8545
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.00010810597604900194,
+      "loss": 0.9897,
+      "step": 8546
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.00010808858200033826,
+      "loss": 0.9027,
+      "step": 8547
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.000108071187705337,
+      "loss": 0.8736,
+      "step": 8548
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.00010805379316452797,
+      "loss": 0.7975,
+      "step": 8549
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.00010803639837844087,
+      "loss": 0.8415,
+      "step": 8550
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.00010801900334760545,
+      "loss": 0.9363,
+      "step": 8551
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.00010800160807255151,
+      "loss": 0.9407,
+      "step": 8552
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.00010798421255380877,
+      "loss": 1.0326,
+      "step": 8553
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.00010796681679190708,
+      "loss": 0.9468,
+      "step": 8554
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.0001079494207873762,
+      "loss": 0.9595,
+      "step": 8555
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.0001079320245407459,
+      "loss": 0.8975,
+      "step": 8556
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.000107914628052546,
+      "loss": 0.9658,
+      "step": 8557
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.00010789723132330631,
+      "loss": 1.0094,
+      "step": 8558
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.00010787983435355664,
+      "loss": 0.8595,
+      "step": 8559
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.00010786243714382684,
+      "loss": 0.8622,
+      "step": 8560
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.00010784503969464675,
+      "loss": 0.874,
+      "step": 8561
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.00010782764200654613,
+      "loss": 0.9134,
+      "step": 8562
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.00010781024408005491,
+      "loss": 0.916,
+      "step": 8563
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.00010779284591570288,
+      "loss": 0.8394,
+      "step": 8564
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.00010777544751401996,
+      "loss": 0.732,
+      "step": 8565
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.000107758048875536,
+      "loss": 0.8608,
+      "step": 8566
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.00010774065000078084,
+      "loss": 0.8698,
+      "step": 8567
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.00010772325089028439,
+      "loss": 0.8822,
+      "step": 8568
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.00010770585154457652,
+      "loss": 0.9598,
+      "step": 8569
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.00010768845196418718,
+      "loss": 0.9031,
+      "step": 8570
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.00010767105214964623,
+      "loss": 0.9246,
+      "step": 8571
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.00010765365210148359,
+      "loss": 0.8971,
+      "step": 8572
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.00010763625182022914,
+      "loss": 0.9295,
+      "step": 8573
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.0001076188513064129,
+      "loss": 0.9116,
+      "step": 8574
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.0001076014505605647,
+      "loss": 0.9035,
+      "step": 8575
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.00010758404958321453,
+      "loss": 0.7641,
+      "step": 8576
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.00010756664837489234,
+      "loss": 0.9156,
+      "step": 8577
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.00010754924693612806,
+      "loss": 0.8636,
+      "step": 8578
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.00010753184526745168,
+      "loss": 0.8587,
+      "step": 8579
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.00010751444336939311,
+      "loss": 0.9121,
+      "step": 8580
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.00010749704124248239,
+      "loss": 0.988,
+      "step": 8581
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.00010747963888724948,
+      "loss": 0.868,
+      "step": 8582
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.00010746223630422438,
+      "loss": 1.0882,
+      "step": 8583
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.00010744483349393703,
+      "loss": 0.8543,
+      "step": 8584
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.00010742743045691752,
+      "loss": 0.8915,
+      "step": 8585
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.00010741002719369577,
+      "loss": 0.921,
+      "step": 8586
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.00010739262370480185,
+      "loss": 0.8888,
+      "step": 8587
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.00010737521999076576,
+      "loss": 0.7854,
+      "step": 8588
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.00010735781605211757,
+      "loss": 1.0676,
+      "step": 8589
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.00010734041188938726,
+      "loss": 0.9285,
+      "step": 8590
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.00010732300750310488,
+      "loss": 0.8506,
+      "step": 8591
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.00010730560289380052,
+      "loss": 0.9536,
+      "step": 8592
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.00010728819806200424,
+      "loss": 0.8137,
+      "step": 8593
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.00010727079300824603,
+      "loss": 0.8999,
+      "step": 8594
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.00010725338773305604,
+      "loss": 1.0139,
+      "step": 8595
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.00010723598223696432,
+      "loss": 0.9267,
+      "step": 8596
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.00010721857652050094,
+      "loss": 0.8985,
+      "step": 8597
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.00010720117058419601,
+      "loss": 0.8607,
+      "step": 8598
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.0001071837644285796,
+      "loss": 0.9102,
+      "step": 8599
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.00010716635805418185,
+      "loss": 0.8441,
+      "step": 8600
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.00010714895146153285,
+      "loss": 0.7765,
+      "step": 8601
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.00010713154465116271,
+      "loss": 1.1226,
+      "step": 8602
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.00010711413762360157,
+      "loss": 0.9417,
+      "step": 8603
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.00010709673037937957,
+      "loss": 0.9638,
+      "step": 8604
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.0001070793229190268,
+      "loss": 0.7825,
+      "step": 8605
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.00010706191524307348,
+      "loss": 0.9,
+      "step": 8606
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.00010704450735204972,
+      "loss": 0.8743,
+      "step": 8607
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.00010702709924648561,
+      "loss": 0.9225,
+      "step": 8608
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.00010700969092691145,
+      "loss": 0.8284,
+      "step": 8609
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.00010699228239385728,
+      "loss": 1.0523,
+      "step": 8610
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.00010697487364785334,
+      "loss": 0.8271,
+      "step": 8611
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.00010695746468942987,
+      "loss": 0.9076,
+      "step": 8612
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.00010694005551911692,
+      "loss": 1.0106,
+      "step": 8613
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.00010692264613744478,
+      "loss": 0.8682,
+      "step": 8614
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.00010690523654494366,
+      "loss": 1.0112,
+      "step": 8615
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.0001068878267421437,
+      "loss": 0.9091,
+      "step": 8616
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.0001068704167295752,
+      "loss": 0.8436,
+      "step": 8617
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.00010685300650776832,
+      "loss": 0.8122,
+      "step": 8618
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.00010683559607725331,
+      "loss": 0.8516,
+      "step": 8619
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.00010681818543856041,
+      "loss": 0.8215,
+      "step": 8620
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.00010680077459221983,
+      "loss": 0.8581,
+      "step": 8621
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.00010678336353876186,
+      "loss": 0.9469,
+      "step": 8622
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.00010676595227871674,
+      "loss": 0.9257,
+      "step": 8623
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.00010674854081261468,
+      "loss": 0.9659,
+      "step": 8624
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.00010673112914098602,
+      "loss": 0.8961,
+      "step": 8625
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.00010671371726436096,
+      "loss": 0.9904,
+      "step": 8626
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.00010669630518326983,
+      "loss": 0.975,
+      "step": 8627
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.00010667889289824293,
+      "loss": 0.9029,
+      "step": 8628
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.00010666148040981053,
+      "loss": 0.9983,
+      "step": 8629
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.00010664406771850286,
+      "loss": 0.9575,
+      "step": 8630
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.00010662665482485033,
+      "loss": 0.7874,
+      "step": 8631
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.00010660924172938318,
+      "loss": 1.0091,
+      "step": 8632
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.00010659182843263176,
+      "loss": 1.0003,
+      "step": 8633
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.00010657441493512636,
+      "loss": 0.9045,
+      "step": 8634
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.00010655700123739735,
+      "loss": 0.9748,
+      "step": 8635
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.00010653958733997504,
+      "loss": 0.9041,
+      "step": 8636
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.00010652217324338976,
+      "loss": 0.953,
+      "step": 8637
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.00010650475894817184,
+      "loss": 0.8511,
+      "step": 8638
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.00010648734445485168,
+      "loss": 0.7305,
+      "step": 8639
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.00010646992976395962,
+      "loss": 0.6733,
+      "step": 8640
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.00010645251487602601,
+      "loss": 0.8655,
+      "step": 8641
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.00010643509979158126,
+      "loss": 0.9751,
+      "step": 8642
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.00010641768451115565,
+      "loss": 0.9116,
+      "step": 8643
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.00010640026903527969,
+      "loss": 0.8232,
+      "step": 8644
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.00010638285336448366,
+      "loss": 0.9599,
+      "step": 8645
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.00010636543749929804,
+      "loss": 0.8922,
+      "step": 8646
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.00010634802144025317,
+      "loss": 0.7954,
+      "step": 8647
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.00010633060518787944,
+      "loss": 0.9886,
+      "step": 8648
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.00010631318874270733,
+      "loss": 0.9708,
+      "step": 8649
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.00010629577210526722,
+      "loss": 0.9866,
+      "step": 8650
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.00010627835527608951,
+      "loss": 0.9668,
+      "step": 8651
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.00010626093825570469,
+      "loss": 0.7588,
+      "step": 8652
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.00010624352104464317,
+      "loss": 0.8263,
+      "step": 8653
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.00010622610364343534,
+      "loss": 0.9526,
+      "step": 8654
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.00010620868605261172,
+      "loss": 0.9847,
+      "step": 8655
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.00010619126827270269,
+      "loss": 1.022,
+      "step": 8656
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.00010617385030423876,
+      "loss": 0.8803,
+      "step": 8657
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.00010615643214775042,
+      "loss": 0.9379,
+      "step": 8658
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.00010613901380376807,
+      "loss": 0.862,
+      "step": 8659
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.0001061215952728222,
+      "loss": 1.0135,
+      "step": 8660
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.00010610417655544333,
+      "loss": 0.8021,
+      "step": 8661
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.00010608675765216192,
+      "loss": 0.8306,
+      "step": 8662
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.00010606933856350846,
+      "loss": 1.0167,
+      "step": 8663
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.00010605191929001348,
+      "loss": 0.8377,
+      "step": 8664
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.00010603449983220741,
+      "loss": 0.8848,
+      "step": 8665
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.00010601708019062084,
+      "loss": 0.7956,
+      "step": 8666
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.00010599966036578423,
+      "loss": 0.9176,
+      "step": 8667
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.00010598224035822813,
+      "loss": 0.9501,
+      "step": 8668
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.00010596482016848309,
+      "loss": 0.9298,
+      "step": 8669
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.00010594739979707955,
+      "loss": 0.8548,
+      "step": 8670
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.00010592997924454813,
+      "loss": 0.8785,
+      "step": 8671
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.00010591255851141935,
+      "loss": 1.0024,
+      "step": 8672
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.00010589513759822374,
+      "loss": 0.8114,
+      "step": 8673
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.0001058777165054919,
+      "loss": 0.9953,
+      "step": 8674
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.00010586029523375433,
+      "loss": 0.9101,
+      "step": 8675
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.00010584287378354162,
+      "loss": 0.9368,
+      "step": 8676
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.00010582545215538438,
+      "loss": 0.9786,
+      "step": 8677
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.00010580803034981313,
+      "loss": 0.6998,
+      "step": 8678
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.00010579060836735846,
+      "loss": 0.7912,
+      "step": 8679
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.00010577318620855094,
+      "loss": 0.8699,
+      "step": 8680
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.00010575576387392124,
+      "loss": 0.9358,
+      "step": 8681
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.00010573834136399988,
+      "loss": 1.0139,
+      "step": 8682
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.00010572091867931746,
+      "loss": 0.9473,
+      "step": 8683
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.00010570349582040462,
+      "loss": 0.9248,
+      "step": 8684
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.00010568607278779199,
+      "loss": 0.9317,
+      "step": 8685
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.00010566864958201012,
+      "loss": 0.806,
+      "step": 8686
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.00010565122620358974,
+      "loss": 0.8834,
+      "step": 8687
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.00010563380265306138,
+      "loss": 1.0351,
+      "step": 8688
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.00010561637893095568,
+      "loss": 0.9686,
+      "step": 8689
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.00010559895503780334,
+      "loss": 0.8875,
+      "step": 8690
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.00010558153097413497,
+      "loss": 1.0213,
+      "step": 8691
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.00010556410674048122,
+      "loss": 0.8784,
+      "step": 8692
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.00010554668233737274,
+      "loss": 0.964,
+      "step": 8693
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.0001055292577653402,
+      "loss": 0.8638,
+      "step": 8694
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.00010551183302491423,
+      "loss": 1.0637,
+      "step": 8695
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.00010549440811662557,
+      "loss": 0.9807,
+      "step": 8696
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.00010547698304100482,
+      "loss": 0.9704,
+      "step": 8697
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.00010545955779858275,
+      "loss": 0.8713,
+      "step": 8698
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.00010544213238988996,
+      "loss": 1.0203,
+      "step": 8699
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.00010542470681545716,
+      "loss": 0.8829,
+      "step": 8700
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.00010540728107581505,
+      "loss": 0.7951,
+      "step": 8701
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.00010538985517149433,
+      "loss": 1.1168,
+      "step": 8702
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.00010537242910302574,
+      "loss": 0.9353,
+      "step": 8703
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.00010535500287093997,
+      "loss": 0.9135,
+      "step": 8704
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.00010533757647576771,
+      "loss": 0.9332,
+      "step": 8705
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.0001053201499180397,
+      "loss": 0.8992,
+      "step": 8706
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.00010530272319828667,
+      "loss": 0.8694,
+      "step": 8707
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.00010528529631703931,
+      "loss": 0.7395,
+      "step": 8708
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.00010526786927482844,
+      "loss": 0.8724,
+      "step": 8709
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.00010525044207218472,
+      "loss": 0.9857,
+      "step": 8710
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.00010523301470963893,
+      "loss": 0.8078,
+      "step": 8711
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.00010521558718772184,
+      "loss": 0.9082,
+      "step": 8712
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.00010519815950696415,
+      "loss": 0.9748,
+      "step": 8713
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.00010518073166789663,
+      "loss": 0.8842,
+      "step": 8714
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.00010516330367105012,
+      "loss": 0.997,
+      "step": 8715
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.00010514587551695529,
+      "loss": 0.8078,
+      "step": 8716
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.00010512844720614294,
+      "loss": 0.9675,
+      "step": 8717
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.00010511101873914391,
+      "loss": 0.9954,
+      "step": 8718
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.00010509359011648891,
+      "loss": 1.0315,
+      "step": 8719
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.00010507616133870877,
+      "loss": 0.885,
+      "step": 8720
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.00010505873240633428,
+      "loss": 0.8406,
+      "step": 8721
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.00010504130331989622,
+      "loss": 0.9288,
+      "step": 8722
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.00010502387407992538,
+      "loss": 0.9492,
+      "step": 8723
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.0001050064446869526,
+      "loss": 0.9185,
+      "step": 8724
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.00010498901514150867,
+      "loss": 0.8768,
+      "step": 8725
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.00010497158544412442,
+      "loss": 0.9856,
+      "step": 8726
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.00010495415559533069,
+      "loss": 0.7832,
+      "step": 8727
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.00010493672559565826,
+      "loss": 0.9017,
+      "step": 8728
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.00010491929544563797,
+      "loss": 0.8593,
+      "step": 8729
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.00010490186514580067,
+      "loss": 0.7972,
+      "step": 8730
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.00010488443469667721,
+      "loss": 0.8983,
+      "step": 8731
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.00010486700409879839,
+      "loss": 0.8288,
+      "step": 8732
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.00010484957335269514,
+      "loss": 0.9761,
+      "step": 8733
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.00010483214245889824,
+      "loss": 0.8232,
+      "step": 8734
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.00010481471141793855,
+      "loss": 0.9077,
+      "step": 8735
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.00010479728023034696,
+      "loss": 1.0103,
+      "step": 8736
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.00010477984889665431,
+      "loss": 0.8374,
+      "step": 8737
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.00010476241741739153,
+      "loss": 0.853,
+      "step": 8738
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.00010474498579308942,
+      "loss": 1.0472,
+      "step": 8739
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.00010472755402427889,
+      "loss": 0.8597,
+      "step": 8740
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.00010471012211149083,
+      "loss": 0.8774,
+      "step": 8741
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.00010469269005525612,
+      "loss": 0.8786,
+      "step": 8742
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.00010467525785610569,
+      "loss": 0.9467,
+      "step": 8743
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.00010465782551457037,
+      "loss": 1.0328,
+      "step": 8744
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.00010464039303118112,
+      "loss": 0.9928,
+      "step": 8745
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.00010462296040646879,
+      "loss": 0.7878,
+      "step": 8746
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.00010460552764096436,
+      "loss": 0.9494,
+      "step": 8747
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.00010458809473519869,
+      "loss": 0.9166,
+      "step": 8748
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.00010457066168970269,
+      "loss": 1.0025,
+      "step": 8749
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.00010455322850500737,
+      "loss": 1.0359,
+      "step": 8750
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.00010453579518164354,
+      "loss": 0.9147,
+      "step": 8751
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.00010451836172014219,
+      "loss": 0.9174,
+      "step": 8752
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.00010450092812103427,
+      "loss": 0.8263,
+      "step": 8753
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.0001044834943848507,
+      "loss": 0.8868,
+      "step": 8754
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.00010446606051212242,
+      "loss": 0.8486,
+      "step": 8755
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.00010444862650338041,
+      "loss": 0.9472,
+      "step": 8756
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.00010443119235915557,
+      "loss": 0.9441,
+      "step": 8757
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.00010441375807997888,
+      "loss": 1.0776,
+      "step": 8758
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.00010439632366638133,
+      "loss": 0.8647,
+      "step": 8759
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.00010437888911889382,
+      "loss": 1.019,
+      "step": 8760
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.00010436145443804741,
+      "loss": 0.9062,
+      "step": 8761
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.00010434401962437297,
+      "loss": 0.8299,
+      "step": 8762
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.00010432658467840152,
+      "loss": 0.9849,
+      "step": 8763
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.00010430914960066408,
+      "loss": 0.8572,
+      "step": 8764
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.00010429171439169159,
+      "loss": 0.8127,
+      "step": 8765
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.00010427427905201505,
+      "loss": 0.9509,
+      "step": 8766
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.00010425684358216548,
+      "loss": 0.9202,
+      "step": 8767
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.00010423940798267382,
+      "loss": 0.9505,
+      "step": 8768
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.0001042219722540711,
+      "loss": 0.9018,
+      "step": 8769
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.00010420453639688834,
+      "loss": 0.8676,
+      "step": 8770
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.00010418710041165651,
+      "loss": 0.9628,
+      "step": 8771
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.0001041696642989067,
+      "loss": 0.7837,
+      "step": 8772
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.00010415222805916984,
+      "loss": 0.9622,
+      "step": 8773
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.00010413479169297698,
+      "loss": 0.9668,
+      "step": 8774
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.00010411735520085914,
+      "loss": 0.8129,
+      "step": 8775
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.00010409991858334736,
+      "loss": 0.945,
+      "step": 8776
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.00010408248184097269,
+      "loss": 1.0637,
+      "step": 8777
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.0001040650449742661,
+      "loss": 0.7852,
+      "step": 8778
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.00010404760798375872,
+      "loss": 0.9067,
+      "step": 8779
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.00010403017086998153,
+      "loss": 0.8442,
+      "step": 8780
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.00010401273363346557,
+      "loss": 0.9792,
+      "step": 8781
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.00010399529627474193,
+      "loss": 0.8966,
+      "step": 8782
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.00010397785879434163,
+      "loss": 0.9381,
+      "step": 8783
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.00010396042119279577,
+      "loss": 0.9608,
+      "step": 8784
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.00010394298347063538,
+      "loss": 0.9733,
+      "step": 8785
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.00010392554562839151,
+      "loss": 0.8774,
+      "step": 8786
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.00010390810766659527,
+      "loss": 0.7602,
+      "step": 8787
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.00010389066958577772,
+      "loss": 0.9103,
+      "step": 8788
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.00010387323138646992,
+      "loss": 1.0784,
+      "step": 8789
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.00010385579306920296,
+      "loss": 0.902,
+      "step": 8790
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.00010383835463450793,
+      "loss": 0.9136,
+      "step": 8791
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.00010382091608291591,
+      "loss": 0.9098,
+      "step": 8792
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.00010380347741495797,
+      "loss": 1.0726,
+      "step": 8793
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.00010378603863116525,
+      "loss": 0.9034,
+      "step": 8794
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.0001037685997320688,
+      "loss": 0.9081,
+      "step": 8795
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.00010375116071819979,
+      "loss": 0.9317,
+      "step": 8796
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.00010373372159008923,
+      "loss": 0.9074,
+      "step": 8797
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.0001037162823482683,
+      "loss": 0.9304,
+      "step": 8798
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.00010369884299326807,
+      "loss": 0.8975,
+      "step": 8799
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.00010368140352561968,
+      "loss": 0.8664,
+      "step": 8800
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.00010366396394585425,
+      "loss": 1.0027,
+      "step": 8801
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.00010364652425450289,
+      "loss": 0.9209,
+      "step": 8802
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.00010362908445209673,
+      "loss": 0.9298,
+      "step": 8803
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.00010361164453916692,
+      "loss": 0.8842,
+      "step": 8804
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.00010359420451624453,
+      "loss": 1.011,
+      "step": 8805
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.00010357676438386074,
+      "loss": 1.2135,
+      "step": 8806
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.00010355932414254675,
+      "loss": 0.9061,
+      "step": 8807
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.00010354188379283356,
+      "loss": 0.8335,
+      "step": 8808
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.00010352444333525241,
+      "loss": 0.9413,
+      "step": 8809
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.00010350700277033446,
+      "loss": 0.9337,
+      "step": 8810
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.0001034895620986108,
+      "loss": 0.8216,
+      "step": 8811
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.00010347212132061264,
+      "loss": 0.9139,
+      "step": 8812
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.00010345468043687112,
+      "loss": 0.9044,
+      "step": 8813
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.00010343723944791736,
+      "loss": 0.9479,
+      "step": 8814
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.00010341979835428262,
+      "loss": 0.7956,
+      "step": 8815
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.00010340235715649795,
+      "loss": 0.8866,
+      "step": 8816
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.0001033849158550946,
+      "loss": 0.9682,
+      "step": 8817
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.00010336747445060375,
+      "loss": 0.7496,
+      "step": 8818
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.00010335003294355657,
+      "loss": 0.987,
+      "step": 8819
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.00010333259133448417,
+      "loss": 0.8733,
+      "step": 8820
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.00010331514962391781,
+      "loss": 0.8782,
+      "step": 8821
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.00010329770781238865,
+      "loss": 0.946,
+      "step": 8822
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.00010328026590042792,
+      "loss": 1.0126,
+      "step": 8823
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.00010326282388856673,
+      "loss": 0.874,
+      "step": 8824
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.00010324538177733635,
+      "loss": 0.864,
+      "step": 8825
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.00010322793956726797,
+      "loss": 0.8668,
+      "step": 8826
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.00010321049725889274,
+      "loss": 0.9618,
+      "step": 8827
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.00010319305485274191,
+      "loss": 0.9031,
+      "step": 8828
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.0001031756123493467,
+      "loss": 1.092,
+      "step": 8829
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.00010315816974923828,
+      "loss": 0.8849,
+      "step": 8830
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.00010314072705294791,
+      "loss": 0.8791,
+      "step": 8831
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.00010312328426100672,
+      "loss": 0.9297,
+      "step": 8832
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.00010310584137394603,
+      "loss": 0.8966,
+      "step": 8833
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.00010308839839229703,
+      "loss": 0.8286,
+      "step": 8834
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.00010307095531659092,
+      "loss": 0.8695,
+      "step": 8835
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.00010305351214735896,
+      "loss": 0.8129,
+      "step": 8836
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.00010303606888513237,
+      "loss": 1.0395,
+      "step": 8837
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.00010301862553044235,
+      "loss": 1.0156,
+      "step": 8838
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.00010300118208382022,
+      "loss": 0.9431,
+      "step": 8839
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.00010298373854579712,
+      "loss": 0.9391,
+      "step": 8840
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.00010296629491690435,
+      "loss": 0.8029,
+      "step": 8841
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.0001029488511976732,
+      "loss": 0.96,
+      "step": 8842
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.00010293140738863478,
+      "loss": 0.8344,
+      "step": 8843
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.00010291396349032048,
+      "loss": 0.9171,
+      "step": 8844
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.0001028965195032615,
+      "loss": 0.8226,
+      "step": 8845
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.00010287907542798905,
+      "loss": 0.8314,
+      "step": 8846
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.00010286163126503447,
+      "loss": 0.9095,
+      "step": 8847
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.000102844187014929,
+      "loss": 0.9891,
+      "step": 8848
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.00010282674267820384,
+      "loss": 0.8806,
+      "step": 8849
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.00010280929825539034,
+      "loss": 0.9038,
+      "step": 8850
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.00010279185374701972,
+      "loss": 0.7275,
+      "step": 8851
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.00010277440915362327,
+      "loss": 0.8704,
+      "step": 8852
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.00010275696447573229,
+      "loss": 1.0471,
+      "step": 8853
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.00010273951971387798,
+      "loss": 0.9038,
+      "step": 8854
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.00010272207486859168,
+      "loss": 0.9284,
+      "step": 8855
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.0001027046299404047,
+      "loss": 0.8987,
+      "step": 8856
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.00010268718492984825,
+      "loss": 0.9952,
+      "step": 8857
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.00010266973983745367,
+      "loss": 0.8715,
+      "step": 8858
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.00010265229466375225,
+      "loss": 0.9103,
+      "step": 8859
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.00010263484940927524,
+      "loss": 0.9314,
+      "step": 8860
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.00010261740407455398,
+      "loss": 1.008,
+      "step": 8861
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.00010259995866011971,
+      "loss": 1.0009,
+      "step": 8862
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.00010258251316650379,
+      "loss": 0.9414,
+      "step": 8863
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.00010256506759423749,
+      "loss": 0.9627,
+      "step": 8864
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.00010254762194385215,
+      "loss": 0.8354,
+      "step": 8865
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.00010253017621587902,
+      "loss": 0.905,
+      "step": 8866
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.00010251273041084948,
+      "loss": 0.8813,
+      "step": 8867
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.00010249528452929475,
+      "loss": 0.8997,
+      "step": 8868
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.00010247783857174622,
+      "loss": 1.095,
+      "step": 8869
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.00010246039253873518,
+      "loss": 0.8981,
+      "step": 8870
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.00010244294643079295,
+      "loss": 0.8694,
+      "step": 8871
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.00010242550024845088,
+      "loss": 0.7968,
+      "step": 8872
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.00010240805399224023,
+      "loss": 0.8985,
+      "step": 8873
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.00010239060766269233,
+      "loss": 0.9301,
+      "step": 8874
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.00010237316126033858,
+      "loss": 0.9397,
+      "step": 8875
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.00010235571478571027,
+      "loss": 1.0654,
+      "step": 8876
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.00010233826823933872,
+      "loss": 0.8913,
+      "step": 8877
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.00010232082162175527,
+      "loss": 0.8905,
+      "step": 8878
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.00010230337493349125,
+      "loss": 0.7676,
+      "step": 8879
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.000102285928175078,
+      "loss": 0.9405,
+      "step": 8880
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.00010226848134704687,
+      "loss": 0.9016,
+      "step": 8881
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.00010225103444992923,
+      "loss": 0.9562,
+      "step": 8882
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.00010223358748425637,
+      "loss": 0.9183,
+      "step": 8883
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.00010221614045055967,
+      "loss": 0.7133,
+      "step": 8884
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.00010219869334937047,
+      "loss": 1.0512,
+      "step": 8885
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.00010218124618122012,
+      "loss": 0.9382,
+      "step": 8886
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.00010216379894663997,
+      "loss": 0.7644,
+      "step": 8887
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.0001021463516461614,
+      "loss": 0.8005,
+      "step": 8888
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.00010212890428031572,
+      "loss": 1.0155,
+      "step": 8889
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.00010211145684963432,
+      "loss": 1.0313,
+      "step": 8890
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.00010209400935464857,
+      "loss": 0.9686,
+      "step": 8891
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.0001020765617958898,
+      "loss": 0.8489,
+      "step": 8892
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.00010205911417388942,
+      "loss": 0.8879,
+      "step": 8893
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.00010204166648917877,
+      "loss": 0.823,
+      "step": 8894
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.0001020242187422892,
+      "loss": 0.8581,
+      "step": 8895
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.00010200677093375212,
+      "loss": 0.8889,
+      "step": 8896
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.00010198932306409887,
+      "loss": 0.9346,
+      "step": 8897
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.00010197187513386084,
+      "loss": 0.9467,
+      "step": 8898
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.00010195442714356942,
+      "loss": 0.9087,
+      "step": 8899
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.00010193697909375598,
+      "loss": 0.972,
+      "step": 8900
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.00010191953098495183,
+      "loss": 0.9281,
+      "step": 8901
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.00010190208281768847,
+      "loss": 0.9398,
+      "step": 8902
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.00010188463459249722,
+      "loss": 0.9434,
+      "step": 8903
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.00010186718630990948,
+      "loss": 0.8389,
+      "step": 8904
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.00010184973797045663,
+      "loss": 1.1125,
+      "step": 8905
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.00010183228957467004,
+      "loss": 0.7196,
+      "step": 8906
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.00010181484112308112,
+      "loss": 0.932,
+      "step": 8907
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.00010179739261622126,
+      "loss": 0.8842,
+      "step": 8908
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.00010177994405462186,
+      "loss": 0.8694,
+      "step": 8909
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.00010176249543881433,
+      "loss": 0.9535,
+      "step": 8910
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.00010174504676933004,
+      "loss": 0.935,
+      "step": 8911
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.00010172759804670037,
+      "loss": 0.9366,
+      "step": 8912
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.00010171014927145677,
+      "loss": 0.7595,
+      "step": 8913
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.0001016927004441306,
+      "loss": 0.8947,
+      "step": 8914
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.00010167525156525331,
+      "loss": 0.9296,
+      "step": 8915
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.00010165780263535624,
+      "loss": 0.8397,
+      "step": 8916
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.00010164035365497089,
+      "loss": 0.8404,
+      "step": 8917
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.00010162290462462858,
+      "loss": 0.8634,
+      "step": 8918
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.00010160545554486075,
+      "loss": 0.7121,
+      "step": 8919
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.0001015880064161988,
+      "loss": 0.7818,
+      "step": 8920
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.00010157055723917417,
+      "loss": 0.8785,
+      "step": 8921
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.00010155310801431828,
+      "loss": 0.7945,
+      "step": 8922
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.00010153565874216251,
+      "loss": 0.951,
+      "step": 8923
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.00010151820942323829,
+      "loss": 0.901,
+      "step": 8924
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.00010150076005807702,
+      "loss": 0.8903,
+      "step": 8925
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.00010148331064721019,
+      "loss": 0.9209,
+      "step": 8926
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.00010146586119116914,
+      "loss": 0.8964,
+      "step": 8927
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.00010144841169048534,
+      "loss": 0.9113,
+      "step": 8928
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.00010143096214569021,
+      "loss": 0.8342,
+      "step": 8929
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.00010141351255731513,
+      "loss": 0.9623,
+      "step": 8930
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.00010139606292589156,
+      "loss": 0.8695,
+      "step": 8931
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.00010137861325195097,
+      "loss": 0.899,
+      "step": 8932
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.00010136116353602474,
+      "loss": 0.9163,
+      "step": 8933
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.00010134371377864431,
+      "loss": 0.9299,
+      "step": 8934
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.00010132626398034111,
+      "loss": 0.8963,
+      "step": 8935
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.00010130881414164654,
+      "loss": 0.9287,
+      "step": 8936
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.00010129136426309213,
+      "loss": 0.8341,
+      "step": 8937
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.00010127391434520921,
+      "loss": 1.0293,
+      "step": 8938
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.0001012564643885293,
+      "loss": 0.9474,
+      "step": 8939
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.00010123901439358379,
+      "loss": 1.026,
+      "step": 8940
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.00010122156436090409,
+      "loss": 0.861,
+      "step": 8941
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.00010120411429102172,
+      "loss": 0.9347,
+      "step": 8942
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.00010118666418446805,
+      "loss": 0.8627,
+      "step": 8943
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.00010116921404177455,
+      "loss": 0.8369,
+      "step": 8944
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.00010115176386347274,
+      "loss": 0.9306,
+      "step": 8945
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.00010113431365009392,
+      "loss": 0.9485,
+      "step": 8946
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.00010111686340216961,
+      "loss": 0.9321,
+      "step": 8947
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.00010109941312023128,
+      "loss": 0.8378,
+      "step": 8948
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.00010108196280481033,
+      "loss": 0.7976,
+      "step": 8949
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.00010106451245643827,
+      "loss": 1.0207,
+      "step": 8950
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.00010104706207564649,
+      "loss": 0.9651,
+      "step": 8951
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.00010102961166296642,
+      "loss": 0.8482,
+      "step": 8952
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.00010101216121892962,
+      "loss": 0.8771,
+      "step": 8953
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.00010099471074406743,
+      "loss": 0.9263,
+      "step": 8954
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.00010097726023891136,
+      "loss": 1.076,
+      "step": 8955
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.00010095980970399288,
+      "loss": 0.7691,
+      "step": 8956
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.0001009423591398434,
+      "loss": 0.9111,
+      "step": 8957
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.00010092490854699439,
+      "loss": 1.0883,
+      "step": 8958
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.00010090745792597733,
+      "loss": 0.8675,
+      "step": 8959
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.00010089000727732365,
+      "loss": 0.909,
+      "step": 8960
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.00010087255660156485,
+      "loss": 0.843,
+      "step": 8961
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.00010085510589923233,
+      "loss": 0.9026,
+      "step": 8962
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.00010083765517085759,
+      "loss": 0.9348,
+      "step": 8963
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.0001008202044169721,
+      "loss": 0.9049,
+      "step": 8964
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.0001008027536381073,
+      "loss": 0.8905,
+      "step": 8965
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.00010078530283479464,
+      "loss": 1.0125,
+      "step": 8966
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.00010076785200756562,
+      "loss": 1.0021,
+      "step": 8967
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.00010075040115695169,
+      "loss": 0.8137,
+      "step": 8968
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.0001007329502834843,
+      "loss": 0.9321,
+      "step": 8969
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.00010071549938769493,
+      "loss": 0.8697,
+      "step": 8970
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.00010069804847011504,
+      "loss": 0.9592,
+      "step": 8971
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.00010068059753127612,
+      "loss": 0.8017,
+      "step": 8972
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.00010066314657170961,
+      "loss": 0.793,
+      "step": 8973
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.00010064569559194698,
+      "loss": 0.9687,
+      "step": 8974
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.00010062824459251974,
+      "loss": 0.9147,
+      "step": 8975
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.00010061079357395929,
+      "loss": 0.8281,
+      "step": 8976
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.00010059334253679715,
+      "loss": 0.9542,
+      "step": 8977
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.00010057589148156479,
+      "loss": 1.0144,
+      "step": 8978
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.00010055844040879366,
+      "loss": 0.9208,
+      "step": 8979
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.00010054098931901523,
+      "loss": 0.9563,
+      "step": 8980
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.00010052353821276102,
+      "loss": 0.952,
+      "step": 8981
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.00010050608709056241,
+      "loss": 0.9214,
+      "step": 8982
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.00010048863595295099,
+      "loss": 0.7481,
+      "step": 8983
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.00010047118480045815,
+      "loss": 0.9581,
+      "step": 8984
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.00010045373363361539,
+      "loss": 0.8519,
+      "step": 8985
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.0001004362824529542,
+      "loss": 0.9326,
+      "step": 8986
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.00010041883125900599,
+      "loss": 0.9131,
+      "step": 8987
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.00010040138005230235,
+      "loss": 0.8939,
+      "step": 8988
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.00010038392883337464,
+      "loss": 0.9036,
+      "step": 8989
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.00010036647760275439,
+      "loss": 0.8501,
+      "step": 8990
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.00010034902636097312,
+      "loss": 0.9254,
+      "step": 8991
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.00010033157510856224,
+      "loss": 0.9409,
+      "step": 8992
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.0001003141238460532,
+      "loss": 0.8696,
+      "step": 8993
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.0001002966725739776,
+      "loss": 0.8675,
+      "step": 8994
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.0001002792212928668,
+      "loss": 0.8698,
+      "step": 8995
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.00010026177000325234,
+      "loss": 0.9916,
+      "step": 8996
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.00010024431870566568,
+      "loss": 0.8651,
+      "step": 8997
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.0001002268674006383,
+      "loss": 0.8656,
+      "step": 8998
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.00010020941608870169,
+      "loss": 0.8664,
+      "step": 8999
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.00010019196477038728,
+      "loss": 0.8973,
+      "step": 9000
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.0001001745134462266,
+      "loss": 0.9508,
+      "step": 9001
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.00010015706211675114,
+      "loss": 0.8655,
+      "step": 9002
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.00010013961078249237,
+      "loss": 0.7916,
+      "step": 9003
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.00010012215944398171,
+      "loss": 0.9255,
+      "step": 9004
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.00010010470810175073,
+      "loss": 0.8755,
+      "step": 9005
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.00010008725675633084,
+      "loss": 0.9684,
+      "step": 9006
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.00010006980540825357,
+      "loss": 0.8637,
+      "step": 9007
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.00010005235405805033,
+      "loss": 0.9232,
+      "step": 9008
+    },
+    {
+      "epoch": 1.98,
+      "eval_loss": 0.5695223212242126,
+      "eval_runtime": 321.9222,
+      "eval_samples_per_second": 56.346,
+      "eval_steps_per_second": 28.175,
+      "step": 9008
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.00010003490270625271,
+      "loss": 0.9345,
+      "step": 9009
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 0.00010001745135339211,
+      "loss": 0.9367,
+      "step": 9010
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 0.0001,
+      "loss": 0.863,
+      "step": 9011
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 9.998254864660793e-05,
+      "loss": 0.9228,
+      "step": 9012
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 9.996509729374731e-05,
+      "loss": 1.016,
+      "step": 9013
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 9.994764594194968e-05,
+      "loss": 0.9231,
+      "step": 9014
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 9.993019459174646e-05,
+      "loss": 0.8454,
+      "step": 9015
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 9.99127432436692e-05,
+      "loss": 0.8749,
+      "step": 9016
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 9.989529189824932e-05,
+      "loss": 0.8484,
+      "step": 9017
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 9.987784055601831e-05,
+      "loss": 0.8858,
+      "step": 9018
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 9.98603892175077e-05,
+      "loss": 0.8304,
+      "step": 9019
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 9.984293788324886e-05,
+      "loss": 0.8533,
+      "step": 9020
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 9.982548655377341e-05,
+      "loss": 0.861,
+      "step": 9021
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 9.980803522961272e-05,
+      "loss": 0.9908,
+      "step": 9022
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 9.979058391129835e-05,
+      "loss": 0.9315,
+      "step": 9023
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 9.977313259936176e-05,
+      "loss": 0.8352,
+      "step": 9024
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 9.975568129433434e-05,
+      "loss": 0.9071,
+      "step": 9025
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 9.973822999674768e-05,
+      "loss": 0.8594,
+      "step": 9026
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 9.97207787071332e-05,
+      "loss": 0.8148,
+      "step": 9027
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 9.970332742602242e-05,
+      "loss": 0.9138,
+      "step": 9028
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 9.96858761539468e-05,
+      "loss": 0.81,
+      "step": 9029
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 9.96684248914378e-05,
+      "loss": 0.9006,
+      "step": 9030
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 9.965097363902692e-05,
+      "loss": 1.0557,
+      "step": 9031
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 9.96335223972456e-05,
+      "loss": 1.0217,
+      "step": 9032
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 9.961607116662538e-05,
+      "loss": 0.9754,
+      "step": 9033
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 9.959861994769767e-05,
+      "loss": 0.8383,
+      "step": 9034
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 9.958116874099402e-05,
+      "loss": 0.8984,
+      "step": 9035
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 9.956371754704585e-05,
+      "loss": 0.8897,
+      "step": 9036
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 9.954626636638464e-05,
+      "loss": 1.0287,
+      "step": 9037
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 9.952881519954188e-05,
+      "loss": 0.942,
+      "step": 9038
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 9.951136404704903e-05,
+      "loss": 0.927,
+      "step": 9039
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 9.94939129094376e-05,
+      "loss": 0.8032,
+      "step": 9040
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 9.947646178723903e-05,
+      "loss": 0.8259,
+      "step": 9041
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 9.94590106809848e-05,
+      "loss": 0.8638,
+      "step": 9042
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 9.944155959120638e-05,
+      "loss": 0.9235,
+      "step": 9043
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 9.942410851843521e-05,
+      "loss": 1.0106,
+      "step": 9044
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 9.940665746320287e-05,
+      "loss": 0.8878,
+      "step": 9045
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 9.938920642604075e-05,
+      "loss": 0.8538,
+      "step": 9046
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 9.937175540748028e-05,
+      "loss": 0.9699,
+      "step": 9047
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 9.935430440805304e-05,
+      "loss": 0.8147,
+      "step": 9048
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 9.93368534282904e-05,
+      "loss": 0.887,
+      "step": 9049
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 9.93194024687239e-05,
+      "loss": 0.819,
+      "step": 9050
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 9.930195152988494e-05,
+      "loss": 0.8187,
+      "step": 9051
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 9.928450061230509e-05,
+      "loss": 0.952,
+      "step": 9052
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 9.926704971651573e-05,
+      "loss": 0.869,
+      "step": 9053
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 9.924959884304834e-05,
+      "loss": 1.0047,
+      "step": 9054
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 9.92321479924344e-05,
+      "loss": 0.7705,
+      "step": 9055
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 9.921469716520536e-05,
+      "loss": 0.8986,
+      "step": 9056
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 9.919724636189272e-05,
+      "loss": 0.9143,
+      "step": 9057
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 9.917979558302793e-05,
+      "loss": 0.9281,
+      "step": 9058
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 9.916234482914242e-05,
+      "loss": 0.8355,
+      "step": 9059
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 9.914489410076769e-05,
+      "loss": 0.7693,
+      "step": 9060
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 9.912744339843517e-05,
+      "loss": 0.8777,
+      "step": 9061
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 9.910999272267636e-05,
+      "loss": 0.9552,
+      "step": 9062
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 9.909254207402269e-05,
+      "loss": 0.7758,
+      "step": 9063
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 9.907509145300562e-05,
+      "loss": 0.9642,
+      "step": 9064
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 9.905764086015663e-05,
+      "loss": 0.8835,
+      "step": 9065
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 9.904019029600713e-05,
+      "loss": 0.8994,
+      "step": 9066
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 9.902273976108865e-05,
+      "loss": 0.8154,
+      "step": 9067
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 9.900528925593257e-05,
+      "loss": 0.9441,
+      "step": 9068
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 9.89878387810704e-05,
+      "loss": 0.905,
+      "step": 9069
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 9.89703883370336e-05,
+      "loss": 0.9829,
+      "step": 9070
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 9.895293792435354e-05,
+      "loss": 0.9056,
+      "step": 9071
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 9.893548754356177e-05,
+      "loss": 0.8696,
+      "step": 9072
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 9.891803719518967e-05,
+      "loss": 0.9187,
+      "step": 9073
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 9.890058687976875e-05,
+      "loss": 1.0014,
+      "step": 9074
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 9.888313659783041e-05,
+      "loss": 0.7922,
+      "step": 9075
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 9.88656863499061e-05,
+      "loss": 0.9941,
+      "step": 9076
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 9.884823613652732e-05,
+      "loss": 0.8399,
+      "step": 9077
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 9.883078595822544e-05,
+      "loss": 0.9536,
+      "step": 9078
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 9.881333581553198e-05,
+      "loss": 0.9282,
+      "step": 9079
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 9.87958857089783e-05,
+      "loss": 0.9111,
+      "step": 9080
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 9.877843563909593e-05,
+      "loss": 0.7613,
+      "step": 9081
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 9.876098560641626e-05,
+      "loss": 0.7953,
+      "step": 9082
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 9.874353561147074e-05,
+      "loss": 0.9848,
+      "step": 9083
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 9.872608565479081e-05,
+      "loss": 0.8865,
+      "step": 9084
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 9.870863573690788e-05,
+      "loss": 0.8518,
+      "step": 9085
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 9.869118585835347e-05,
+      "loss": 0.8968,
+      "step": 9086
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 9.867373601965894e-05,
+      "loss": 0.95,
+      "step": 9087
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 9.865628622135572e-05,
+      "loss": 0.8017,
+      "step": 9088
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 9.863883646397529e-05,
+      "loss": 0.9573,
+      "step": 9089
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 9.862138674804902e-05,
+      "loss": 0.6905,
+      "step": 9090
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 9.860393707410844e-05,
+      "loss": 1.101,
+      "step": 9091
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 9.858648744268491e-05,
+      "loss": 0.9559,
+      "step": 9092
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 9.856903785430981e-05,
+      "loss": 0.8199,
+      "step": 9093
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 9.85515883095147e-05,
+      "loss": 0.8676,
+      "step": 9094
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 9.853413880883087e-05,
+      "loss": 0.7658,
+      "step": 9095
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 9.851668935278983e-05,
+      "loss": 0.9297,
+      "step": 9096
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 9.849923994192296e-05,
+      "loss": 0.9278,
+      "step": 9097
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 9.848179057676174e-05,
+      "loss": 0.8535,
+      "step": 9098
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 9.846434125783752e-05,
+      "loss": 0.8753,
+      "step": 9099
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 9.844689198568174e-05,
+      "loss": 0.9684,
+      "step": 9100
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 9.842944276082584e-05,
+      "loss": 0.8567,
+      "step": 9101
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 9.84119935838012e-05,
+      "loss": 0.8823,
+      "step": 9102
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 9.839454445513928e-05,
+      "loss": 1.1022,
+      "step": 9103
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 9.837709537537147e-05,
+      "loss": 0.8651,
+      "step": 9104
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 9.835964634502915e-05,
+      "loss": 0.8916,
+      "step": 9105
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 9.834219736464377e-05,
+      "loss": 0.9802,
+      "step": 9106
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 9.83247484347467e-05,
+      "loss": 0.8004,
+      "step": 9107
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 9.830729955586941e-05,
+      "loss": 0.8151,
+      "step": 9108
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 9.828985072854328e-05,
+      "loss": 0.8177,
+      "step": 9109
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 9.827240195329965e-05,
+      "loss": 0.985,
+      "step": 9110
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 9.825495323067001e-05,
+      "loss": 0.8548,
+      "step": 9111
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 9.823750456118568e-05,
+      "loss": 0.8288,
+      "step": 9112
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 9.822005594537816e-05,
+      "loss": 0.8209,
+      "step": 9113
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 9.820260738377874e-05,
+      "loss": 0.8584,
+      "step": 9114
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 9.818515887691889e-05,
+      "loss": 0.966,
+      "step": 9115
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 9.816771042533001e-05,
+      "loss": 0.923,
+      "step": 9116
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 9.81502620295434e-05,
+      "loss": 0.8221,
+      "step": 9117
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 9.813281369009054e-05,
+      "loss": 0.9635,
+      "step": 9118
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 9.811536540750278e-05,
+      "loss": 0.7983,
+      "step": 9119
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 9.809791718231154e-05,
+      "loss": 0.8849,
+      "step": 9120
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 9.808046901504818e-05,
+      "loss": 0.8759,
+      "step": 9121
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 9.806302090624406e-05,
+      "loss": 0.9175,
+      "step": 9122
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 9.804557285643061e-05,
+      "loss": 0.9725,
+      "step": 9123
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 9.802812486613917e-05,
+      "loss": 0.9348,
+      "step": 9124
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 9.801067693590115e-05,
+      "loss": 1.0382,
+      "step": 9125
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 9.79932290662479e-05,
+      "loss": 0.8879,
+      "step": 9126
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 9.797578125771082e-05,
+      "loss": 0.8357,
+      "step": 9127
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 9.795833351082128e-05,
+      "loss": 0.7942,
+      "step": 9128
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 9.794088582611061e-05,
+      "loss": 0.8717,
+      "step": 9129
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 9.792343820411022e-05,
+      "loss": 0.9924,
+      "step": 9130
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 9.790599064535143e-05,
+      "loss": 0.8021,
+      "step": 9131
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 9.78885431503657e-05,
+      "loss": 0.994,
+      "step": 9132
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 9.787109571968432e-05,
+      "loss": 0.989,
+      "step": 9133
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 9.785364835383862e-05,
+      "loss": 0.9127,
+      "step": 9134
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 9.783620105336007e-05,
+      "loss": 1.0566,
+      "step": 9135
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 9.781875381877989e-05,
+      "loss": 0.972,
+      "step": 9136
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 9.780130665062954e-05,
+      "loss": 0.8027,
+      "step": 9137
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 9.778385954944037e-05,
+      "loss": 0.8605,
+      "step": 9138
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 9.776641251574364e-05,
+      "loss": 0.9515,
+      "step": 9139
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 9.774896555007079e-05,
+      "loss": 0.9309,
+      "step": 9140
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 9.773151865295312e-05,
+      "loss": 0.9563,
+      "step": 9141
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 9.771407182492201e-05,
+      "loss": 1.0432,
+      "step": 9142
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 9.769662506650876e-05,
+      "loss": 0.8185,
+      "step": 9143
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 9.767917837824475e-05,
+      "loss": 0.8354,
+      "step": 9144
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 9.766173176066131e-05,
+      "loss": 0.9198,
+      "step": 9145
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 9.764428521428974e-05,
+      "loss": 0.8984,
+      "step": 9146
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 9.762683873966143e-05,
+      "loss": 0.9391,
+      "step": 9147
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 9.760939233730765e-05,
+      "loss": 0.8243,
+      "step": 9148
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 9.75919460077598e-05,
+      "loss": 0.9809,
+      "step": 9149
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 9.757449975154917e-05,
+      "loss": 0.8454,
+      "step": 9150
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 9.755705356920705e-05,
+      "loss": 0.9816,
+      "step": 9151
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 9.753960746126483e-05,
+      "loss": 0.9393,
+      "step": 9152
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 9.752216142825379e-05,
+      "loss": 0.7671,
+      "step": 9153
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 9.750471547070527e-05,
+      "loss": 0.8859,
+      "step": 9154
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 9.748726958915057e-05,
+      "loss": 0.8595,
+      "step": 9155
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 9.7469823784121e-05,
+      "loss": 0.9051,
+      "step": 9156
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 9.74523780561479e-05,
+      "loss": 0.724,
+      "step": 9157
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 9.74349324057625e-05,
+      "loss": 0.8953,
+      "step": 9158
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 9.741748683349624e-05,
+      "loss": 0.9,
+      "step": 9159
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 9.740004133988029e-05,
+      "loss": 0.9105,
+      "step": 9160
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 9.738259592544606e-05,
+      "loss": 0.8502,
+      "step": 9161
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 9.736515059072481e-05,
+      "loss": 0.7966,
+      "step": 9162
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 9.734770533624777e-05,
+      "loss": 0.8204,
+      "step": 9163
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 9.733026016254634e-05,
+      "loss": 0.9315,
+      "step": 9164
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 9.731281507015175e-05,
+      "loss": 0.7864,
+      "step": 9165
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 9.729537005959531e-05,
+      "loss": 0.8945,
+      "step": 9166
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 9.727792513140833e-05,
+      "loss": 0.7707,
+      "step": 9167
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 9.726048028612203e-05,
+      "loss": 0.8332,
+      "step": 9168
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 9.724303552426776e-05,
+      "loss": 0.8459,
+      "step": 9169
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 9.722559084637674e-05,
+      "loss": 0.8752,
+      "step": 9170
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 9.72081462529803e-05,
+      "loss": 0.8492,
+      "step": 9171
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 9.719070174460968e-05,
+      "loss": 0.9444,
+      "step": 9172
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 9.717325732179618e-05,
+      "loss": 0.8797,
+      "step": 9173
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 9.715581298507106e-05,
+      "loss": 0.8689,
+      "step": 9174
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 9.713836873496555e-05,
+      "loss": 0.9615,
+      "step": 9175
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 9.712092457201097e-05,
+      "loss": 0.9741,
+      "step": 9176
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 9.710348049673852e-05,
+      "loss": 0.7806,
+      "step": 9177
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 9.708603650967956e-05,
+      "loss": 0.8842,
+      "step": 9178
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 9.706859261136525e-05,
+      "loss": 0.8061,
+      "step": 9179
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 9.705114880232684e-05,
+      "loss": 0.8502,
+      "step": 9180
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 9.703370508309569e-05,
+      "loss": 0.7806,
+      "step": 9181
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 9.701626145420289e-05,
+      "loss": 0.8474,
+      "step": 9182
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 9.699881791617982e-05,
+      "loss": 0.8375,
+      "step": 9183
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 9.698137446955768e-05,
+      "loss": 0.8834,
+      "step": 9184
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 9.696393111486766e-05,
+      "loss": 0.9122,
+      "step": 9185
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 9.694648785264106e-05,
+      "loss": 0.8359,
+      "step": 9186
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 9.692904468340909e-05,
+      "loss": 0.8599,
+      "step": 9187
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 9.6911601607703e-05,
+      "loss": 0.8689,
+      "step": 9188
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 9.689415862605397e-05,
+      "loss": 0.9361,
+      "step": 9189
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 9.687671573899329e-05,
+      "loss": 0.8314,
+      "step": 9190
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 9.685927294705214e-05,
+      "loss": 0.7136,
+      "step": 9191
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 9.684183025076174e-05,
+      "loss": 0.8997,
+      "step": 9192
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 9.682438765065332e-05,
+      "loss": 0.9348,
+      "step": 9193
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 9.680694514725809e-05,
+      "loss": 0.8347,
+      "step": 9194
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 9.678950274110727e-05,
+      "loss": 0.9215,
+      "step": 9195
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 9.677206043273208e-05,
+      "loss": 0.834,
+      "step": 9196
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 9.675461822266366e-05,
+      "loss": 0.7393,
+      "step": 9197
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 9.673717611143329e-05,
+      "loss": 0.8085,
+      "step": 9198
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 9.671973409957211e-05,
+      "loss": 0.8424,
+      "step": 9199
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 9.670229218761137e-05,
+      "loss": 0.862,
+      "step": 9200
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 9.668485037608223e-05,
+      "loss": 0.964,
+      "step": 9201
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 9.666740866551586e-05,
+      "loss": 0.8139,
+      "step": 9202
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 9.664996705644348e-05,
+      "loss": 0.9623,
+      "step": 9203
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 9.663252554939624e-05,
+      "loss": 0.864,
+      "step": 9204
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 9.661508414490542e-05,
+      "loss": 0.771,
+      "step": 9205
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 9.659764284350205e-05,
+      "loss": 0.781,
+      "step": 9206
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 9.658020164571742e-05,
+      "loss": 0.8127,
+      "step": 9207
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 9.656276055208267e-05,
+      "loss": 0.9138,
+      "step": 9208
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 9.654531956312891e-05,
+      "loss": 0.9822,
+      "step": 9209
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 9.652787867938738e-05,
+      "loss": 0.9021,
+      "step": 9210
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 9.65104379013892e-05,
+      "loss": 0.817,
+      "step": 9211
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 9.649299722966555e-05,
+      "loss": 0.9172,
+      "step": 9212
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 9.64755566647476e-05,
+      "loss": 0.8459,
+      "step": 9213
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 9.645811620716645e-05,
+      "loss": 0.8567,
+      "step": 9214
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 9.64406758574533e-05,
+      "loss": 0.9847,
+      "step": 9215
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 9.642323561613925e-05,
+      "loss": 0.9021,
+      "step": 9216
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 9.64057954837555e-05,
+      "loss": 0.8848,
+      "step": 9217
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 9.63883554608331e-05,
+      "loss": 0.8567,
+      "step": 9218
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 9.637091554790328e-05,
+      "loss": 0.9186,
+      "step": 9219
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 9.635347574549713e-05,
+      "loss": 0.7722,
+      "step": 9220
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 9.633603605414576e-05,
+      "loss": 0.6716,
+      "step": 9221
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 9.631859647438034e-05,
+      "loss": 0.8354,
+      "step": 9222
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 9.630115700673193e-05,
+      "loss": 0.8964,
+      "step": 9223
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 9.628371765173174e-05,
+      "loss": 0.8917,
+      "step": 9224
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 9.626627840991081e-05,
+      "loss": 0.7382,
+      "step": 9225
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 9.624883928180024e-05,
+      "loss": 0.8865,
+      "step": 9226
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 9.623140026793122e-05,
+      "loss": 0.8283,
+      "step": 9227
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 9.621396136883475e-05,
+      "loss": 0.8829,
+      "step": 9228
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 9.619652258504204e-05,
+      "loss": 0.9099,
+      "step": 9229
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 9.617908391708414e-05,
+      "loss": 0.8639,
+      "step": 9230
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 9.616164536549207e-05,
+      "loss": 0.7575,
+      "step": 9231
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 9.614420693079706e-05,
+      "loss": 1.0355,
+      "step": 9232
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 9.612676861353009e-05,
+      "loss": 0.746,
+      "step": 9233
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 9.610933041422229e-05,
+      "loss": 0.7286,
+      "step": 9234
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 9.609189233340473e-05,
+      "loss": 0.8258,
+      "step": 9235
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 9.60744543716085e-05,
+      "loss": 0.8474,
+      "step": 9236
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 9.605701652936465e-05,
+      "loss": 0.6694,
+      "step": 9237
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 9.603957880720424e-05,
+      "loss": 0.8038,
+      "step": 9238
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 9.60221412056584e-05,
+      "loss": 0.9344,
+      "step": 9239
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 9.600470372525809e-05,
+      "loss": 0.8039,
+      "step": 9240
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 9.598726636653446e-05,
+      "loss": 0.9195,
+      "step": 9241
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 9.596982913001852e-05,
+      "loss": 0.7579,
+      "step": 9242
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 9.59523920162413e-05,
+      "loss": 0.9369,
+      "step": 9243
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 9.593495502573392e-05,
+      "loss": 0.9278,
+      "step": 9244
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 9.591751815902735e-05,
+      "loss": 0.8384,
+      "step": 9245
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 9.590008141665266e-05,
+      "loss": 0.9736,
+      "step": 9246
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 9.58826447991409e-05,
+      "loss": 0.8384,
+      "step": 9247
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 9.586520830702306e-05,
+      "loss": 0.9404,
+      "step": 9248
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 9.58477719408302e-05,
+      "loss": 0.9448,
+      "step": 9249
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 9.583033570109332e-05,
+      "loss": 0.8117,
+      "step": 9250
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 9.581289958834351e-05,
+      "loss": 0.9657,
+      "step": 9251
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 9.579546360311166e-05,
+      "loss": 0.8478,
+      "step": 9252
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 9.57780277459289e-05,
+      "loss": 0.8674,
+      "step": 9253
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 9.576059201732623e-05,
+      "loss": 0.8187,
+      "step": 9254
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 9.574315641783453e-05,
+      "loss": 0.8672,
+      "step": 9255
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 9.572572094798496e-05,
+      "loss": 0.9466,
+      "step": 9256
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 9.57082856083084e-05,
+      "loss": 0.9103,
+      "step": 9257
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 9.569085039933592e-05,
+      "loss": 0.8486,
+      "step": 9258
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 9.567341532159849e-05,
+      "loss": 0.9575,
+      "step": 9259
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 9.565598037562705e-05,
+      "loss": 0.807,
+      "step": 9260
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 9.563854556195264e-05,
+      "loss": 0.6905,
+      "step": 9261
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 9.562111088110619e-05,
+      "loss": 1.073,
+      "step": 9262
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 9.560367633361871e-05,
+      "loss": 0.877,
+      "step": 9263
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 9.558624192002113e-05,
+      "loss": 0.855,
+      "step": 9264
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 9.556880764084447e-05,
+      "loss": 0.8623,
+      "step": 9265
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 9.555137349661964e-05,
+      "loss": 0.7972,
+      "step": 9266
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 9.553393948787758e-05,
+      "loss": 0.7508,
+      "step": 9267
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 9.551650561514932e-05,
+      "loss": 0.9213,
+      "step": 9268
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 9.549907187896571e-05,
+      "loss": 0.7294,
+      "step": 9269
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 9.548163827985783e-05,
+      "loss": 0.839,
+      "step": 9270
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 9.546420481835651e-05,
+      "loss": 0.8432,
+      "step": 9271
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 9.544677149499265e-05,
+      "loss": 0.8462,
+      "step": 9272
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 9.542933831029734e-05,
+      "loss": 0.8901,
+      "step": 9273
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 9.541190526480132e-05,
+      "loss": 0.76,
+      "step": 9274
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 9.539447235903566e-05,
+      "loss": 0.8264,
+      "step": 9275
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 9.537703959353124e-05,
+      "loss": 0.7898,
+      "step": 9276
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 9.535960696881891e-05,
+      "loss": 0.911,
+      "step": 9277
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 9.534217448542964e-05,
+      "loss": 0.9086,
+      "step": 9278
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 9.532474214389432e-05,
+      "loss": 0.8535,
+      "step": 9279
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 9.530730994474389e-05,
+      "loss": 0.8118,
+      "step": 9280
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 9.528987788850918e-05,
+      "loss": 0.9378,
+      "step": 9281
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 9.527244597572112e-05,
+      "loss": 0.7811,
+      "step": 9282
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 9.525501420691061e-05,
+      "loss": 0.793,
+      "step": 9283
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 9.52375825826085e-05,
+      "loss": 0.828,
+      "step": 9284
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 9.52201511033457e-05,
+      "loss": 0.9671,
+      "step": 9285
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 9.520271976965306e-05,
+      "loss": 0.8732,
+      "step": 9286
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 9.518528858206148e-05,
+      "loss": 0.8616,
+      "step": 9287
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 9.516785754110181e-05,
+      "loss": 0.8365,
+      "step": 9288
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 9.51504266473049e-05,
+      "loss": 0.9981,
+      "step": 9289
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 9.513299590120162e-05,
+      "loss": 0.809,
+      "step": 9290
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 9.51155653033228e-05,
+      "loss": 1.0196,
+      "step": 9291
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 9.509813485419936e-05,
+      "loss": 0.9185,
+      "step": 9292
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 9.508070455436208e-05,
+      "loss": 0.8594,
+      "step": 9293
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 9.506327440434176e-05,
+      "loss": 0.8963,
+      "step": 9294
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 9.504584440466936e-05,
+      "loss": 0.8106,
+      "step": 9295
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 9.502841455587557e-05,
+      "loss": 0.8813,
+      "step": 9296
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 9.501098485849134e-05,
+      "loss": 0.8992,
+      "step": 9297
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 9.499355531304741e-05,
+      "loss": 0.9666,
+      "step": 9298
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 9.497612592007463e-05,
+      "loss": 0.8005,
+      "step": 9299
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 9.495869668010382e-05,
+      "loss": 0.7018,
+      "step": 9300
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 9.494126759366574e-05,
+      "loss": 0.8531,
+      "step": 9301
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 9.492383866129124e-05,
+      "loss": 0.8613,
+      "step": 9302
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 9.49064098835111e-05,
+      "loss": 0.915,
+      "step": 9303
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 9.488898126085612e-05,
+      "loss": 0.6951,
+      "step": 9304
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 9.487155279385707e-05,
+      "loss": 0.8305,
+      "step": 9305
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 9.485412448304473e-05,
+      "loss": 0.9687,
+      "step": 9306
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 9.483669632894993e-05,
+      "loss": 0.952,
+      "step": 9307
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 9.481926833210338e-05,
+      "loss": 0.8362,
+      "step": 9308
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 9.480184049303589e-05,
+      "loss": 0.9471,
+      "step": 9309
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 9.47844128122782e-05,
+      "loss": 0.7982,
+      "step": 9310
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 9.47669852903611e-05,
+      "loss": 0.8555,
+      "step": 9311
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 9.47495579278153e-05,
+      "loss": 0.8944,
+      "step": 9312
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 9.473213072517159e-05,
+      "loss": 0.928,
+      "step": 9313
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 9.47147036829607e-05,
+      "loss": 0.7876,
+      "step": 9314
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 9.469727680171334e-05,
+      "loss": 0.8792,
+      "step": 9315
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 9.467985008196033e-05,
+      "loss": 0.987,
+      "step": 9316
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 9.466242352423234e-05,
+      "loss": 0.8537,
+      "step": 9317
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 9.464499712906004e-05,
+      "loss": 0.8477,
+      "step": 9318
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 9.46275708969743e-05,
+      "loss": 0.8135,
+      "step": 9319
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 9.461014482850565e-05,
+      "loss": 0.8809,
+      "step": 9320
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 9.459271892418496e-05,
+      "loss": 0.8379,
+      "step": 9321
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 9.45752931845429e-05,
+      "loss": 0.8937,
+      "step": 9322
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 9.455786761011007e-05,
+      "loss": 0.8154,
+      "step": 9323
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 9.454044220141729e-05,
+      "loss": 0.8729,
+      "step": 9324
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 9.452301695899516e-05,
+      "loss": 0.801,
+      "step": 9325
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 9.450559188337446e-05,
+      "loss": 0.8548,
+      "step": 9326
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 9.448816697508577e-05,
+      "loss": 0.8666,
+      "step": 9327
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 9.447074223465982e-05,
+      "loss": 0.8657,
+      "step": 9328
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 9.44533176626273e-05,
+      "loss": 1.0138,
+      "step": 9329
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 9.44358932595188e-05,
+      "loss": 0.8529,
+      "step": 9330
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 9.441846902586505e-05,
+      "loss": 0.9256,
+      "step": 9331
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 9.440104496219668e-05,
+      "loss": 0.9368,
+      "step": 9332
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 9.438362106904433e-05,
+      "loss": 0.857,
+      "step": 9333
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 9.436619734693866e-05,
+      "loss": 0.7512,
+      "step": 9334
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 9.43487737964103e-05,
+      "loss": 0.7417,
+      "step": 9335
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 9.433135041798989e-05,
+      "loss": 1.0121,
+      "step": 9336
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 9.4313927212208e-05,
+      "loss": 0.8305,
+      "step": 9337
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 9.429650417959539e-05,
+      "loss": 0.9278,
+      "step": 9338
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 9.427908132068257e-05,
+      "loss": 0.9258,
+      "step": 9339
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 9.426165863600013e-05,
+      "loss": 0.8781,
+      "step": 9340
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 9.424423612607881e-05,
+      "loss": 0.8303,
+      "step": 9341
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 9.422681379144904e-05,
+      "loss": 0.8254,
+      "step": 9342
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 9.420939163264156e-05,
+      "loss": 0.9515,
+      "step": 9343
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 9.419196965018688e-05,
+      "loss": 0.8235,
+      "step": 9344
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 9.417454784461563e-05,
+      "loss": 0.8121,
+      "step": 9345
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 9.415712621645839e-05,
+      "loss": 0.8321,
+      "step": 9346
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 9.413970476624568e-05,
+      "loss": 0.7843,
+      "step": 9347
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 9.412228349450812e-05,
+      "loss": 0.8064,
+      "step": 9348
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 9.410486240177626e-05,
+      "loss": 0.7919,
+      "step": 9349
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 9.408744148858067e-05,
+      "loss": 0.8652,
+      "step": 9350
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 9.407002075545191e-05,
+      "loss": 0.7878,
+      "step": 9351
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 9.405260020292047e-05,
+      "loss": 0.8761,
+      "step": 9352
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 9.403517983151698e-05,
+      "loss": 0.8004,
+      "step": 9353
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 9.401775964177188e-05,
+      "loss": 0.8951,
+      "step": 9354
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 9.400033963421579e-05,
+      "loss": 0.9939,
+      "step": 9355
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 9.398291980937918e-05,
+      "loss": 0.6669,
+      "step": 9356
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 9.396550016779261e-05,
+      "loss": 0.8821,
+      "step": 9357
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 9.394808070998658e-05,
+      "loss": 0.934,
+      "step": 9358
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 9.393066143649156e-05,
+      "loss": 0.779,
+      "step": 9359
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 9.391324234783812e-05,
+      "loss": 0.9533,
+      "step": 9360
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 9.389582344455666e-05,
+      "loss": 1.0328,
+      "step": 9361
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 9.387840472717782e-05,
+      "loss": 0.8469,
+      "step": 9362
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 9.386098619623198e-05,
+      "loss": 0.8916,
+      "step": 9363
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 9.384356785224959e-05,
+      "loss": 0.8901,
+      "step": 9364
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 9.382614969576126e-05,
+      "loss": 0.8903,
+      "step": 9365
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 9.38087317272973e-05,
+      "loss": 0.7384,
+      "step": 9366
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 9.37913139473883e-05,
+      "loss": 0.8831,
+      "step": 9367
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 9.377389635656471e-05,
+      "loss": 0.9074,
+      "step": 9368
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 9.375647895535685e-05,
+      "loss": 0.7979,
+      "step": 9369
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 9.373906174429532e-05,
+      "loss": 0.7955,
+      "step": 9370
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 9.372164472391047e-05,
+      "loss": 0.8539,
+      "step": 9371
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 9.37042278947328e-05,
+      "loss": 0.8224,
+      "step": 9372
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 9.368681125729268e-05,
+      "loss": 0.8543,
+      "step": 9373
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 9.366939481212056e-05,
+      "loss": 0.8529,
+      "step": 9374
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 9.365197855974688e-05,
+      "loss": 0.9018,
+      "step": 9375
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 9.3634562500702e-05,
+      "loss": 0.8581,
+      "step": 9376
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 9.361714663551635e-05,
+      "loss": 0.9175,
+      "step": 9377
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 9.359973096472033e-05,
+      "loss": 0.9161,
+      "step": 9378
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 9.358231548884436e-05,
+      "loss": 0.88,
+      "step": 9379
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 9.35649002084188e-05,
+      "loss": 0.7517,
+      "step": 9380
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 9.3547485123974e-05,
+      "loss": 0.8309,
+      "step": 9381
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 9.353007023604041e-05,
+      "loss": 0.7891,
+      "step": 9382
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 9.351265554514831e-05,
+      "loss": 0.7626,
+      "step": 9383
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 9.349524105182818e-05,
+      "loss": 0.8831,
+      "step": 9384
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 9.347782675661029e-05,
+      "loss": 0.9147,
+      "step": 9385
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 9.346041266002498e-05,
+      "loss": 0.8198,
+      "step": 9386
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 9.344299876260268e-05,
+      "loss": 0.9146,
+      "step": 9387
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 9.342558506487363e-05,
+      "loss": 0.85,
+      "step": 9388
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 9.340817156736825e-05,
+      "loss": 0.8902,
+      "step": 9389
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 9.339075827061681e-05,
+      "loss": 0.8571,
+      "step": 9390
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 9.337334517514968e-05,
+      "loss": 0.8199,
+      "step": 9391
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 9.335593228149715e-05,
+      "loss": 0.844,
+      "step": 9392
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 9.333851959018951e-05,
+      "loss": 0.8747,
+      "step": 9393
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 9.332110710175709e-05,
+      "loss": 0.755,
+      "step": 9394
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 9.330369481673016e-05,
+      "loss": 0.9194,
+      "step": 9395
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 9.328628273563906e-05,
+      "loss": 0.8552,
+      "step": 9396
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 9.326887085901404e-05,
+      "loss": 0.7203,
+      "step": 9397
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 9.325145918738535e-05,
+      "loss": 0.7884,
+      "step": 9398
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 9.323404772128333e-05,
+      "loss": 0.8122,
+      "step": 9399
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 9.321663646123817e-05,
+      "loss": 0.7976,
+      "step": 9400
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 9.31992254077802e-05,
+      "loss": 0.9034,
+      "step": 9401
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 9.318181456143962e-05,
+      "loss": 0.8425,
+      "step": 9402
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 9.316440392274671e-05,
+      "loss": 0.9372,
+      "step": 9403
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 9.314699349223171e-05,
+      "loss": 0.9125,
+      "step": 9404
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 9.312958327042482e-05,
+      "loss": 0.9622,
+      "step": 9405
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 9.311217325785632e-05,
+      "loss": 0.8711,
+      "step": 9406
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 9.309476345505635e-05,
+      "loss": 0.8374,
+      "step": 9407
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 9.307735386255524e-05,
+      "loss": 0.7401,
+      "step": 9408
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 9.305994448088312e-05,
+      "loss": 0.7881,
+      "step": 9409
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 9.304253531057017e-05,
+      "loss": 0.8908,
+      "step": 9410
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 9.302512635214668e-05,
+      "loss": 0.7721,
+      "step": 9411
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 9.300771760614272e-05,
+      "loss": 0.8593,
+      "step": 9412
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 9.299030907308858e-05,
+      "loss": 0.7926,
+      "step": 9413
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 9.297290075351441e-05,
+      "loss": 0.7965,
+      "step": 9414
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 9.295549264795031e-05,
+      "loss": 0.8048,
+      "step": 9415
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 9.293808475692653e-05,
+      "loss": 0.8927,
+      "step": 9416
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 9.292067708097318e-05,
+      "loss": 0.8785,
+      "step": 9417
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 9.290326962062044e-05,
+      "loss": 0.8084,
+      "step": 9418
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 9.288586237639843e-05,
+      "loss": 0.798,
+      "step": 9419
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 9.28684553488373e-05,
+      "loss": 0.8265,
+      "step": 9420
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 9.285104853846717e-05,
+      "loss": 0.8188,
+      "step": 9421
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 9.283364194581817e-05,
+      "loss": 1.0,
+      "step": 9422
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 9.281623557142043e-05,
+      "loss": 1.0227,
+      "step": 9423
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 9.279882941580401e-05,
+      "loss": 0.8404,
+      "step": 9424
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 9.278142347949909e-05,
+      "loss": 0.753,
+      "step": 9425
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 9.276401776303572e-05,
+      "loss": 0.8962,
+      "step": 9426
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 9.274661226694398e-05,
+      "loss": 0.7333,
+      "step": 9427
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 9.272920699175399e-05,
+      "loss": 0.8839,
+      "step": 9428
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 9.271180193799577e-05,
+      "loss": 0.9387,
+      "step": 9429
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 9.26943971061995e-05,
+      "loss": 0.9418,
+      "step": 9430
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 9.267699249689515e-05,
+      "loss": 0.8542,
+      "step": 9431
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 9.265958811061276e-05,
+      "loss": 0.8953,
+      "step": 9432
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 9.264218394788249e-05,
+      "loss": 0.8033,
+      "step": 9433
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 9.262478000923423e-05,
+      "loss": 0.8138,
+      "step": 9434
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 9.260737629519817e-05,
+      "loss": 0.7952,
+      "step": 9435
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 9.258997280630424e-05,
+      "loss": 0.7444,
+      "step": 9436
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 9.257256954308252e-05,
+      "loss": 0.8514,
+      "step": 9437
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 9.255516650606299e-05,
+      "loss": 0.8569,
+      "step": 9438
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 9.253776369577563e-05,
+      "loss": 0.808,
+      "step": 9439
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 9.252036111275053e-05,
+      "loss": 0.897,
+      "step": 9440
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 9.250295875751761e-05,
+      "loss": 0.8376,
+      "step": 9441
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 9.24855566306069e-05,
+      "loss": 0.9035,
+      "step": 9442
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 9.246815473254837e-05,
+      "loss": 0.7207,
+      "step": 9443
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 9.245075306387196e-05,
+      "loss": 0.8416,
+      "step": 9444
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 9.24333516251077e-05,
+      "loss": 0.9837,
+      "step": 9445
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 9.241595041678549e-05,
+      "loss": 0.8841,
+      "step": 9446
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 9.239854943943533e-05,
+      "loss": 0.8267,
+      "step": 9447
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 9.238114869358711e-05,
+      "loss": 0.974,
+      "step": 9448
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 9.236374817977087e-05,
+      "loss": 0.9702,
+      "step": 9449
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 9.234634789851646e-05,
+      "loss": 0.6313,
+      "step": 9450
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 9.232894785035378e-05,
+      "loss": 0.8273,
+      "step": 9451
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 9.231154803581285e-05,
+      "loss": 1.0129,
+      "step": 9452
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 9.229414845542346e-05,
+      "loss": 0.8744,
+      "step": 9453
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 9.227674910971565e-05,
+      "loss": 0.8891,
+      "step": 9454
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 9.225934999921921e-05,
+      "loss": 0.7956,
+      "step": 9455
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 9.224195112446402e-05,
+      "loss": 0.8391,
+      "step": 9456
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 9.222455248598008e-05,
+      "loss": 0.9374,
+      "step": 9457
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 9.220715408429711e-05,
+      "loss": 0.9614,
+      "step": 9458
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 9.218975591994512e-05,
+      "loss": 0.8875,
+      "step": 9459
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 9.217235799345392e-05,
+      "loss": 0.8895,
+      "step": 9460
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 9.215496030535328e-05,
+      "loss": 0.8136,
+      "step": 9461
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 9.213756285617317e-05,
+      "loss": 0.9371,
+      "step": 9462
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 9.212016564644336e-05,
+      "loss": 0.8163,
+      "step": 9463
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 9.21027686766937e-05,
+      "loss": 0.8026,
+      "step": 9464
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 9.208537194745401e-05,
+      "loss": 0.9454,
+      "step": 9465
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 9.206797545925413e-05,
+      "loss": 0.7979,
+      "step": 9466
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 9.205057921262384e-05,
+      "loss": 0.8468,
+      "step": 9467
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 9.203318320809293e-05,
+      "loss": 0.8572,
+      "step": 9468
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 9.201578744619125e-05,
+      "loss": 0.8678,
+      "step": 9469
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 9.199839192744853e-05,
+      "loss": 0.8362,
+      "step": 9470
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 9.198099665239458e-05,
+      "loss": 0.9985,
+      "step": 9471
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 9.196360162155918e-05,
+      "loss": 0.7387,
+      "step": 9472
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 9.194620683547207e-05,
+      "loss": 0.8405,
+      "step": 9473
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 9.192881229466302e-05,
+      "loss": 0.7541,
+      "step": 9474
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 9.191141799966175e-05,
+      "loss": 0.8823,
+      "step": 9475
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 9.189402395099808e-05,
+      "loss": 0.794,
+      "step": 9476
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 9.18766301492017e-05,
+      "loss": 0.8511,
+      "step": 9477
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 9.185923659480226e-05,
+      "loss": 1.0053,
+      "step": 9478
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 9.184184328832966e-05,
+      "loss": 0.9286,
+      "step": 9479
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 9.182445023031342e-05,
+      "loss": 0.9108,
+      "step": 9480
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 9.180705742128337e-05,
+      "loss": 0.8532,
+      "step": 9481
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 9.178966486176917e-05,
+      "loss": 0.7668,
+      "step": 9482
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 9.177227255230051e-05,
+      "loss": 0.8815,
+      "step": 9483
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 9.17548804934071e-05,
+      "loss": 0.7061,
+      "step": 9484
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 9.173748868561856e-05,
+      "loss": 0.8943,
+      "step": 9485
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 9.172009712946458e-05,
+      "loss": 0.9144,
+      "step": 9486
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 9.170270582547483e-05,
+      "loss": 1.0448,
+      "step": 9487
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 9.168531477417897e-05,
+      "loss": 0.8361,
+      "step": 9488
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 9.166792397610665e-05,
+      "loss": 0.7622,
+      "step": 9489
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 9.165053343178745e-05,
+      "loss": 0.726,
+      "step": 9490
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 9.163314314175104e-05,
+      "loss": 0.8056,
+      "step": 9491
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 9.161575310652703e-05,
+      "loss": 0.8163,
+      "step": 9492
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 9.159836332664507e-05,
+      "loss": 0.939,
+      "step": 9493
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 9.158097380263467e-05,
+      "loss": 0.9337,
+      "step": 9494
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 9.156358453502557e-05,
+      "loss": 0.8965,
+      "step": 9495
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 9.154619552434725e-05,
+      "loss": 0.8466,
+      "step": 9496
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 9.15288067711293e-05,
+      "loss": 0.8272,
+      "step": 9497
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 9.151141827590137e-05,
+      "loss": 0.7673,
+      "step": 9498
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 9.14940300391929e-05,
+      "loss": 0.9576,
+      "step": 9499
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 9.147664206153357e-05,
+      "loss": 0.7949,
+      "step": 9500
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 9.145925434345291e-05,
+      "loss": 0.8784,
+      "step": 9501
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 9.144186688548036e-05,
+      "loss": 0.7277,
+      "step": 9502
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 9.142447968814558e-05,
+      "loss": 0.6797,
+      "step": 9503
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 9.140709275197803e-05,
+      "loss": 1.018,
+      "step": 9504
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 9.138970607750726e-05,
+      "loss": 0.7816,
+      "step": 9505
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 9.137231966526276e-05,
+      "loss": 0.9137,
+      "step": 9506
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 9.135493351577402e-05,
+      "loss": 0.8201,
+      "step": 9507
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 9.133754762957058e-05,
+      "loss": 0.864,
+      "step": 9508
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 9.132016200718188e-05,
+      "loss": 0.7791,
+      "step": 9509
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 9.130277664913743e-05,
+      "loss": 0.9532,
+      "step": 9510
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 9.128539155596668e-05,
+      "loss": 0.8468,
+      "step": 9511
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 9.12680067281991e-05,
+      "loss": 0.7826,
+      "step": 9512
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 9.125062216636417e-05,
+      "loss": 0.7995,
+      "step": 9513
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 9.123323787099128e-05,
+      "loss": 0.9723,
+      "step": 9514
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 9.121585384260992e-05,
+      "loss": 0.7603,
+      "step": 9515
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 9.119847008174949e-05,
+      "loss": 0.8934,
+      "step": 9516
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 9.118108658893944e-05,
+      "loss": 0.8375,
+      "step": 9517
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 9.116370336470916e-05,
+      "loss": 0.9389,
+      "step": 9518
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 9.114632040958803e-05,
+      "loss": 0.8911,
+      "step": 9519
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 9.112893772410552e-05,
+      "loss": 0.8632,
+      "step": 9520
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 9.111155530879094e-05,
+      "loss": 0.903,
+      "step": 9521
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 9.109417316417377e-05,
+      "loss": 0.886,
+      "step": 9522
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 9.10767912907833e-05,
+      "loss": 0.9061,
+      "step": 9523
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 9.105940968914885e-05,
+      "loss": 0.7266,
+      "step": 9524
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 9.104202835979994e-05,
+      "loss": 0.7968,
+      "step": 9525
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 9.102464730326574e-05,
+      "loss": 0.8381,
+      "step": 9526
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 9.100726652007572e-05,
+      "loss": 0.8544,
+      "step": 9527
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 9.098988601075914e-05,
+      "loss": 0.8946,
+      "step": 9528
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 9.097250577584537e-05,
+      "loss": 0.9197,
+      "step": 9529
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 9.095512581586368e-05,
+      "loss": 0.9552,
+      "step": 9530
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 9.093774613134341e-05,
+      "loss": 0.9612,
+      "step": 9531
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 9.092036672281384e-05,
+      "loss": 0.7734,
+      "step": 9532
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 9.090298759080424e-05,
+      "loss": 0.7768,
+      "step": 9533
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 9.088560873584394e-05,
+      "loss": 0.8719,
+      "step": 9534
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 9.086823015846219e-05,
+      "loss": 0.8995,
+      "step": 9535
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 9.085085185918822e-05,
+      "loss": 0.9312,
+      "step": 9536
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 9.083347383855135e-05,
+      "loss": 0.9178,
+      "step": 9537
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 9.081609609708077e-05,
+      "loss": 0.9728,
+      "step": 9538
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 9.079871863530577e-05,
+      "loss": 0.9307,
+      "step": 9539
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 9.07813414537555e-05,
+      "loss": 0.76,
+      "step": 9540
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 9.07639645529593e-05,
+      "loss": 0.8864,
+      "step": 9541
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 9.074658793344629e-05,
+      "loss": 0.928,
+      "step": 9542
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 9.072921159574566e-05,
+      "loss": 0.8531,
+      "step": 9543
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 9.071183554038672e-05,
+      "loss": 0.7766,
+      "step": 9544
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 9.069445976789853e-05,
+      "loss": 0.9358,
+      "step": 9545
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 9.067708427881034e-05,
+      "loss": 0.7979,
+      "step": 9546
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 9.065970907365134e-05,
+      "loss": 0.7009,
+      "step": 9547
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 9.06423341529506e-05,
+      "loss": 0.8802,
+      "step": 9548
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 9.062495951723736e-05,
+      "loss": 0.9481,
+      "step": 9549
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 9.06075851670407e-05,
+      "loss": 0.8634,
+      "step": 9550
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 9.059021110288982e-05,
+      "loss": 0.7927,
+      "step": 9551
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 9.057283732531381e-05,
+      "loss": 0.9092,
+      "step": 9552
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 9.055546383484177e-05,
+      "loss": 0.8663,
+      "step": 9553
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 9.053809063200286e-05,
+      "loss": 0.9355,
+      "step": 9554
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 9.052071771732613e-05,
+      "loss": 0.801,
+      "step": 9555
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 9.050334509134071e-05,
+      "loss": 0.9453,
+      "step": 9556
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 9.048597275457565e-05,
+      "loss": 0.8242,
+      "step": 9557
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 9.046860070756004e-05,
+      "loss": 0.8703,
+      "step": 9558
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 9.045122895082296e-05,
+      "loss": 0.836,
+      "step": 9559
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 9.043385748489344e-05,
+      "loss": 0.8683,
+      "step": 9560
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 9.041648631030055e-05,
+      "loss": 0.8531,
+      "step": 9561
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 9.03991154275733e-05,
+      "loss": 0.7686,
+      "step": 9562
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 9.038174483724075e-05,
+      "loss": 0.975,
+      "step": 9563
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 9.036437453983193e-05,
+      "loss": 0.7266,
+      "step": 9564
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 9.034700453587579e-05,
+      "loss": 0.8531,
+      "step": 9565
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 9.03296348259014e-05,
+      "loss": 0.8284,
+      "step": 9566
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 9.031226541043769e-05,
+      "loss": 0.872,
+      "step": 9567
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 9.029489629001374e-05,
+      "loss": 0.8758,
+      "step": 9568
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 9.027752746515845e-05,
+      "loss": 0.897,
+      "step": 9569
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 9.026015893640076e-05,
+      "loss": 0.8053,
+      "step": 9570
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 9.024279070426974e-05,
+      "loss": 0.8314,
+      "step": 9571
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 9.022542276929421e-05,
+      "loss": 0.8075,
+      "step": 9572
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 9.020805513200321e-05,
+      "loss": 0.9046,
+      "step": 9573
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 9.019068779292562e-05,
+      "loss": 0.8403,
+      "step": 9574
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 9.01733207525904e-05,
+      "loss": 0.8119,
+      "step": 9575
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 9.015595401152643e-05,
+      "loss": 0.7871,
+      "step": 9576
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 9.01385875702626e-05,
+      "loss": 1.0029,
+      "step": 9577
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 9.012122142932786e-05,
+      "loss": 0.7667,
+      "step": 9578
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 9.010385558925103e-05,
+      "loss": 0.8702,
+      "step": 9579
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 9.008649005056105e-05,
+      "loss": 0.8715,
+      "step": 9580
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 9.006912481378674e-05,
+      "loss": 0.8638,
+      "step": 9581
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 9.005175987945696e-05,
+      "loss": 0.7992,
+      "step": 9582
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 9.003439524810059e-05,
+      "loss": 0.9877,
+      "step": 9583
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 9.001703092024644e-05,
+      "loss": 0.8311,
+      "step": 9584
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 8.999966689642337e-05,
+      "loss": 0.8648,
+      "step": 9585
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 8.998230317716011e-05,
+      "loss": 0.7693,
+      "step": 9586
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 8.996493976298565e-05,
+      "loss": 0.8723,
+      "step": 9587
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 8.994757665442865e-05,
+      "loss": 0.8784,
+      "step": 9588
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 8.993021385201788e-05,
+      "loss": 0.7448,
+      "step": 9589
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 8.991285135628227e-05,
+      "loss": 0.7414,
+      "step": 9590
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 8.989548916775043e-05,
+      "loss": 0.8569,
+      "step": 9591
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 8.987812728695125e-05,
+      "loss": 1.023,
+      "step": 9592
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 8.986076571441346e-05,
+      "loss": 0.8629,
+      "step": 9593
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 8.984340445066572e-05,
+      "loss": 0.9029,
+      "step": 9594
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 8.982604349623686e-05,
+      "loss": 0.9584,
+      "step": 9595
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 8.980868285165558e-05,
+      "loss": 0.827,
+      "step": 9596
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 8.979132251745061e-05,
+      "loss": 0.8142,
+      "step": 9597
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 8.977396249415065e-05,
+      "loss": 0.9412,
+      "step": 9598
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 8.975660278228435e-05,
+      "loss": 0.8304,
+      "step": 9599
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 8.973924338238048e-05,
+      "loss": 0.8787,
+      "step": 9600
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 8.972188429496766e-05,
+      "loss": 0.8032,
+      "step": 9601
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 8.97045255205746e-05,
+      "loss": 0.8362,
+      "step": 9602
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 8.968716705972994e-05,
+      "loss": 0.7963,
+      "step": 9603
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 8.966980891296234e-05,
+      "loss": 0.764,
+      "step": 9604
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 8.965245108080045e-05,
+      "loss": 0.8806,
+      "step": 9605
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 8.963509356377286e-05,
+      "loss": 0.7378,
+      "step": 9606
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 8.961773636240826e-05,
+      "loss": 0.8012,
+      "step": 9607
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 8.960037947723518e-05,
+      "loss": 0.9188,
+      "step": 9608
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 8.95830229087823e-05,
+      "loss": 0.9015,
+      "step": 9609
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 8.956566665757821e-05,
+      "loss": 0.7174,
+      "step": 9610
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 8.954831072415142e-05,
+      "loss": 0.8704,
+      "step": 9611
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 8.953095510903058e-05,
+      "loss": 0.8575,
+      "step": 9612
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 8.951359981274419e-05,
+      "loss": 0.7783,
+      "step": 9613
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 8.949624483582091e-05,
+      "loss": 0.8612,
+      "step": 9614
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 8.947889017878919e-05,
+      "loss": 0.8165,
+      "step": 9615
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 8.946153584217753e-05,
+      "loss": 0.9379,
+      "step": 9616
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 8.944418182651461e-05,
+      "loss": 0.8743,
+      "step": 9617
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 8.942682813232879e-05,
+      "loss": 0.724,
+      "step": 9618
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 8.940947476014869e-05,
+      "loss": 0.9783,
+      "step": 9619
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 8.93921217105027e-05,
+      "loss": 0.8774,
+      "step": 9620
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 8.937476898391942e-05,
+      "loss": 0.8979,
+      "step": 9621
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 8.935741658092727e-05,
+      "loss": 0.7772,
+      "step": 9622
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 8.93400645020547e-05,
+      "loss": 0.8864,
+      "step": 9623
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 8.932271274783018e-05,
+      "loss": 0.777,
+      "step": 9624
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 8.930536131878217e-05,
+      "loss": 0.808,
+      "step": 9625
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 8.928801021543911e-05,
+      "loss": 0.8643,
+      "step": 9626
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 8.927065943832941e-05,
+      "loss": 0.8278,
+      "step": 9627
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 8.925330898798149e-05,
+      "loss": 0.849,
+      "step": 9628
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 8.923595886492378e-05,
+      "loss": 0.9093,
+      "step": 9629
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 8.921860906968463e-05,
+      "loss": 0.9085,
+      "step": 9630
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 8.920125960279248e-05,
+      "loss": 0.9597,
+      "step": 9631
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 8.918391046477565e-05,
+      "loss": 0.906,
+      "step": 9632
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 8.916656165616261e-05,
+      "loss": 0.8093,
+      "step": 9633
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 8.914921317748163e-05,
+      "loss": 0.876,
+      "step": 9634
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 8.913186502926103e-05,
+      "loss": 0.864,
+      "step": 9635
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 8.911451721202926e-05,
+      "loss": 0.9044,
+      "step": 9636
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 8.909716972631451e-05,
+      "loss": 0.8423,
+      "step": 9637
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 8.907982257264522e-05,
+      "loss": 0.857,
+      "step": 9638
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 8.906247575154967e-05,
+      "loss": 0.8355,
+      "step": 9639
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 8.904512926355606e-05,
+      "loss": 0.8582,
+      "step": 9640
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 8.902778310919281e-05,
+      "loss": 0.8854,
+      "step": 9641
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 8.90104372889881e-05,
+      "loss": 0.865,
+      "step": 9642
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 8.899309180347027e-05,
+      "loss": 0.975,
+      "step": 9643
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 8.897574665316752e-05,
+      "loss": 0.9552,
+      "step": 9644
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 8.895840183860812e-05,
+      "loss": 0.9425,
+      "step": 9645
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 8.89410573603203e-05,
+      "loss": 0.8772,
+      "step": 9646
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 8.892371321883229e-05,
+      "loss": 0.788,
+      "step": 9647
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 8.89063694146723e-05,
+      "loss": 0.8326,
+      "step": 9648
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 8.888902594836853e-05,
+      "loss": 0.8301,
+      "step": 9649
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 8.887168282044921e-05,
+      "loss": 0.8098,
+      "step": 9650
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 8.88543400314425e-05,
+      "loss": 0.7509,
+      "step": 9651
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 8.883699758187654e-05,
+      "loss": 0.8833,
+      "step": 9652
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 8.881965547227954e-05,
+      "loss": 0.7119,
+      "step": 9653
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 8.880231370317961e-05,
+      "loss": 0.7578,
+      "step": 9654
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 8.878497227510499e-05,
+      "loss": 0.9316,
+      "step": 9655
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 8.87676311885837e-05,
+      "loss": 0.7693,
+      "step": 9656
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 8.875029044414388e-05,
+      "loss": 0.9323,
+      "step": 9657
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 8.873295004231373e-05,
+      "loss": 0.8966,
+      "step": 9658
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 8.871560998362124e-05,
+      "loss": 0.8521,
+      "step": 9659
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 8.869827026859458e-05,
+      "loss": 0.9803,
+      "step": 9660
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 8.868093089776183e-05,
+      "loss": 0.852,
+      "step": 9661
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 8.866359187165096e-05,
+      "loss": 0.802,
+      "step": 9662
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 8.864625319079016e-05,
+      "loss": 0.7357,
+      "step": 9663
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 8.862891485570739e-05,
+      "loss": 0.8415,
+      "step": 9664
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 8.861157686693074e-05,
+      "loss": 0.8999,
+      "step": 9665
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 8.859423922498819e-05,
+      "loss": 0.8739,
+      "step": 9666
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 8.857690193040781e-05,
+      "loss": 0.7477,
+      "step": 9667
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 8.85595649837176e-05,
+      "loss": 0.7967,
+      "step": 9668
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 8.85422283854455e-05,
+      "loss": 1.0135,
+      "step": 9669
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 8.852489213611956e-05,
+      "loss": 0.8508,
+      "step": 9670
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 8.85075562362677e-05,
+      "loss": 0.9165,
+      "step": 9671
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 8.849022068641794e-05,
+      "loss": 0.8541,
+      "step": 9672
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 8.847288548709821e-05,
+      "loss": 0.8579,
+      "step": 9673
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 8.845555063883643e-05,
+      "loss": 0.7972,
+      "step": 9674
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 8.843821614216056e-05,
+      "loss": 0.8054,
+      "step": 9675
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 8.84208819975985e-05,
+      "loss": 0.8569,
+      "step": 9676
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 8.840354820567821e-05,
+      "loss": 0.8532,
+      "step": 9677
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 8.83862147669275e-05,
+      "loss": 0.8276,
+      "step": 9678
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 8.836888168187437e-05,
+      "loss": 0.8663,
+      "step": 9679
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 8.835154895104664e-05,
+      "loss": 0.7997,
+      "step": 9680
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 8.833421657497213e-05,
+      "loss": 0.9155,
+      "step": 9681
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 8.83168845541788e-05,
+      "loss": 0.9086,
+      "step": 9682
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 8.829955288919439e-05,
+      "loss": 0.8934,
+      "step": 9683
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 8.828222158054685e-05,
+      "loss": 0.8322,
+      "step": 9684
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 8.826489062876394e-05,
+      "loss": 0.8697,
+      "step": 9685
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 8.824756003437342e-05,
+      "loss": 0.764,
+      "step": 9686
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 8.82302297979032e-05,
+      "loss": 0.7583,
+      "step": 9687
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 8.8212899919881e-05,
+      "loss": 0.8231,
+      "step": 9688
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 8.819557040083464e-05,
+      "loss": 0.7997,
+      "step": 9689
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 8.817824124129188e-05,
+      "loss": 0.8427,
+      "step": 9690
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 8.816091244178044e-05,
+      "loss": 0.8234,
+      "step": 9691
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 8.814358400282814e-05,
+      "loss": 0.8371,
+      "step": 9692
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 8.812625592496264e-05,
+      "loss": 0.7813,
+      "step": 9693
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 8.810892820871173e-05,
+      "loss": 0.9245,
+      "step": 9694
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 8.809160085460309e-05,
+      "loss": 0.9024,
+      "step": 9695
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 8.807427386316443e-05,
+      "loss": 0.8521,
+      "step": 9696
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 8.805694723492347e-05,
+      "loss": 0.9189,
+      "step": 9697
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 8.803962097040783e-05,
+      "loss": 0.9178,
+      "step": 9698
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 8.802229507014526e-05,
+      "loss": 0.8517,
+      "step": 9699
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 8.800496953466332e-05,
+      "loss": 0.8323,
+      "step": 9700
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 8.798764436448978e-05,
+      "loss": 0.7998,
+      "step": 9701
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 8.79703195601522e-05,
+      "loss": 0.9047,
+      "step": 9702
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 8.79529951221782e-05,
+      "loss": 0.8433,
+      "step": 9703
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 8.793567105109546e-05,
+      "loss": 0.9102,
+      "step": 9704
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 8.791834734743148e-05,
+      "loss": 0.7278,
+      "step": 9705
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 8.790102401171395e-05,
+      "loss": 0.8719,
+      "step": 9706
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 8.788370104447044e-05,
+      "loss": 0.8481,
+      "step": 9707
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 8.786637844622845e-05,
+      "loss": 0.9238,
+      "step": 9708
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 8.78490562175156e-05,
+      "loss": 0.8236,
+      "step": 9709
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 8.783173435885943e-05,
+      "loss": 0.8353,
+      "step": 9710
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 8.781441287078748e-05,
+      "loss": 0.9998,
+      "step": 9711
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 8.779709175382725e-05,
+      "loss": 0.7579,
+      "step": 9712
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 8.777977100850628e-05,
+      "loss": 0.8662,
+      "step": 9713
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 8.776245063535206e-05,
+      "loss": 0.8579,
+      "step": 9714
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 8.774513063489209e-05,
+      "loss": 0.7754,
+      "step": 9715
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 8.772781100765385e-05,
+      "loss": 0.9439,
+      "step": 9716
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 8.771049175416479e-05,
+      "loss": 0.8464,
+      "step": 9717
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 8.769317287495238e-05,
+      "loss": 0.9304,
+      "step": 9718
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 8.767585437054408e-05,
+      "loss": 0.9094,
+      "step": 9719
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 8.76585362414673e-05,
+      "loss": 0.8912,
+      "step": 9720
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 8.764121848824949e-05,
+      "loss": 0.8526,
+      "step": 9721
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 8.762390111141801e-05,
+      "loss": 0.7969,
+      "step": 9722
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 8.760658411150034e-05,
+      "loss": 0.885,
+      "step": 9723
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 8.758926748902377e-05,
+      "loss": 0.8032,
+      "step": 9724
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 8.75719512445158e-05,
+      "loss": 0.8809,
+      "step": 9725
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 8.755463537850371e-05,
+      "loss": 0.9223,
+      "step": 9726
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 8.753731989151483e-05,
+      "loss": 0.9113,
+      "step": 9727
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 8.752000478407662e-05,
+      "loss": 1.0056,
+      "step": 9728
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 8.750269005671625e-05,
+      "loss": 1.0199,
+      "step": 9729
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 8.748537570996118e-05,
+      "loss": 0.8823,
+      "step": 9730
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 8.74680617443387e-05,
+      "loss": 0.8905,
+      "step": 9731
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 8.745074816037599e-05,
+      "loss": 0.8652,
+      "step": 9732
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 8.743343495860046e-05,
+      "loss": 0.9511,
+      "step": 9733
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 8.741612213953932e-05,
+      "loss": 0.7697,
+      "step": 9734
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 8.739880970371986e-05,
+      "loss": 0.9735,
+      "step": 9735
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 8.738149765166933e-05,
+      "loss": 0.8393,
+      "step": 9736
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 8.736418598391494e-05,
+      "loss": 0.7601,
+      "step": 9737
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 8.734687470098395e-05,
+      "loss": 0.7457,
+      "step": 9738
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 8.732956380340353e-05,
+      "loss": 1.0151,
+      "step": 9739
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 8.731225329170095e-05,
+      "loss": 0.924,
+      "step": 9740
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 8.729494316640333e-05,
+      "loss": 0.9732,
+      "step": 9741
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 8.727763342803791e-05,
+      "loss": 0.8483,
+      "step": 9742
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 8.726032407713183e-05,
+      "loss": 0.7784,
+      "step": 9743
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 8.724301511421223e-05,
+      "loss": 0.8655,
+      "step": 9744
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 8.722570653980627e-05,
+      "loss": 0.917,
+      "step": 9745
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 8.720839835444105e-05,
+      "loss": 0.9633,
+      "step": 9746
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 8.71910905586438e-05,
+      "loss": 0.7847,
+      "step": 9747
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 8.717378315294152e-05,
+      "loss": 0.9421,
+      "step": 9748
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 8.715647613786129e-05,
+      "loss": 0.7969,
+      "step": 9749
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 8.713916951393031e-05,
+      "loss": 0.7177,
+      "step": 9750
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 8.712186328167551e-05,
+      "loss": 0.8451,
+      "step": 9751
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 8.710455744162408e-05,
+      "loss": 1.0369,
+      "step": 9752
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 8.708725199430303e-05,
+      "loss": 0.8831,
+      "step": 9753
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 8.706994694023932e-05,
+      "loss": 0.8145,
+      "step": 9754
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 8.705264227996009e-05,
+      "loss": 0.9863,
+      "step": 9755
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 8.703533801399227e-05,
+      "loss": 0.8305,
+      "step": 9756
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 8.701803414286292e-05,
+      "loss": 0.7726,
+      "step": 9757
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 8.700073066709896e-05,
+      "loss": 0.802,
+      "step": 9758
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 8.698342758722744e-05,
+      "loss": 0.837,
+      "step": 9759
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 8.696612490377528e-05,
+      "loss": 0.8496,
+      "step": 9760
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 8.694882261726944e-05,
+      "loss": 0.8131,
+      "step": 9761
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 8.693152072823687e-05,
+      "loss": 0.7514,
+      "step": 9762
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 8.691421923720447e-05,
+      "loss": 0.8767,
+      "step": 9763
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 8.689691814469923e-05,
+      "loss": 0.8562,
+      "step": 9764
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 8.687961745124797e-05,
+      "loss": 0.8505,
+      "step": 9765
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 8.68623171573776e-05,
+      "loss": 0.7994,
+      "step": 9766
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 8.684501726361506e-05,
+      "loss": 0.8714,
+      "step": 9767
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 8.682771777048714e-05,
+      "loss": 0.716,
+      "step": 9768
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 8.681041867852074e-05,
+      "loss": 0.8125,
+      "step": 9769
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 8.679311998824267e-05,
+      "loss": 0.8282,
+      "step": 9770
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 8.677582170017984e-05,
+      "loss": 0.8631,
+      "step": 9771
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 8.675852381485899e-05,
+      "loss": 0.7894,
+      "step": 9772
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 8.67412263328069e-05,
+      "loss": 0.7995,
+      "step": 9773
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 8.672392925455048e-05,
+      "loss": 0.7633,
+      "step": 9774
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 8.670663258061638e-05,
+      "loss": 0.9602,
+      "step": 9775
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 8.66893363115315e-05,
+      "loss": 0.7952,
+      "step": 9776
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 8.667204044782253e-05,
+      "loss": 0.9204,
+      "step": 9777
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 8.665474499001617e-05,
+      "loss": 0.9309,
+      "step": 9778
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 8.663744993863923e-05,
+      "loss": 0.9138,
+      "step": 9779
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 8.66201552942184e-05,
+      "loss": 0.8473,
+      "step": 9780
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 8.660286105728039e-05,
+      "loss": 0.8501,
+      "step": 9781
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 8.65855672283519e-05,
+      "loss": 0.783,
+      "step": 9782
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 8.65682738079596e-05,
+      "loss": 1.0083,
+      "step": 9783
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 8.655098079663018e-05,
+      "loss": 0.8325,
+      "step": 9784
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 8.653368819489026e-05,
+      "loss": 0.8398,
+      "step": 9785
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 8.651639600326654e-05,
+      "loss": 0.8557,
+      "step": 9786
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 8.649910422228561e-05,
+      "loss": 0.7631,
+      "step": 9787
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 8.648181285247412e-05,
+      "loss": 0.8779,
+      "step": 9788
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 8.646452189435868e-05,
+      "loss": 0.7511,
+      "step": 9789
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 8.644723134846582e-05,
+      "loss": 0.8816,
+      "step": 9790
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 8.642994121532222e-05,
+      "loss": 0.8307,
+      "step": 9791
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 8.641265149545434e-05,
+      "loss": 0.8018,
+      "step": 9792
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 8.639536218938889e-05,
+      "loss": 0.9253,
+      "step": 9793
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 8.637807329765228e-05,
+      "loss": 0.8245,
+      "step": 9794
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 8.636078482077107e-05,
+      "loss": 0.8191,
+      "step": 9795
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 8.634349675927186e-05,
+      "loss": 0.9742,
+      "step": 9796
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 8.632620911368103e-05,
+      "loss": 0.8224,
+      "step": 9797
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 8.630892188452518e-05,
+      "loss": 0.9026,
+      "step": 9798
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 8.629163507233077e-05,
+      "loss": 0.9108,
+      "step": 9799
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 8.627434867762422e-05,
+      "loss": 0.7074,
+      "step": 9800
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 8.625706270093204e-05,
+      "loss": 0.872,
+      "step": 9801
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 8.623977714278064e-05,
+      "loss": 0.8484,
+      "step": 9802
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 8.622249200369648e-05,
+      "loss": 0.8396,
+      "step": 9803
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 8.620520728420596e-05,
+      "loss": 0.8465,
+      "step": 9804
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 8.618792298483549e-05,
+      "loss": 0.9522,
+      "step": 9805
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 8.617063910611147e-05,
+      "loss": 0.8852,
+      "step": 9806
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 8.615335564856026e-05,
+      "loss": 0.8998,
+      "step": 9807
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 8.613607261270826e-05,
+      "loss": 0.6916,
+      "step": 9808
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 8.611878999908178e-05,
+      "loss": 0.8828,
+      "step": 9809
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 8.610150780820721e-05,
+      "loss": 0.8302,
+      "step": 9810
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 8.608422604061086e-05,
+      "loss": 0.9678,
+      "step": 9811
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 8.6066944696819e-05,
+      "loss": 0.9484,
+      "step": 9812
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 8.6049663777358e-05,
+      "loss": 0.8079,
+      "step": 9813
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 8.603238328275409e-05,
+      "loss": 0.9162,
+      "step": 9814
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 8.601510321353365e-05,
+      "loss": 0.7811,
+      "step": 9815
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 8.599782357022278e-05,
+      "loss": 0.8423,
+      "step": 9816
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 8.598054435334792e-05,
+      "loss": 0.8617,
+      "step": 9817
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 8.596326556343519e-05,
+      "loss": 0.9062,
+      "step": 9818
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 8.594598720101079e-05,
+      "loss": 0.9481,
+      "step": 9819
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 8.592870926660105e-05,
+      "loss": 0.8909,
+      "step": 9820
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 8.591143176073204e-05,
+      "loss": 0.9348,
+      "step": 9821
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 8.589415468393003e-05,
+      "loss": 0.8138,
+      "step": 9822
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 8.58768780367212e-05,
+      "loss": 0.8049,
+      "step": 9823
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 8.585960181963162e-05,
+      "loss": 0.8635,
+      "step": 9824
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 8.584232603318753e-05,
+      "loss": 0.8105,
+      "step": 9825
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 8.5825050677915e-05,
+      "loss": 0.906,
+      "step": 9826
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 8.580777575434023e-05,
+      "loss": 0.8214,
+      "step": 9827
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 8.579050126298925e-05,
+      "loss": 0.7909,
+      "step": 9828
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 8.577322720438816e-05,
+      "loss": 0.8526,
+      "step": 9829
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 8.575595357906309e-05,
+      "loss": 0.8757,
+      "step": 9830
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 8.573868038754008e-05,
+      "loss": 0.7978,
+      "step": 9831
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 8.572140763034517e-05,
+      "loss": 0.9212,
+      "step": 9832
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 8.570413530800442e-05,
+      "loss": 0.7896,
+      "step": 9833
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 8.568686342104386e-05,
+      "loss": 0.6553,
+      "step": 9834
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 8.566959196998952e-05,
+      "loss": 0.8855,
+      "step": 9835
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 8.565232095536733e-05,
+      "loss": 0.8428,
+      "step": 9836
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 8.563505037770337e-05,
+      "loss": 0.8947,
+      "step": 9837
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 8.561778023752352e-05,
+      "loss": 0.698,
+      "step": 9838
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 8.560051053535387e-05,
+      "loss": 0.8349,
+      "step": 9839
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 8.558324127172026e-05,
+      "loss": 0.8316,
+      "step": 9840
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 8.556597244714863e-05,
+      "loss": 0.8568,
+      "step": 9841
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 8.554870406216499e-05,
+      "loss": 0.8151,
+      "step": 9842
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 8.553143611729511e-05,
+      "loss": 0.8603,
+      "step": 9843
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 8.551416861306503e-05,
+      "loss": 0.8272,
+      "step": 9844
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 8.549690155000056e-05,
+      "loss": 0.8443,
+      "step": 9845
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 8.547963492862752e-05,
+      "loss": 0.9952,
+      "step": 9846
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 8.546236874947186e-05,
+      "loss": 0.8731,
+      "step": 9847
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 8.544510301305937e-05,
+      "loss": 0.8132,
+      "step": 9848
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 8.542783771991589e-05,
+      "loss": 0.7448,
+      "step": 9849
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 8.541057287056722e-05,
+      "loss": 0.7993,
+      "step": 9850
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 8.539330846553919e-05,
+      "loss": 0.9146,
+      "step": 9851
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 8.537604450535756e-05,
+      "loss": 0.9462,
+      "step": 9852
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 8.535878099054809e-05,
+      "loss": 0.905,
+      "step": 9853
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 8.534151792163658e-05,
+      "loss": 0.8118,
+      "step": 9854
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 8.532425529914875e-05,
+      "loss": 0.8191,
+      "step": 9855
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 8.530699312361034e-05,
+      "loss": 0.9497,
+      "step": 9856
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 8.528973139554709e-05,
+      "loss": 0.7949,
+      "step": 9857
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 8.527247011548465e-05,
+      "loss": 0.8146,
+      "step": 9858
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 8.525520928394878e-05,
+      "loss": 0.9186,
+      "step": 9859
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 8.523794890146506e-05,
+      "loss": 0.8525,
+      "step": 9860
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 8.52206889685593e-05,
+      "loss": 0.8031,
+      "step": 9861
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 8.520342948575701e-05,
+      "loss": 0.8991,
+      "step": 9862
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 8.518617045358391e-05,
+      "loss": 0.8474,
+      "step": 9863
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 8.516891187256564e-05,
+      "loss": 0.8346,
+      "step": 9864
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 8.515165374322768e-05,
+      "loss": 0.8318,
+      "step": 9865
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 8.513439606609579e-05,
+      "loss": 0.8891,
+      "step": 9866
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 8.511713884169543e-05,
+      "loss": 0.8657,
+      "step": 9867
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 8.509988207055225e-05,
+      "loss": 0.9026,
+      "step": 9868
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 8.508262575319177e-05,
+      "loss": 0.8088,
+      "step": 9869
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 8.506536989013951e-05,
+      "loss": 0.8175,
+      "step": 9870
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 8.504811448192104e-05,
+      "loss": 0.9217,
+      "step": 9871
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 8.503085952906182e-05,
+      "loss": 0.7965,
+      "step": 9872
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 8.501360503208741e-05,
+      "loss": 0.8277,
+      "step": 9873
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 8.499635099152328e-05,
+      "loss": 0.8678,
+      "step": 9874
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 8.497909740789485e-05,
+      "loss": 0.8712,
+      "step": 9875
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 8.496184428172763e-05,
+      "loss": 0.8715,
+      "step": 9876
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 8.494459161354703e-05,
+      "loss": 0.9035,
+      "step": 9877
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 8.492733940387852e-05,
+      "loss": 0.8688,
+      "step": 9878
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 8.491008765324747e-05,
+      "loss": 0.745,
+      "step": 9879
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 8.489283636217933e-05,
+      "loss": 0.874,
+      "step": 9880
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 8.487558553119947e-05,
+      "loss": 0.8056,
+      "step": 9881
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 8.485833516083322e-05,
+      "loss": 0.9211,
+      "step": 9882
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 8.4841085251606e-05,
+      "loss": 0.8096,
+      "step": 9883
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 8.482383580404308e-05,
+      "loss": 0.7862,
+      "step": 9884
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 8.480658681866992e-05,
+      "loss": 0.9003,
+      "step": 9885
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 8.478933829601173e-05,
+      "loss": 0.7788,
+      "step": 9886
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 8.47720902365938e-05,
+      "loss": 0.783,
+      "step": 9887
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 8.475484264094153e-05,
+      "loss": 0.8219,
+      "step": 9888
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 8.473759550958007e-05,
+      "loss": 0.8021,
+      "step": 9889
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 8.472034884303478e-05,
+      "loss": 0.8249,
+      "step": 9890
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 8.470310264183087e-05,
+      "loss": 0.7784,
+      "step": 9891
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 8.468585690649352e-05,
+      "loss": 0.8851,
+      "step": 9892
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 8.466861163754803e-05,
+      "loss": 0.9271,
+      "step": 9893
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 8.465136683551955e-05,
+      "loss": 0.8024,
+      "step": 9894
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 8.463412250093332e-05,
+      "loss": 0.8047,
+      "step": 9895
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 8.461687863431445e-05,
+      "loss": 0.8742,
+      "step": 9896
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 8.459963523618816e-05,
+      "loss": 0.8324,
+      "step": 9897
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 8.458239230707959e-05,
+      "loss": 0.7925,
+      "step": 9898
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 8.456514984751383e-05,
+      "loss": 0.7271,
+      "step": 9899
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 8.454790785801605e-05,
+      "loss": 0.8888,
+      "step": 9900
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 8.45306663391113e-05,
+      "loss": 0.899,
+      "step": 9901
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 8.451342529132474e-05,
+      "loss": 0.8273,
+      "step": 9902
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 8.449618471518138e-05,
+      "loss": 0.8456,
+      "step": 9903
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 8.44789446112063e-05,
+      "loss": 1.02,
+      "step": 9904
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 8.446170497992457e-05,
+      "loss": 0.7819,
+      "step": 9905
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 8.444446582186118e-05,
+      "loss": 0.8237,
+      "step": 9906
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 8.442722713754122e-05,
+      "loss": 0.8733,
+      "step": 9907
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 8.440998892748959e-05,
+      "loss": 0.8422,
+      "step": 9908
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 8.439275119223139e-05,
+      "loss": 0.9031,
+      "step": 9909
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 8.437551393229154e-05,
+      "loss": 0.8909,
+      "step": 9910
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 8.435827714819497e-05,
+      "loss": 0.7606,
+      "step": 9911
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 8.43410408404667e-05,
+      "loss": 0.8419,
+      "step": 9912
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 8.432380500963157e-05,
+      "loss": 0.8296,
+      "step": 9913
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 8.430656965621459e-05,
+      "loss": 0.8464,
+      "step": 9914
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 8.428933478074061e-05,
+      "loss": 0.8191,
+      "step": 9915
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 8.427210038373452e-05,
+      "loss": 0.907,
+      "step": 9916
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 8.42548664657212e-05,
+      "loss": 0.8664,
+      "step": 9917
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 8.423763302722551e-05,
+      "loss": 0.9401,
+      "step": 9918
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 8.42204000687723e-05,
+      "loss": 0.8244,
+      "step": 9919
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 8.42031675908864e-05,
+      "loss": 1.0091,
+      "step": 9920
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 8.418593559409258e-05,
+      "loss": 0.9267,
+      "step": 9921
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 8.416870407891572e-05,
+      "loss": 0.7777,
+      "step": 9922
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 8.415147304588053e-05,
+      "loss": 0.8889,
+      "step": 9923
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 8.413424249551183e-05,
+      "loss": 0.9328,
+      "step": 9924
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 8.411701242833433e-05,
+      "loss": 0.9801,
+      "step": 9925
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 8.409978284487284e-05,
+      "loss": 0.813,
+      "step": 9926
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 8.408255374565203e-05,
+      "loss": 0.6895,
+      "step": 9927
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 8.40653251311966e-05,
+      "loss": 0.8748,
+      "step": 9928
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 8.404809700203132e-05,
+      "loss": 0.8312,
+      "step": 9929
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 8.403086935868077e-05,
+      "loss": 0.8842,
+      "step": 9930
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 8.401364220166974e-05,
+      "loss": 0.7914,
+      "step": 9931
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 8.399641553152279e-05,
+      "loss": 0.8305,
+      "step": 9932
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 8.397918934876456e-05,
+      "loss": 0.9611,
+      "step": 9933
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 8.396196365391975e-05,
+      "loss": 0.94,
+      "step": 9934
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 8.394473844751285e-05,
+      "loss": 0.8688,
+      "step": 9935
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 8.392751373006857e-05,
+      "loss": 0.8134,
+      "step": 9936
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 8.391028950211145e-05,
+      "loss": 0.7461,
+      "step": 9937
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 8.389306576416598e-05,
+      "loss": 0.8305,
+      "step": 9938
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 8.387584251675682e-05,
+      "loss": 0.8804,
+      "step": 9939
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 8.385861976040842e-05,
+      "loss": 0.8576,
+      "step": 9940
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 8.384139749564535e-05,
+      "loss": 0.9145,
+      "step": 9941
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 8.38241757229921e-05,
+      "loss": 0.7881,
+      "step": 9942
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 8.380695444297315e-05,
+      "loss": 0.9252,
+      "step": 9943
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 8.378973365611297e-05,
+      "loss": 0.9615,
+      "step": 9944
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 8.377251336293603e-05,
+      "loss": 0.8848,
+      "step": 9945
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 8.375529356396678e-05,
+      "loss": 0.9072,
+      "step": 9946
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 8.373807425972962e-05,
+      "loss": 0.966,
+      "step": 9947
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 8.372085545074899e-05,
+      "loss": 0.9273,
+      "step": 9948
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 8.37036371375493e-05,
+      "loss": 0.8239,
+      "step": 9949
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 8.368641932065487e-05,
+      "loss": 0.8717,
+      "step": 9950
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 8.366920200059015e-05,
+      "loss": 0.8548,
+      "step": 9951
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 8.36519851778794e-05,
+      "loss": 0.8017,
+      "step": 9952
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 8.363476885304708e-05,
+      "loss": 0.8578,
+      "step": 9953
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 8.361755302661739e-05,
+      "loss": 0.8531,
+      "step": 9954
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 8.360033769911472e-05,
+      "loss": 0.967,
+      "step": 9955
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 8.358312287106338e-05,
+      "loss": 0.8402,
+      "step": 9956
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 8.356590854298751e-05,
+      "loss": 0.8266,
+      "step": 9957
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 8.354869471541152e-05,
+      "loss": 0.8708,
+      "step": 9958
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 8.353148138885958e-05,
+      "loss": 0.7572,
+      "step": 9959
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 8.351426856385597e-05,
+      "loss": 0.8773,
+      "step": 9960
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 8.349705624092487e-05,
+      "loss": 0.8201,
+      "step": 9961
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 8.347984442059046e-05,
+      "loss": 0.829,
+      "step": 9962
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 8.346263310337698e-05,
+      "loss": 0.788,
+      "step": 9963
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 8.344542228980856e-05,
+      "loss": 0.8569,
+      "step": 9964
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 8.34282119804094e-05,
+      "loss": 0.8366,
+      "step": 9965
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 8.341100217570359e-05,
+      "loss": 0.8941,
+      "step": 9966
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 8.339379287621526e-05,
+      "loss": 0.8191,
+      "step": 9967
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 8.337658408246856e-05,
+      "loss": 0.8437,
+      "step": 9968
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 8.335937579498752e-05,
+      "loss": 0.8647,
+      "step": 9969
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 8.334216801429629e-05,
+      "loss": 0.7851,
+      "step": 9970
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 8.332496074091886e-05,
+      "loss": 0.8662,
+      "step": 9971
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 8.330775397537934e-05,
+      "loss": 0.8017,
+      "step": 9972
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 8.329054771820174e-05,
+      "loss": 0.7566,
+      "step": 9973
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 8.327334196991003e-05,
+      "loss": 0.8916,
+      "step": 9974
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 8.32561367310283e-05,
+      "loss": 0.894,
+      "step": 9975
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 8.323893200208043e-05,
+      "loss": 0.8398,
+      "step": 9976
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 8.322172778359051e-05,
+      "loss": 0.8062,
+      "step": 9977
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 8.320452407608242e-05,
+      "loss": 0.9422,
+      "step": 9978
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 8.318732088008005e-05,
+      "loss": 0.8293,
+      "step": 9979
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 8.317011819610745e-05,
+      "loss": 0.9589,
+      "step": 9980
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 8.315291602468839e-05,
+      "loss": 0.8731,
+      "step": 9981
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 8.313571436634689e-05,
+      "loss": 0.9478,
+      "step": 9982
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 8.311851322160676e-05,
+      "loss": 0.7867,
+      "step": 9983
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 8.310131259099182e-05,
+      "loss": 0.9475,
+      "step": 9984
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 8.308411247502602e-05,
+      "loss": 0.7359,
+      "step": 9985
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 8.306691287423307e-05,
+      "loss": 0.8652,
+      "step": 9986
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 8.30497137891369e-05,
+      "loss": 0.9748,
+      "step": 9987
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 8.30325152202612e-05,
+      "loss": 0.8744,
+      "step": 9988
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 8.301531716812986e-05,
+      "loss": 0.8987,
+      "step": 9989
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 8.299811963326657e-05,
+      "loss": 0.7613,
+      "step": 9990
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 8.298092261619508e-05,
+      "loss": 0.9183,
+      "step": 9991
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 8.296372611743916e-05,
+      "loss": 0.7853,
+      "step": 9992
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 8.29465301375225e-05,
+      "loss": 0.742,
+      "step": 9993
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 8.292933467696886e-05,
+      "loss": 0.8903,
+      "step": 9994
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 8.291213973630186e-05,
+      "loss": 0.8524,
+      "step": 9995
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 8.289494531604516e-05,
+      "loss": 0.8829,
+      "step": 9996
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 8.287775141672249e-05,
+      "loss": 0.8474,
+      "step": 9997
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 8.286055803885742e-05,
+      "loss": 0.9417,
+      "step": 9998
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 8.284336518297366e-05,
+      "loss": 0.7771,
+      "step": 9999
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 8.28261728495947e-05,
+      "loss": 0.7642,
+      "step": 10000
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 8.280898103924422e-05,
+      "loss": 1.0008,
+      "step": 10001
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 8.279178975244581e-05,
+      "loss": 0.8835,
+      "step": 10002
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 8.277459898972292e-05,
+      "loss": 0.8814,
+      "step": 10003
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 8.275740875159923e-05,
+      "loss": 0.9206,
+      "step": 10004
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 8.274021903859817e-05,
+      "loss": 0.92,
+      "step": 10005
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 8.272302985124329e-05,
+      "loss": 0.8771,
+      "step": 10006
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 8.27058411900581e-05,
+      "loss": 0.8664,
+      "step": 10007
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 8.268865305556604e-05,
+      "loss": 0.8359,
+      "step": 10008
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 8.267146544829061e-05,
+      "loss": 0.9528,
+      "step": 10009
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 8.265427836875524e-05,
+      "loss": 0.8329,
+      "step": 10010
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 8.263709181748337e-05,
+      "loss": 0.8948,
+      "step": 10011
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 8.261990579499842e-05,
+      "loss": 0.9722,
+      "step": 10012
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 8.260272030182377e-05,
+      "loss": 0.7198,
+      "step": 10013
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 8.258553533848282e-05,
+      "loss": 0.7641,
+      "step": 10014
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 8.256835090549893e-05,
+      "loss": 0.9694,
+      "step": 10015
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 8.255116700339546e-05,
+      "loss": 0.8534,
+      "step": 10016
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 8.253398363269571e-05,
+      "loss": 0.9225,
+      "step": 10017
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 8.25168007939231e-05,
+      "loss": 0.8816,
+      "step": 10018
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 8.249961848760083e-05,
+      "loss": 0.8981,
+      "step": 10019
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 8.248243671425218e-05,
+      "loss": 0.9217,
+      "step": 10020
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 8.246525547440051e-05,
+      "loss": 0.8599,
+      "step": 10021
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 8.244807476856899e-05,
+      "loss": 0.8215,
+      "step": 10022
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 8.243089459728092e-05,
+      "loss": 0.8824,
+      "step": 10023
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 8.241371496105951e-05,
+      "loss": 1.006,
+      "step": 10024
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 8.239653586042789e-05,
+      "loss": 0.8389,
+      "step": 10025
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 8.237935729590935e-05,
+      "loss": 0.9262,
+      "step": 10026
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 8.236217926802701e-05,
+      "loss": 0.9527,
+      "step": 10027
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 8.234500177730404e-05,
+      "loss": 0.923,
+      "step": 10028
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 8.232782482426356e-05,
+      "loss": 0.8402,
+      "step": 10029
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 8.231064840942875e-05,
+      "loss": 0.9007,
+      "step": 10030
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 8.229347253332267e-05,
+      "loss": 0.8845,
+      "step": 10031
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 8.227629719646839e-05,
+      "loss": 0.8412,
+      "step": 10032
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 8.225912239938905e-05,
+      "loss": 0.8971,
+      "step": 10033
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 8.224194814260763e-05,
+      "loss": 1.0019,
+      "step": 10034
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 8.222477442664724e-05,
+      "loss": 0.8292,
+      "step": 10035
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 8.220760125203089e-05,
+      "loss": 0.8562,
+      "step": 10036
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 8.219042861928154e-05,
+      "loss": 0.8102,
+      "step": 10037
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 8.217325652892225e-05,
+      "loss": 0.9081,
+      "step": 10038
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 8.215608498147595e-05,
+      "loss": 0.8732,
+      "step": 10039
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 8.213891397746563e-05,
+      "loss": 0.8857,
+      "step": 10040
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 8.21217435174142e-05,
+      "loss": 0.9409,
+      "step": 10041
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 8.210457360184461e-05,
+      "loss": 0.8454,
+      "step": 10042
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 8.208740423127976e-05,
+      "loss": 0.8319,
+      "step": 10043
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 8.20702354062425e-05,
+      "loss": 0.8466,
+      "step": 10044
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 8.205306712725582e-05,
+      "loss": 0.7915,
+      "step": 10045
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 8.203589939484245e-05,
+      "loss": 0.9119,
+      "step": 10046
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 8.201873220952534e-05,
+      "loss": 0.9168,
+      "step": 10047
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 8.200156557182728e-05,
+      "loss": 0.7989,
+      "step": 10048
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 8.1984399482271e-05,
+      "loss": 0.9121,
+      "step": 10049
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 8.196723394137942e-05,
+      "loss": 1.0409,
+      "step": 10050
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 8.195006894967522e-05,
+      "loss": 0.834,
+      "step": 10051
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 8.193290450768123e-05,
+      "loss": 0.8175,
+      "step": 10052
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 8.191574061592014e-05,
+      "loss": 0.7859,
+      "step": 10053
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 8.189857727491469e-05,
+      "loss": 0.8987,
+      "step": 10054
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 8.188141448518761e-05,
+      "loss": 0.7923,
+      "step": 10055
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 8.186425224726156e-05,
+      "loss": 0.8394,
+      "step": 10056
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 8.184709056165925e-05,
+      "loss": 0.7984,
+      "step": 10057
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 8.182992942890331e-05,
+      "loss": 0.8019,
+      "step": 10058
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 8.181276884951638e-05,
+      "loss": 0.8212,
+      "step": 10059
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 8.179560882402113e-05,
+      "loss": 0.8491,
+      "step": 10060
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 8.17784493529401e-05,
+      "loss": 0.8833,
+      "step": 10061
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 8.176129043679594e-05,
+      "loss": 0.7266,
+      "step": 10062
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 8.174413207611115e-05,
+      "loss": 0.884,
+      "step": 10063
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 8.172697427140843e-05,
+      "loss": 0.9393,
+      "step": 10064
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 8.17098170232102e-05,
+      "loss": 0.8056,
+      "step": 10065
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 8.169266033203896e-05,
+      "loss": 0.7935,
+      "step": 10066
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 8.167550419841732e-05,
+      "loss": 0.7986,
+      "step": 10067
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 8.165834862286766e-05,
+      "loss": 0.8817,
+      "step": 10068
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 8.164119360591257e-05,
+      "loss": 1.0272,
+      "step": 10069
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 8.162403914807446e-05,
+      "loss": 0.9107,
+      "step": 10070
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 8.160688524987569e-05,
+      "loss": 0.8409,
+      "step": 10071
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 8.158973191183879e-05,
+      "loss": 0.8398,
+      "step": 10072
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 8.15725791344861e-05,
+      "loss": 0.8979,
+      "step": 10073
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 8.155542691834005e-05,
+      "loss": 0.8989,
+      "step": 10074
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 8.153827526392297e-05,
+      "loss": 0.9595,
+      "step": 10075
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 8.152112417175726e-05,
+      "loss": 0.9259,
+      "step": 10076
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 8.150397364236521e-05,
+      "loss": 0.9191,
+      "step": 10077
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 8.148682367626916e-05,
+      "loss": 0.8219,
+      "step": 10078
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 8.146967427399142e-05,
+      "loss": 0.8664,
+      "step": 10079
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 8.145252543605422e-05,
+      "loss": 0.7773,
+      "step": 10080
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 8.143537716297992e-05,
+      "loss": 0.9805,
+      "step": 10081
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 8.141822945529072e-05,
+      "loss": 0.9369,
+      "step": 10082
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 8.140108231350882e-05,
+      "loss": 0.8244,
+      "step": 10083
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 8.13839357381565e-05,
+      "loss": 0.7649,
+      "step": 10084
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 8.13667897297559e-05,
+      "loss": 0.7898,
+      "step": 10085
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 8.134964428882925e-05,
+      "loss": 0.8185,
+      "step": 10086
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 8.13324994158987e-05,
+      "loss": 1.0114,
+      "step": 10087
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 8.131535511148638e-05,
+      "loss": 0.7354,
+      "step": 10088
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 8.129821137611444e-05,
+      "loss": 0.8446,
+      "step": 10089
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 8.128106821030493e-05,
+      "loss": 0.844,
+      "step": 10090
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 8.126392561458007e-05,
+      "loss": 0.8685,
+      "step": 10091
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 8.12467835894618e-05,
+      "loss": 0.8515,
+      "step": 10092
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 8.122964213547228e-05,
+      "loss": 0.8454,
+      "step": 10093
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 8.121250125313354e-05,
+      "loss": 0.8162,
+      "step": 10094
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 8.119536094296753e-05,
+      "loss": 0.797,
+      "step": 10095
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 8.117822120549634e-05,
+      "loss": 0.8506,
+      "step": 10096
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 8.116108204124191e-05,
+      "loss": 0.792,
+      "step": 10097
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 8.114394345072624e-05,
+      "loss": 0.7746,
+      "step": 10098
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 8.11268054344713e-05,
+      "loss": 0.8337,
+      "step": 10099
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 8.110966799299897e-05,
+      "loss": 0.7318,
+      "step": 10100
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 8.109253112683123e-05,
+      "loss": 0.809,
+      "step": 10101
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 8.107539483648994e-05,
+      "loss": 0.9833,
+      "step": 10102
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 8.105825912249701e-05,
+      "loss": 0.9334,
+      "step": 10103
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 8.104112398537432e-05,
+      "loss": 0.8947,
+      "step": 10104
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 8.102398942564367e-05,
+      "loss": 0.8402,
+      "step": 10105
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 8.100685544382696e-05,
+      "loss": 0.9008,
+      "step": 10106
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 8.098972204044593e-05,
+      "loss": 0.8274,
+      "step": 10107
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 8.097258921602245e-05,
+      "loss": 0.8188,
+      "step": 10108
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 8.09554569710782e-05,
+      "loss": 0.7782,
+      "step": 10109
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 8.09383253061351e-05,
+      "loss": 0.9156,
+      "step": 10110
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 8.092119422171477e-05,
+      "loss": 0.7167,
+      "step": 10111
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 8.090406371833892e-05,
+      "loss": 0.8558,
+      "step": 10112
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 8.088693379652938e-05,
+      "loss": 0.7534,
+      "step": 10113
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 8.08698044568077e-05,
+      "loss": 0.9002,
+      "step": 10114
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 8.085267569969568e-05,
+      "loss": 0.848,
+      "step": 10115
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 8.083554752571493e-05,
+      "loss": 0.8982,
+      "step": 10116
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 8.081841993538701e-05,
+      "loss": 0.8688,
+      "step": 10117
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 8.080129292923366e-05,
+      "loss": 0.8216,
+      "step": 10118
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 8.07841665077764e-05,
+      "loss": 0.8109,
+      "step": 10119
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 8.076704067153687e-05,
+      "loss": 0.8582,
+      "step": 10120
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 8.074991542103658e-05,
+      "loss": 0.8286,
+      "step": 10121
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 8.073279075679714e-05,
+      "loss": 0.8233,
+      "step": 10122
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 8.071566667934006e-05,
+      "loss": 0.8309,
+      "step": 10123
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 8.069854318918682e-05,
+      "loss": 0.9493,
+      "step": 10124
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 8.068142028685895e-05,
+      "loss": 0.9943,
+      "step": 10125
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 8.066429797287791e-05,
+      "loss": 0.7309,
+      "step": 10126
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 8.064717624776519e-05,
+      "loss": 0.8855,
+      "step": 10127
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 8.06300551120422e-05,
+      "loss": 0.9125,
+      "step": 10128
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 8.061293456623036e-05,
+      "loss": 0.8142,
+      "step": 10129
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 8.059581461085109e-05,
+      "loss": 0.9759,
+      "step": 10130
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 8.057869524642578e-05,
+      "loss": 0.8065,
+      "step": 10131
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 8.056157647347581e-05,
+      "loss": 0.8918,
+      "step": 10132
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 8.054445829252252e-05,
+      "loss": 0.8901,
+      "step": 10133
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 8.052734070408721e-05,
+      "loss": 0.8507,
+      "step": 10134
+    },
+    {
+      "epoch": 2.22,
+      "eval_loss": 0.570040762424469,
+      "eval_runtime": 321.9396,
+      "eval_samples_per_second": 56.343,
+      "eval_steps_per_second": 28.173,
+      "step": 10134
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 8.051022370869126e-05,
+      "loss": 0.7621,
+      "step": 10135
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 8.049310730685588e-05,
+      "loss": 0.8964,
+      "step": 10136
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 8.047599149910248e-05,
+      "loss": 0.8233,
+      "step": 10137
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 8.045887628595219e-05,
+      "loss": 0.8047,
+      "step": 10138
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 8.044176166792635e-05,
+      "loss": 0.8424,
+      "step": 10139
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 8.042464764554616e-05,
+      "loss": 0.8455,
+      "step": 10140
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 8.040753421933274e-05,
+      "loss": 0.9393,
+      "step": 10141
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 8.03904213898074e-05,
+      "loss": 0.8456,
+      "step": 10142
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 8.037330915749124e-05,
+      "loss": 0.864,
+      "step": 10143
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 8.035619752290546e-05,
+      "loss": 0.8522,
+      "step": 10144
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 8.033908648657116e-05,
+      "loss": 0.8832,
+      "step": 10145
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 8.032197604900946e-05,
+      "loss": 0.8436,
+      "step": 10146
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 8.030486621074148e-05,
+      "loss": 0.8594,
+      "step": 10147
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 8.028775697228825e-05,
+      "loss": 0.9358,
+      "step": 10148
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 8.02706483341709e-05,
+      "loss": 0.9397,
+      "step": 10149
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 8.025354029691042e-05,
+      "loss": 0.852,
+      "step": 10150
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 8.023643286102785e-05,
+      "loss": 0.8647,
+      "step": 10151
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 8.02193260270442e-05,
+      "loss": 0.9031,
+      "step": 10152
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 8.020221979548044e-05,
+      "loss": 0.9302,
+      "step": 10153
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 8.018511416685759e-05,
+      "loss": 0.8508,
+      "step": 10154
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 8.016800914169652e-05,
+      "loss": 0.9383,
+      "step": 10155
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 8.015090472051827e-05,
+      "loss": 0.7449,
+      "step": 10156
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 8.013380090384368e-05,
+      "loss": 0.9097,
+      "step": 10157
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 8.011669769219363e-05,
+      "loss": 0.8737,
+      "step": 10158
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 8.009959508608907e-05,
+      "loss": 0.8613,
+      "step": 10159
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 8.008249308605076e-05,
+      "loss": 0.8625,
+      "step": 10160
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 8.006539169259966e-05,
+      "loss": 0.9076,
+      "step": 10161
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 8.004829090625654e-05,
+      "loss": 0.8537,
+      "step": 10162
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 8.003119072754212e-05,
+      "loss": 0.8904,
+      "step": 10163
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 8.001409115697734e-05,
+      "loss": 0.8285,
+      "step": 10164
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 7.999699219508285e-05,
+      "loss": 0.7113,
+      "step": 10165
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 7.997989384237945e-05,
+      "loss": 0.8578,
+      "step": 10166
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 7.996279609938783e-05,
+      "loss": 0.9177,
+      "step": 10167
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 7.994569896662878e-05,
+      "loss": 0.8828,
+      "step": 10168
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 7.992860244462293e-05,
+      "loss": 0.749,
+      "step": 10169
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 7.991150653389094e-05,
+      "loss": 0.9785,
+      "step": 10170
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 7.989441123495351e-05,
+      "loss": 0.9277,
+      "step": 10171
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 7.987731654833124e-05,
+      "loss": 0.8689,
+      "step": 10172
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 7.986022247454479e-05,
+      "loss": 0.847,
+      "step": 10173
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 7.984312901411473e-05,
+      "loss": 0.8359,
+      "step": 10174
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 7.982603616756163e-05,
+      "loss": 0.7061,
+      "step": 10175
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 7.980894393540608e-05,
+      "loss": 0.9508,
+      "step": 10176
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 7.97918523181686e-05,
+      "loss": 0.7734,
+      "step": 10177
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 7.977476131636977e-05,
+      "loss": 0.8185,
+      "step": 10178
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 7.975767093053002e-05,
+      "loss": 0.8387,
+      "step": 10179
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 7.974058116116984e-05,
+      "loss": 0.9729,
+      "step": 10180
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 7.97234920088098e-05,
+      "loss": 0.8757,
+      "step": 10181
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 7.97064034739702e-05,
+      "loss": 0.7893,
+      "step": 10182
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 7.968931555717162e-05,
+      "loss": 0.7823,
+      "step": 10183
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 7.967222825893434e-05,
+      "loss": 0.8639,
+      "step": 10184
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 7.965514157977886e-05,
+      "loss": 0.7951,
+      "step": 10185
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 7.963805552022551e-05,
+      "loss": 0.7791,
+      "step": 10186
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 7.962097008079461e-05,
+      "loss": 0.9142,
+      "step": 10187
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 7.960388526200657e-05,
+      "loss": 0.8678,
+      "step": 10188
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 7.958680106438164e-05,
+      "loss": 0.9237,
+      "step": 10189
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 7.956971748844016e-05,
+      "loss": 0.8896,
+      "step": 10190
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 7.955263453470241e-05,
+      "loss": 0.9471,
+      "step": 10191
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 7.953555220368861e-05,
+      "loss": 0.8225,
+      "step": 10192
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 7.951847049591905e-05,
+      "loss": 0.9187,
+      "step": 10193
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 7.950138941191391e-05,
+      "loss": 0.9038,
+      "step": 10194
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 7.948430895219346e-05,
+      "loss": 0.7982,
+      "step": 10195
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 7.946722911727783e-05,
+      "loss": 0.8605,
+      "step": 10196
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 7.945014990768718e-05,
+      "loss": 0.8687,
+      "step": 10197
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 7.943307132394169e-05,
+      "loss": 0.8841,
+      "step": 10198
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 7.941599336656145e-05,
+      "loss": 0.8459,
+      "step": 10199
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 7.939891603606662e-05,
+      "loss": 0.8674,
+      "step": 10200
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 7.938183933297722e-05,
+      "loss": 0.8871,
+      "step": 10201
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 7.936476325781342e-05,
+      "loss": 0.8439,
+      "step": 10202
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 7.934768781109519e-05,
+      "loss": 0.8198,
+      "step": 10203
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 7.933061299334254e-05,
+      "loss": 0.8127,
+      "step": 10204
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 7.93135388050756e-05,
+      "loss": 0.8694,
+      "step": 10205
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 7.929646524681423e-05,
+      "loss": 0.8439,
+      "step": 10206
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 7.927939231907851e-05,
+      "loss": 0.8468,
+      "step": 10207
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 7.926232002238836e-05,
+      "loss": 0.7631,
+      "step": 10208
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 7.924524835726367e-05,
+      "loss": 0.905,
+      "step": 10209
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 7.922817732422443e-05,
+      "loss": 0.984,
+      "step": 10210
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 7.921110692379048e-05,
+      "loss": 0.8401,
+      "step": 10211
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 7.919403715648175e-05,
+      "loss": 0.697,
+      "step": 10212
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 7.917696802281805e-05,
+      "loss": 0.9709,
+      "step": 10213
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 7.915989952331925e-05,
+      "loss": 0.791,
+      "step": 10214
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 7.914283165850517e-05,
+      "loss": 0.6417,
+      "step": 10215
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 7.912576442889559e-05,
+      "loss": 0.8268,
+      "step": 10216
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 7.910869783501032e-05,
+      "loss": 0.8371,
+      "step": 10217
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 7.909163187736909e-05,
+      "loss": 0.7353,
+      "step": 10218
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 7.907456655649169e-05,
+      "loss": 0.9207,
+      "step": 10219
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 7.905750187289781e-05,
+      "loss": 0.7741,
+      "step": 10220
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 7.904043782710715e-05,
+      "loss": 0.9217,
+      "step": 10221
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 7.902337441963941e-05,
+      "loss": 0.8207,
+      "step": 10222
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 7.900631165101421e-05,
+      "loss": 0.7774,
+      "step": 10223
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 7.898924952175131e-05,
+      "loss": 0.8539,
+      "step": 10224
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 7.897218803237025e-05,
+      "loss": 0.8844,
+      "step": 10225
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 7.89551271833906e-05,
+      "loss": 0.716,
+      "step": 10226
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 7.893806697533205e-05,
+      "loss": 0.9191,
+      "step": 10227
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 7.892100740871408e-05,
+      "loss": 0.9564,
+      "step": 10228
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 7.89039484840563e-05,
+      "loss": 0.9114,
+      "step": 10229
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 7.88868902018782e-05,
+      "loss": 0.8978,
+      "step": 10230
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 7.886983256269932e-05,
+      "loss": 0.7946,
+      "step": 10231
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 7.885277556703914e-05,
+      "loss": 0.8835,
+      "step": 10232
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 7.883571921541711e-05,
+      "loss": 0.8247,
+      "step": 10233
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 7.88186635083527e-05,
+      "loss": 0.9008,
+      "step": 10234
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 7.880160844636533e-05,
+      "loss": 0.8545,
+      "step": 10235
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 7.878455402997444e-05,
+      "loss": 0.7258,
+      "step": 10236
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 7.87675002596994e-05,
+      "loss": 0.9143,
+      "step": 10237
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 7.875044713605956e-05,
+      "loss": 0.7645,
+      "step": 10238
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 7.873339465957431e-05,
+      "loss": 0.8422,
+      "step": 10239
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 7.871634283076295e-05,
+      "loss": 0.8352,
+      "step": 10240
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 7.869929165014483e-05,
+      "loss": 0.953,
+      "step": 10241
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 7.868224111823923e-05,
+      "loss": 0.9451,
+      "step": 10242
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 7.86651912355654e-05,
+      "loss": 0.887,
+      "step": 10243
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 7.86481420026426e-05,
+      "loss": 0.8457,
+      "step": 10244
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 7.863109341999009e-05,
+      "loss": 0.8762,
+      "step": 10245
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 7.861404548812708e-05,
+      "loss": 0.8842,
+      "step": 10246
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 7.85969982075727e-05,
+      "loss": 0.8413,
+      "step": 10247
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 7.857995157884626e-05,
+      "loss": 0.7951,
+      "step": 10248
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 7.85629056024668e-05,
+      "loss": 0.7831,
+      "step": 10249
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 7.854586027895345e-05,
+      "loss": 0.8141,
+      "step": 10250
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 7.852881560882541e-05,
+      "loss": 0.8937,
+      "step": 10251
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 7.851177159260168e-05,
+      "loss": 0.8508,
+      "step": 10252
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 7.849472823080142e-05,
+      "loss": 0.8905,
+      "step": 10253
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 7.847768552394367e-05,
+      "loss": 0.8997,
+      "step": 10254
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 7.846064347254739e-05,
+      "loss": 0.8658,
+      "step": 10255
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 7.844360207713166e-05,
+      "loss": 0.8609,
+      "step": 10256
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 7.842656133821546e-05,
+      "loss": 0.9245,
+      "step": 10257
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 7.840952125631779e-05,
+      "loss": 0.8025,
+      "step": 10258
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 7.839248183195756e-05,
+      "loss": 0.8887,
+      "step": 10259
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 7.837544306565375e-05,
+      "loss": 0.966,
+      "step": 10260
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 7.835840495792526e-05,
+      "loss": 0.9455,
+      "step": 10261
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 7.834136750929096e-05,
+      "loss": 0.8955,
+      "step": 10262
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 7.832433072026975e-05,
+      "loss": 0.6968,
+      "step": 10263
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 7.830729459138046e-05,
+      "loss": 0.7574,
+      "step": 10264
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 7.829025912314197e-05,
+      "loss": 0.858,
+      "step": 10265
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 7.827322431607307e-05,
+      "loss": 0.8365,
+      "step": 10266
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 7.825619017069254e-05,
+      "loss": 0.9163,
+      "step": 10267
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 7.823915668751917e-05,
+      "loss": 0.8437,
+      "step": 10268
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 7.822212386707168e-05,
+      "loss": 0.8495,
+      "step": 10269
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 7.82050917098689e-05,
+      "loss": 0.8066,
+      "step": 10270
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 7.818806021642946e-05,
+      "loss": 0.8746,
+      "step": 10271
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 7.817102938727202e-05,
+      "loss": 0.8029,
+      "step": 10272
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 7.815399922291538e-05,
+      "loss": 0.7877,
+      "step": 10273
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 7.813696972387806e-05,
+      "loss": 0.8379,
+      "step": 10274
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 7.811994089067879e-05,
+      "loss": 0.961,
+      "step": 10275
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 7.810291272383613e-05,
+      "loss": 0.8836,
+      "step": 10276
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 7.80858852238687e-05,
+      "loss": 0.8163,
+      "step": 10277
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 7.806885839129507e-05,
+      "loss": 0.9352,
+      "step": 10278
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 7.805183222663376e-05,
+      "loss": 0.9154,
+      "step": 10279
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 7.803480673040334e-05,
+      "loss": 0.7691,
+      "step": 10280
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 7.801778190312228e-05,
+      "loss": 0.9033,
+      "step": 10281
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 7.800075774530912e-05,
+      "loss": 0.9802,
+      "step": 10282
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 7.79837342574823e-05,
+      "loss": 0.9013,
+      "step": 10283
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 7.796671144016027e-05,
+      "loss": 0.8162,
+      "step": 10284
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 7.794968929386149e-05,
+      "loss": 0.8303,
+      "step": 10285
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 7.79326678191043e-05,
+      "loss": 0.8543,
+      "step": 10286
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 7.791564701640717e-05,
+      "loss": 0.8535,
+      "step": 10287
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 7.789862688628842e-05,
+      "loss": 0.8771,
+      "step": 10288
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 7.788160742926639e-05,
+      "loss": 0.8038,
+      "step": 10289
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 7.786458864585944e-05,
+      "loss": 0.9172,
+      "step": 10290
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 7.784757053658586e-05,
+      "loss": 0.8914,
+      "step": 10291
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 7.783055310196395e-05,
+      "loss": 0.8273,
+      "step": 10292
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 7.78135363425119e-05,
+      "loss": 0.9125,
+      "step": 10293
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 7.779652025874811e-05,
+      "loss": 0.9669,
+      "step": 10294
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 7.777950485119068e-05,
+      "loss": 0.7707,
+      "step": 10295
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 7.776249012035781e-05,
+      "loss": 0.8729,
+      "step": 10296
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 7.774547606676778e-05,
+      "loss": 0.8537,
+      "step": 10297
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 7.772846269093862e-05,
+      "loss": 0.8113,
+      "step": 10298
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 7.77114499933886e-05,
+      "loss": 0.8779,
+      "step": 10299
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 7.76944379746358e-05,
+      "loss": 0.7652,
+      "step": 10300
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 7.767742663519823e-05,
+      "loss": 0.8594,
+      "step": 10301
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 7.766041597559411e-05,
+      "loss": 0.8984,
+      "step": 10302
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 7.76434059963414e-05,
+      "loss": 0.8222,
+      "step": 10303
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 7.76263966979582e-05,
+      "loss": 0.7736,
+      "step": 10304
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 7.760938808096249e-05,
+      "loss": 0.8454,
+      "step": 10305
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 7.759238014587229e-05,
+      "loss": 0.8561,
+      "step": 10306
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 7.757537289320557e-05,
+      "loss": 0.6554,
+      "step": 10307
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 7.755836632348025e-05,
+      "loss": 0.8485,
+      "step": 10308
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 7.754136043721432e-05,
+      "loss": 0.9186,
+      "step": 10309
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 7.752435523492563e-05,
+      "loss": 1.0076,
+      "step": 10310
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 7.750735071713215e-05,
+      "loss": 0.9407,
+      "step": 10311
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 7.749034688435173e-05,
+      "loss": 0.915,
+      "step": 10312
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 7.747334373710216e-05,
+      "loss": 0.9199,
+      "step": 10313
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 7.745634127590135e-05,
+      "loss": 0.9354,
+      "step": 10314
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 7.743933950126701e-05,
+      "loss": 0.8752,
+      "step": 10315
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 7.742233841371707e-05,
+      "loss": 0.9224,
+      "step": 10316
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 7.740533801376922e-05,
+      "loss": 0.8395,
+      "step": 10317
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 7.738833830194113e-05,
+      "loss": 0.7887,
+      "step": 10318
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 7.737133927875068e-05,
+      "loss": 0.769,
+      "step": 10319
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 7.735434094471546e-05,
+      "loss": 0.9151,
+      "step": 10320
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 7.73373433003532e-05,
+      "loss": 0.844,
+      "step": 10321
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 7.732034634618155e-05,
+      "loss": 0.9153,
+      "step": 10322
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 7.730335008271817e-05,
+      "loss": 0.9332,
+      "step": 10323
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 7.728635451048067e-05,
+      "loss": 0.846,
+      "step": 10324
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 7.726935962998663e-05,
+      "loss": 0.809,
+      "step": 10325
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 7.725236544175367e-05,
+      "loss": 0.881,
+      "step": 10326
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 7.723537194629931e-05,
+      "loss": 0.8786,
+      "step": 10327
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 7.72183791441411e-05,
+      "loss": 0.846,
+      "step": 10328
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 7.720138703579655e-05,
+      "loss": 0.9569,
+      "step": 10329
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 7.718439562178316e-05,
+      "loss": 0.8048,
+      "step": 10330
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 7.716740490261842e-05,
+      "loss": 0.8877,
+      "step": 10331
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 7.715041487881974e-05,
+      "loss": 0.8643,
+      "step": 10332
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 7.71334255509046e-05,
+      "loss": 0.8756,
+      "step": 10333
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 7.711643691939038e-05,
+      "loss": 0.9876,
+      "step": 10334
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 7.709944898479445e-05,
+      "loss": 0.9168,
+      "step": 10335
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 7.708246174763423e-05,
+      "loss": 0.8823,
+      "step": 10336
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 7.706547520842699e-05,
+      "loss": 0.8965,
+      "step": 10337
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 7.704848936769015e-05,
+      "loss": 0.8252,
+      "step": 10338
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 7.703150422594091e-05,
+      "loss": 0.8543,
+      "step": 10339
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 7.701451978369667e-05,
+      "loss": 0.872,
+      "step": 10340
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 7.699753604147462e-05,
+      "loss": 0.8221,
+      "step": 10341
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 7.698055299979194e-05,
+      "loss": 0.7926,
+      "step": 10342
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 7.6963570659166e-05,
+      "loss": 0.9704,
+      "step": 10343
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 7.694658902011383e-05,
+      "loss": 0.8213,
+      "step": 10344
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 7.692960808315273e-05,
+      "loss": 0.8271,
+      "step": 10345
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 7.691262784879983e-05,
+      "loss": 1.0399,
+      "step": 10346
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 7.689564831757218e-05,
+      "loss": 0.8897,
+      "step": 10347
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 7.687866948998701e-05,
+      "loss": 0.8534,
+      "step": 10348
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 7.686169136656133e-05,
+      "loss": 0.9415,
+      "step": 10349
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 7.684471394781224e-05,
+      "loss": 0.8963,
+      "step": 10350
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 7.682773723425677e-05,
+      "loss": 0.8793,
+      "step": 10351
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 7.681076122641197e-05,
+      "loss": 0.8672,
+      "step": 10352
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 7.679378592479483e-05,
+      "loss": 0.8645,
+      "step": 10353
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 7.677681132992231e-05,
+      "loss": 0.84,
+      "step": 10354
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 7.675983744231141e-05,
+      "loss": 0.8311,
+      "step": 10355
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 7.674286426247903e-05,
+      "loss": 0.9147,
+      "step": 10356
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 7.672589179094213e-05,
+      "loss": 0.9604,
+      "step": 10357
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 7.670892002821759e-05,
+      "loss": 0.8789,
+      "step": 10358
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 7.669194897482224e-05,
+      "loss": 0.9263,
+      "step": 10359
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 7.667497863127302e-05,
+      "loss": 0.76,
+      "step": 10360
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 7.665800899808666e-05,
+      "loss": 0.9341,
+      "step": 10361
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 7.664104007578008e-05,
+      "loss": 0.7889,
+      "step": 10362
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 7.662407186486999e-05,
+      "loss": 0.9034,
+      "step": 10363
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 7.660710436587314e-05,
+      "loss": 0.836,
+      "step": 10364
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 7.659013757930637e-05,
+      "loss": 0.7474,
+      "step": 10365
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 7.65731715056863e-05,
+      "loss": 0.9121,
+      "step": 10366
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 7.65562061455297e-05,
+      "loss": 0.9339,
+      "step": 10367
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 7.653924149935321e-05,
+      "loss": 0.8263,
+      "step": 10368
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 7.652227756767354e-05,
+      "loss": 0.8431,
+      "step": 10369
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 7.650531435100727e-05,
+      "loss": 0.7718,
+      "step": 10370
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 7.648835184987102e-05,
+      "loss": 0.7642,
+      "step": 10371
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 7.647139006478143e-05,
+      "loss": 0.8755,
+      "step": 10372
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 7.645442899625501e-05,
+      "loss": 0.9176,
+      "step": 10373
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 7.643746864480836e-05,
+      "loss": 0.8607,
+      "step": 10374
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 7.642050901095797e-05,
+      "loss": 0.8612,
+      "step": 10375
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 7.640355009522035e-05,
+      "loss": 0.908,
+      "step": 10376
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 7.638659189811201e-05,
+      "loss": 0.9677,
+      "step": 10377
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 7.636963442014938e-05,
+      "loss": 0.9157,
+      "step": 10378
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 7.635267766184892e-05,
+      "loss": 0.9052,
+      "step": 10379
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 7.633572162372705e-05,
+      "loss": 0.9247,
+      "step": 10380
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 7.631876630630014e-05,
+      "loss": 0.928,
+      "step": 10381
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 7.630181171008459e-05,
+      "loss": 0.9305,
+      "step": 10382
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 7.628485783559671e-05,
+      "loss": 0.8818,
+      "step": 10383
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 7.626790468335294e-05,
+      "loss": 0.8989,
+      "step": 10384
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 7.625095225386942e-05,
+      "loss": 0.8273,
+      "step": 10385
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 7.623400054766257e-05,
+      "loss": 0.8952,
+      "step": 10386
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 7.621704956524864e-05,
+      "loss": 0.908,
+      "step": 10387
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 7.620009930714378e-05,
+      "loss": 0.8706,
+      "step": 10388
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 7.618314977386432e-05,
+      "loss": 0.9101,
+      "step": 10389
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 7.616620096592637e-05,
+      "loss": 0.8846,
+      "step": 10390
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 7.614925288384617e-05,
+      "loss": 0.9186,
+      "step": 10391
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 7.613230552813985e-05,
+      "loss": 0.9301,
+      "step": 10392
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 7.611535889932352e-05,
+      "loss": 0.8506,
+      "step": 10393
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 7.609841299791332e-05,
+      "loss": 0.8106,
+      "step": 10394
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 7.608146782442531e-05,
+      "loss": 0.8333,
+      "step": 10395
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 7.60645233793756e-05,
+      "loss": 0.8465,
+      "step": 10396
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 7.604757966328017e-05,
+      "loss": 0.7717,
+      "step": 10397
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 7.603063667665508e-05,
+      "loss": 0.8208,
+      "step": 10398
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 7.601369442001634e-05,
+      "loss": 0.8412,
+      "step": 10399
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 7.599675289387989e-05,
+      "loss": 0.9059,
+      "step": 10400
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 7.597981209876172e-05,
+      "loss": 1.0812,
+      "step": 10401
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 7.596287203517771e-05,
+      "loss": 0.9003,
+      "step": 10402
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 7.594593270364383e-05,
+      "loss": 0.8382,
+      "step": 10403
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 7.592899410467593e-05,
+      "loss": 0.9142,
+      "step": 10404
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 7.591205623878987e-05,
+      "loss": 0.8222,
+      "step": 10405
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 7.589511910650152e-05,
+      "loss": 0.8879,
+      "step": 10406
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 7.587818270832665e-05,
+      "loss": 0.8275,
+      "step": 10407
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 7.586124704478114e-05,
+      "loss": 0.7586,
+      "step": 10408
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 7.584431211638073e-05,
+      "loss": 0.9712,
+      "step": 10409
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 7.582737792364109e-05,
+      "loss": 0.7745,
+      "step": 10410
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 7.581044446707809e-05,
+      "loss": 0.8241,
+      "step": 10411
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 7.57935117472073e-05,
+      "loss": 0.8553,
+      "step": 10412
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 7.577657976454453e-05,
+      "loss": 1.0018,
+      "step": 10413
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 7.575964851960536e-05,
+      "loss": 0.8586,
+      "step": 10414
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 7.574271801290547e-05,
+      "loss": 0.9306,
+      "step": 10415
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 7.572578824496048e-05,
+      "loss": 0.8517,
+      "step": 10416
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 7.570885921628594e-05,
+      "loss": 0.9341,
+      "step": 10417
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 7.569193092739748e-05,
+      "loss": 0.799,
+      "step": 10418
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 7.567500337881061e-05,
+      "loss": 0.9171,
+      "step": 10419
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 7.565807657104089e-05,
+      "loss": 0.733,
+      "step": 10420
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 7.56411505046038e-05,
+      "loss": 0.9633,
+      "step": 10421
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 7.562422518001485e-05,
+      "loss": 0.9143,
+      "step": 10422
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 7.560730059778947e-05,
+      "loss": 0.8469,
+      "step": 10423
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 7.559037675844309e-05,
+      "loss": 0.968,
+      "step": 10424
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 7.557345366249116e-05,
+      "loss": 0.8088,
+      "step": 10425
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 7.555653131044908e-05,
+      "loss": 0.831,
+      "step": 10426
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 7.553960970283216e-05,
+      "loss": 0.782,
+      "step": 10427
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 7.552268884015581e-05,
+      "loss": 0.886,
+      "step": 10428
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 7.550576872293527e-05,
+      "loss": 1.0081,
+      "step": 10429
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 7.548884935168598e-05,
+      "loss": 0.8222,
+      "step": 10430
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 7.547193072692309e-05,
+      "loss": 0.782,
+      "step": 10431
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 7.545501284916194e-05,
+      "loss": 0.7997,
+      "step": 10432
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 7.543809571891773e-05,
+      "loss": 0.9084,
+      "step": 10433
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 7.542117933670562e-05,
+      "loss": 0.7378,
+      "step": 10434
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 7.54042637030409e-05,
+      "loss": 0.8137,
+      "step": 10435
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 7.538734881843865e-05,
+      "loss": 0.9795,
+      "step": 10436
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 7.537043468341407e-05,
+      "loss": 0.8733,
+      "step": 10437
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 7.535352129848227e-05,
+      "loss": 0.8923,
+      "step": 10438
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 7.53366086641583e-05,
+      "loss": 0.8795,
+      "step": 10439
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 7.531969678095727e-05,
+      "loss": 0.9956,
+      "step": 10440
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 7.530278564939422e-05,
+      "loss": 0.8564,
+      "step": 10441
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 7.528587526998421e-05,
+      "loss": 0.8756,
+      "step": 10442
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 7.526896564324219e-05,
+      "loss": 0.7645,
+      "step": 10443
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 7.525205676968321e-05,
+      "loss": 0.8465,
+      "step": 10444
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 7.523514864982217e-05,
+      "loss": 0.8136,
+      "step": 10445
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 7.521824128417402e-05,
+      "loss": 0.7701,
+      "step": 10446
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 7.52013346732537e-05,
+      "loss": 0.9587,
+      "step": 10447
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 7.518442881757606e-05,
+      "loss": 0.7961,
+      "step": 10448
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 7.5167523717656e-05,
+      "loss": 0.7932,
+      "step": 10449
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 7.515061937400836e-05,
+      "loss": 0.9506,
+      "step": 10450
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 7.513371578714795e-05,
+      "loss": 0.8778,
+      "step": 10451
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 7.511681295758958e-05,
+      "loss": 0.694,
+      "step": 10452
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 7.509991088584798e-05,
+      "loss": 0.8847,
+      "step": 10453
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 7.5083009572438e-05,
+      "loss": 0.8985,
+      "step": 10454
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 7.506610901787428e-05,
+      "loss": 0.8598,
+      "step": 10455
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 7.504920922267153e-05,
+      "loss": 0.7269,
+      "step": 10456
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 7.503231018734452e-05,
+      "loss": 0.7856,
+      "step": 10457
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 7.501541191240777e-05,
+      "loss": 0.9406,
+      "step": 10458
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 7.499851439837604e-05,
+      "loss": 0.8112,
+      "step": 10459
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 7.498161764576389e-05,
+      "loss": 0.875,
+      "step": 10460
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 7.496472165508592e-05,
+      "loss": 0.8556,
+      "step": 10461
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 7.494782642685672e-05,
+      "loss": 0.7595,
+      "step": 10462
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 7.493093196159076e-05,
+      "loss": 0.896,
+      "step": 10463
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 7.491403825980266e-05,
+      "loss": 0.9191,
+      "step": 10464
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 7.489714532200683e-05,
+      "loss": 0.9346,
+      "step": 10465
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 7.488025314871778e-05,
+      "loss": 0.9178,
+      "step": 10466
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 7.486336174044999e-05,
+      "loss": 0.7992,
+      "step": 10467
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 7.484647109771782e-05,
+      "loss": 0.8852,
+      "step": 10468
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 7.482958122103576e-05,
+      "loss": 0.9954,
+      "step": 10469
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 7.48126921109181e-05,
+      "loss": 0.8255,
+      "step": 10470
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 7.479580376787927e-05,
+      "loss": 0.8341,
+      "step": 10471
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 7.477891619243356e-05,
+      "loss": 0.9199,
+      "step": 10472
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 7.47620293850953e-05,
+      "loss": 0.7757,
+      "step": 10473
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 7.474514334637877e-05,
+      "loss": 0.7648,
+      "step": 10474
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 7.472825807679819e-05,
+      "loss": 0.9203,
+      "step": 10475
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 7.471137357686794e-05,
+      "loss": 0.8405,
+      "step": 10476
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 7.469448984710205e-05,
+      "loss": 0.8579,
+      "step": 10477
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 7.467760688801488e-05,
+      "loss": 0.79,
+      "step": 10478
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 7.466072470012053e-05,
+      "loss": 0.8946,
+      "step": 10479
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 7.464384328393308e-05,
+      "loss": 0.7889,
+      "step": 10480
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 7.462696263996676e-05,
+      "loss": 0.987,
+      "step": 10481
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 7.461008276873561e-05,
+      "loss": 0.8345,
+      "step": 10482
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 7.459320367075375e-05,
+      "loss": 0.9527,
+      "step": 10483
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 7.45763253465352e-05,
+      "loss": 0.8864,
+      "step": 10484
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 7.455944779659396e-05,
+      "loss": 0.7748,
+      "step": 10485
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 7.454257102144411e-05,
+      "loss": 0.8116,
+      "step": 10486
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 7.452569502159957e-05,
+      "loss": 0.9038,
+      "step": 10487
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 7.450881979757432e-05,
+      "loss": 0.8548,
+      "step": 10488
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 7.449194534988228e-05,
+      "loss": 0.8273,
+      "step": 10489
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 7.447507167903741e-05,
+      "loss": 0.8537,
+      "step": 10490
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 7.445819878555355e-05,
+      "loss": 0.9028,
+      "step": 10491
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 7.444132666994457e-05,
+      "loss": 0.8729,
+      "step": 10492
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 7.442445533272432e-05,
+      "loss": 1.003,
+      "step": 10493
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 7.440758477440659e-05,
+      "loss": 0.7422,
+      "step": 10494
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 7.439071499550522e-05,
+      "loss": 0.788,
+      "step": 10495
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 7.437384599653396e-05,
+      "loss": 0.8661,
+      "step": 10496
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 7.435697777800651e-05,
+      "loss": 0.7812,
+      "step": 10497
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 7.434011034043667e-05,
+      "loss": 0.7747,
+      "step": 10498
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 7.432324368433803e-05,
+      "loss": 0.7635,
+      "step": 10499
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 7.430637781022442e-05,
+      "loss": 0.8893,
+      "step": 10500
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 7.428951271860936e-05,
+      "loss": 0.8555,
+      "step": 10501
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 7.427264841000647e-05,
+      "loss": 0.8258,
+      "step": 10502
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 7.425578488492948e-05,
+      "loss": 0.789,
+      "step": 10503
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 7.423892214389181e-05,
+      "loss": 0.9407,
+      "step": 10504
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 7.422206018740714e-05,
+      "loss": 0.8709,
+      "step": 10505
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 7.420519901598891e-05,
+      "loss": 0.735,
+      "step": 10506
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 7.418833863015071e-05,
+      "loss": 0.9117,
+      "step": 10507
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 7.417147903040596e-05,
+      "loss": 0.7766,
+      "step": 10508
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 7.415462021726814e-05,
+      "loss": 0.8727,
+      "step": 10509
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 7.413776219125068e-05,
+      "loss": 0.8578,
+      "step": 10510
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 7.412090495286698e-05,
+      "loss": 0.7441,
+      "step": 10511
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 7.410404850263047e-05,
+      "loss": 0.7214,
+      "step": 10512
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 7.408719284105447e-05,
+      "loss": 0.9249,
+      "step": 10513
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 7.407033796865233e-05,
+      "loss": 0.816,
+      "step": 10514
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 7.405348388593739e-05,
+      "loss": 0.9271,
+      "step": 10515
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 7.403663059342288e-05,
+      "loss": 0.8631,
+      "step": 10516
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 7.401977809162212e-05,
+      "loss": 0.7918,
+      "step": 10517
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 7.400292638104834e-05,
+      "loss": 0.925,
+      "step": 10518
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 7.398607546221474e-05,
+      "loss": 0.9851,
+      "step": 10519
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 7.396922533563455e-05,
+      "loss": 0.9517,
+      "step": 10520
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 7.395237600182088e-05,
+      "loss": 0.8189,
+      "step": 10521
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 7.393552746128696e-05,
+      "loss": 0.846,
+      "step": 10522
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 7.391867971454583e-05,
+      "loss": 0.9207,
+      "step": 10523
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 7.390183276211065e-05,
+      "loss": 0.7633,
+      "step": 10524
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 7.388498660449449e-05,
+      "loss": 0.8792,
+      "step": 10525
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 7.386814124221032e-05,
+      "loss": 0.7643,
+      "step": 10526
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 7.385129667577127e-05,
+      "loss": 1.0039,
+      "step": 10527
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 7.383445290569026e-05,
+      "loss": 0.8374,
+      "step": 10528
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 7.381760993248032e-05,
+      "loss": 0.8971,
+      "step": 10529
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 7.380076775665439e-05,
+      "loss": 0.8661,
+      "step": 10530
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 7.378392637872537e-05,
+      "loss": 0.9527,
+      "step": 10531
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 7.376708579920619e-05,
+      "loss": 0.8711,
+      "step": 10532
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 7.375024601860973e-05,
+      "loss": 0.8257,
+      "step": 10533
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 7.373340703744884e-05,
+      "loss": 0.863,
+      "step": 10534
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 7.371656885623634e-05,
+      "loss": 0.7856,
+      "step": 10535
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 7.369973147548506e-05,
+      "loss": 0.9025,
+      "step": 10536
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 7.368289489570777e-05,
+      "loss": 0.9602,
+      "step": 10537
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 7.36660591174172e-05,
+      "loss": 0.8921,
+      "step": 10538
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 7.364922414112612e-05,
+      "loss": 0.821,
+      "step": 10539
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 7.363238996734721e-05,
+      "loss": 0.971,
+      "step": 10540
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 7.36155565965932e-05,
+      "loss": 0.844,
+      "step": 10541
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 7.359872402937673e-05,
+      "loss": 0.8412,
+      "step": 10542
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 7.358189226621038e-05,
+      "loss": 0.7801,
+      "step": 10543
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 7.356506130760685e-05,
+      "loss": 0.7861,
+      "step": 10544
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 7.354823115407865e-05,
+      "loss": 0.8928,
+      "step": 10545
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 7.35314018061384e-05,
+      "loss": 0.9417,
+      "step": 10546
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 7.351457326429865e-05,
+      "loss": 0.8245,
+      "step": 10547
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 7.349774552907179e-05,
+      "loss": 0.7871,
+      "step": 10548
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 7.348091860097048e-05,
+      "loss": 0.8593,
+      "step": 10549
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 7.346409248050702e-05,
+      "loss": 0.966,
+      "step": 10550
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 7.344726716819397e-05,
+      "loss": 0.8224,
+      "step": 10551
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 7.343044266454369e-05,
+      "loss": 0.854,
+      "step": 10552
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 7.34136189700686e-05,
+      "loss": 1.025,
+      "step": 10553
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 7.339679608528104e-05,
+      "loss": 0.7935,
+      "step": 10554
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 7.337997401069334e-05,
+      "loss": 0.8884,
+      "step": 10555
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 7.336315274681785e-05,
+      "loss": 0.8749,
+      "step": 10556
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 7.334633229416683e-05,
+      "loss": 0.8387,
+      "step": 10557
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 7.332951265325257e-05,
+      "loss": 0.7961,
+      "step": 10558
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 7.331269382458732e-05,
+      "loss": 0.8826,
+      "step": 10559
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 7.329587580868325e-05,
+      "loss": 0.9191,
+      "step": 10560
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 7.32790586060526e-05,
+      "loss": 0.8991,
+      "step": 10561
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 7.326224221720749e-05,
+      "loss": 0.9498,
+      "step": 10562
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 7.324542664266012e-05,
+      "loss": 0.9226,
+      "step": 10563
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 7.322861188292258e-05,
+      "loss": 0.7254,
+      "step": 10564
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 7.321179793850694e-05,
+      "loss": 0.8957,
+      "step": 10565
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 7.31949848099253e-05,
+      "loss": 0.7487,
+      "step": 10566
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 7.317817249768969e-05,
+      "loss": 0.8335,
+      "step": 10567
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 7.316136100231215e-05,
+      "loss": 0.9037,
+      "step": 10568
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 7.314455032430463e-05,
+      "loss": 0.8656,
+      "step": 10569
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 7.312774046417914e-05,
+      "loss": 0.7746,
+      "step": 10570
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 7.311093142244764e-05,
+      "loss": 0.7569,
+      "step": 10571
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 7.309412319962196e-05,
+      "loss": 0.7868,
+      "step": 10572
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 7.307731579621408e-05,
+      "loss": 0.8536,
+      "step": 10573
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 7.306050921273584e-05,
+      "loss": 0.7198,
+      "step": 10574
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 7.304370344969908e-05,
+      "loss": 0.8744,
+      "step": 10575
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 7.302689850761564e-05,
+      "loss": 0.8227,
+      "step": 10576
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 7.301009438699729e-05,
+      "loss": 0.8209,
+      "step": 10577
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 7.299329108835581e-05,
+      "loss": 0.8301,
+      "step": 10578
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 7.297648861220292e-05,
+      "loss": 0.713,
+      "step": 10579
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 7.295968695905039e-05,
+      "loss": 0.8702,
+      "step": 10580
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 7.294288612940985e-05,
+      "loss": 0.8734,
+      "step": 10581
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 7.292608612379303e-05,
+      "loss": 0.9522,
+      "step": 10582
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 7.290928694271155e-05,
+      "loss": 0.8358,
+      "step": 10583
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 7.289248858667701e-05,
+      "loss": 0.8218,
+      "step": 10584
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 7.287569105620103e-05,
+      "loss": 0.8136,
+      "step": 10585
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 7.285889435179512e-05,
+      "loss": 0.9574,
+      "step": 10586
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 7.284209847397094e-05,
+      "loss": 0.905,
+      "step": 10587
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 7.282530342323992e-05,
+      "loss": 0.7453,
+      "step": 10588
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 7.280850920011352e-05,
+      "loss": 0.8279,
+      "step": 10589
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 7.279171580510333e-05,
+      "loss": 0.8803,
+      "step": 10590
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 7.277492323872065e-05,
+      "loss": 0.8632,
+      "step": 10591
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 7.275813150147704e-05,
+      "loss": 0.8582,
+      "step": 10592
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 7.27413405938838e-05,
+      "loss": 0.8554,
+      "step": 10593
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 7.272455051645227e-05,
+      "loss": 0.8308,
+      "step": 10594
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 7.270776126969388e-05,
+      "loss": 0.9454,
+      "step": 10595
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 7.269097285411989e-05,
+      "loss": 0.8564,
+      "step": 10596
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 7.267418527024161e-05,
+      "loss": 0.793,
+      "step": 10597
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 7.26573985185703e-05,
+      "loss": 0.7322,
+      "step": 10598
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 7.26406125996172e-05,
+      "loss": 0.7655,
+      "step": 10599
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 7.262382751389354e-05,
+      "loss": 0.9553,
+      "step": 10600
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 7.260704326191048e-05,
+      "loss": 0.8302,
+      "step": 10601
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 7.25902598441792e-05,
+      "loss": 0.7595,
+      "step": 10602
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 7.257347726121083e-05,
+      "loss": 0.8041,
+      "step": 10603
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 7.255669551351652e-05,
+      "loss": 0.7871,
+      "step": 10604
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 7.253991460160729e-05,
+      "loss": 0.8509,
+      "step": 10605
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 7.252313452599426e-05,
+      "loss": 0.9768,
+      "step": 10606
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 7.250635528718845e-05,
+      "loss": 0.8118,
+      "step": 10607
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 7.248957688570083e-05,
+      "loss": 0.8619,
+      "step": 10608
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 7.247279932204247e-05,
+      "loss": 0.8479,
+      "step": 10609
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 7.245602259672428e-05,
+      "loss": 0.8303,
+      "step": 10610
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 7.243924671025718e-05,
+      "loss": 0.7781,
+      "step": 10611
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 7.242247166315212e-05,
+      "loss": 0.8067,
+      "step": 10612
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 7.24056974559199e-05,
+      "loss": 0.819,
+      "step": 10613
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 7.238892408907152e-05,
+      "loss": 0.8898,
+      "step": 10614
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 7.237215156311768e-05,
+      "loss": 0.7945,
+      "step": 10615
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 7.235537987856927e-05,
+      "loss": 0.8821,
+      "step": 10616
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 7.233860903593705e-05,
+      "loss": 0.8321,
+      "step": 10617
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 7.232183903573171e-05,
+      "loss": 0.921,
+      "step": 10618
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 7.230506987846408e-05,
+      "loss": 0.8625,
+      "step": 10619
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 7.22883015646448e-05,
+      "loss": 0.9708,
+      "step": 10620
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 7.227153409478458e-05,
+      "loss": 0.7569,
+      "step": 10621
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 7.225476746939407e-05,
+      "loss": 0.9181,
+      "step": 10622
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 7.223800168898387e-05,
+      "loss": 0.9161,
+      "step": 10623
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 7.222123675406461e-05,
+      "loss": 0.8982,
+      "step": 10624
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 7.220447266514685e-05,
+      "loss": 0.9137,
+      "step": 10625
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 7.218770942274115e-05,
+      "loss": 0.9734,
+      "step": 10626
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 7.217094702735802e-05,
+      "loss": 0.8499,
+      "step": 10627
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 7.215418547950799e-05,
+      "loss": 0.8548,
+      "step": 10628
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 7.213742477970149e-05,
+      "loss": 0.8302,
+      "step": 10629
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 7.212066492844899e-05,
+      "loss": 0.824,
+      "step": 10630
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 7.210390592626091e-05,
+      "loss": 0.7282,
+      "step": 10631
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 7.208714777364762e-05,
+      "loss": 0.8418,
+      "step": 10632
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 7.207039047111957e-05,
+      "loss": 0.893,
+      "step": 10633
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 7.205363401918701e-05,
+      "loss": 0.8319,
+      "step": 10634
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 7.203687841836027e-05,
+      "loss": 0.863,
+      "step": 10635
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 7.202012366914971e-05,
+      "loss": 0.8409,
+      "step": 10636
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 7.20033697720655e-05,
+      "loss": 0.8163,
+      "step": 10637
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 7.198661672761796e-05,
+      "loss": 0.7412,
+      "step": 10638
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 7.19698645363173e-05,
+      "loss": 0.8812,
+      "step": 10639
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 7.195311319867361e-05,
+      "loss": 0.7569,
+      "step": 10640
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 7.193636271519718e-05,
+      "loss": 0.8861,
+      "step": 10641
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 7.191961308639806e-05,
+      "loss": 1.0223,
+      "step": 10642
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 7.190286431278638e-05,
+      "loss": 0.8543,
+      "step": 10643
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 7.188611639487224e-05,
+      "loss": 0.8935,
+      "step": 10644
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 7.186936933316571e-05,
+      "loss": 0.8516,
+      "step": 10645
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 7.185262312817677e-05,
+      "loss": 0.8746,
+      "step": 10646
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 7.183587778041544e-05,
+      "loss": 0.894,
+      "step": 10647
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 7.181913329039176e-05,
+      "loss": 0.9036,
+      "step": 10648
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 7.180238965861561e-05,
+      "loss": 0.8369,
+      "step": 10649
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 7.178564688559695e-05,
+      "loss": 0.8125,
+      "step": 10650
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 7.176890497184568e-05,
+      "loss": 0.8006,
+      "step": 10651
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 7.175216391787163e-05,
+      "loss": 0.9469,
+      "step": 10652
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 7.173542372418473e-05,
+      "loss": 0.8528,
+      "step": 10653
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 7.171868439129471e-05,
+      "loss": 0.7991,
+      "step": 10654
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 7.170194591971147e-05,
+      "loss": 0.9315,
+      "step": 10655
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 7.168520830994472e-05,
+      "loss": 0.8265,
+      "step": 10656
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 7.166847156250415e-05,
+      "loss": 0.9063,
+      "step": 10657
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 7.16517356778996e-05,
+      "loss": 0.7198,
+      "step": 10658
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 7.163500065664064e-05,
+      "loss": 0.926,
+      "step": 10659
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 7.161826649923704e-05,
+      "loss": 0.7465,
+      "step": 10660
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 7.160153320619833e-05,
+      "loss": 0.9263,
+      "step": 10661
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 7.158480077803424e-05,
+      "loss": 0.8706,
+      "step": 10662
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 7.15680692152543e-05,
+      "loss": 0.9881,
+      "step": 10663
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 7.155133851836801e-05,
+      "loss": 0.8869,
+      "step": 10664
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 7.153460868788499e-05,
+      "loss": 0.7333,
+      "step": 10665
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 7.15178797243147e-05,
+      "loss": 0.8978,
+      "step": 10666
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 7.150115162816667e-05,
+      "loss": 0.8593,
+      "step": 10667
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 7.148442439995032e-05,
+      "loss": 0.8364,
+      "step": 10668
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 7.146769804017503e-05,
+      "loss": 0.874,
+      "step": 10669
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 7.145097254935029e-05,
+      "loss": 0.8502,
+      "step": 10670
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 7.14342479279854e-05,
+      "loss": 0.7636,
+      "step": 10671
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 7.141752417658976e-05,
+      "loss": 0.7795,
+      "step": 10672
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 7.140080129567266e-05,
+      "loss": 0.9655,
+      "step": 10673
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 7.138407928574342e-05,
+      "loss": 0.9494,
+      "step": 10674
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 7.13673581473113e-05,
+      "loss": 0.8488,
+      "step": 10675
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 7.13506378808855e-05,
+      "loss": 1.0184,
+      "step": 10676
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 7.133391848697531e-05,
+      "loss": 0.8675,
+      "step": 10677
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 7.131719996608982e-05,
+      "loss": 0.9517,
+      "step": 10678
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 7.130048231873833e-05,
+      "loss": 0.7773,
+      "step": 10679
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 7.128376554542987e-05,
+      "loss": 0.8485,
+      "step": 10680
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 7.126704964667354e-05,
+      "loss": 0.83,
+      "step": 10681
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 7.125033462297852e-05,
+      "loss": 0.7173,
+      "step": 10682
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 7.123362047485373e-05,
+      "loss": 0.7953,
+      "step": 10683
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 7.121690720280834e-05,
+      "loss": 0.8531,
+      "step": 10684
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 7.120019480735129e-05,
+      "loss": 0.8181,
+      "step": 10685
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 7.11834832889915e-05,
+      "loss": 0.9664,
+      "step": 10686
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 7.1166772648238e-05,
+      "loss": 0.752,
+      "step": 10687
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 7.115006288559965e-05,
+      "loss": 0.8998,
+      "step": 10688
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 7.113335400158542e-05,
+      "loss": 0.7842,
+      "step": 10689
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 7.11166459967041e-05,
+      "loss": 0.8526,
+      "step": 10690
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 7.10999388714646e-05,
+      "loss": 0.908,
+      "step": 10691
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 7.10832326263757e-05,
+      "loss": 0.7792,
+      "step": 10692
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 7.106652726194617e-05,
+      "loss": 1.0111,
+      "step": 10693
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 7.104982277868481e-05,
+      "loss": 0.8967,
+      "step": 10694
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 7.103311917710033e-05,
+      "loss": 0.7334,
+      "step": 10695
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 7.101641645770145e-05,
+      "loss": 0.8435,
+      "step": 10696
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 7.099971462099686e-05,
+      "loss": 0.7541,
+      "step": 10697
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 7.098301366749516e-05,
+      "loss": 0.923,
+      "step": 10698
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 7.096631359770505e-05,
+      "loss": 0.9229,
+      "step": 10699
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 7.094961441213508e-05,
+      "loss": 0.9065,
+      "step": 10700
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 7.093291611129387e-05,
+      "loss": 0.8867,
+      "step": 10701
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 7.091621869568992e-05,
+      "loss": 0.7917,
+      "step": 10702
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 7.089952216583176e-05,
+      "loss": 0.7926,
+      "step": 10703
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 7.088282652222789e-05,
+      "loss": 0.8466,
+      "step": 10704
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 7.086613176538675e-05,
+      "loss": 0.8416,
+      "step": 10705
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 7.084943789581684e-05,
+      "loss": 0.8402,
+      "step": 10706
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 7.083274491402649e-05,
+      "loss": 0.9166,
+      "step": 10707
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 7.081605282052417e-05,
+      "loss": 0.8764,
+      "step": 10708
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 7.079936161581822e-05,
+      "loss": 0.935,
+      "step": 10709
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 7.078267130041687e-05,
+      "loss": 0.7434,
+      "step": 10710
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 7.076598187482854e-05,
+      "loss": 0.77,
+      "step": 10711
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 7.074929333956146e-05,
+      "loss": 0.7919,
+      "step": 10712
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 7.07326056951239e-05,
+      "loss": 0.7806,
+      "step": 10713
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 7.071591894202407e-05,
+      "loss": 0.8609,
+      "step": 10714
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 7.069923308077012e-05,
+      "loss": 0.9231,
+      "step": 10715
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 7.068254811187029e-05,
+      "loss": 0.834,
+      "step": 10716
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 7.066586403583266e-05,
+      "loss": 0.8479,
+      "step": 10717
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 7.06491808531654e-05,
+      "loss": 0.8202,
+      "step": 10718
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 7.063249856437655e-05,
+      "loss": 0.8454,
+      "step": 10719
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 7.061581716997421e-05,
+      "loss": 0.799,
+      "step": 10720
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 7.059913667046636e-05,
+      "loss": 0.8343,
+      "step": 10721
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 7.058245706636105e-05,
+      "loss": 0.88,
+      "step": 10722
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 7.056577835816622e-05,
+      "loss": 0.7641,
+      "step": 10723
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 7.054910054638981e-05,
+      "loss": 0.8548,
+      "step": 10724
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 7.053242363153984e-05,
+      "loss": 0.7551,
+      "step": 10725
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 7.05157476141241e-05,
+      "loss": 0.7879,
+      "step": 10726
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 7.049907249465046e-05,
+      "loss": 0.9193,
+      "step": 10727
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 7.048239827362685e-05,
+      "loss": 0.7816,
+      "step": 10728
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 7.046572495156098e-05,
+      "loss": 0.9631,
+      "step": 10729
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 7.04490525289607e-05,
+      "loss": 0.8495,
+      "step": 10730
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 7.043238100633378e-05,
+      "loss": 0.9453,
+      "step": 10731
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 7.041571038418786e-05,
+      "loss": 0.8249,
+      "step": 10732
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 7.039904066303075e-05,
+      "loss": 0.7661,
+      "step": 10733
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 7.038237184337005e-05,
+      "loss": 0.7802,
+      "step": 10734
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 7.036570392571346e-05,
+      "loss": 0.8077,
+      "step": 10735
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 7.034903691056857e-05,
+      "loss": 0.7985,
+      "step": 10736
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 7.033237079844299e-05,
+      "loss": 0.8988,
+      "step": 10737
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 7.031570558984428e-05,
+      "loss": 0.9615,
+      "step": 10738
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 7.029904128527996e-05,
+      "loss": 0.8737,
+      "step": 10739
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 7.028237788525758e-05,
+      "loss": 0.7179,
+      "step": 10740
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 7.026571539028459e-05,
+      "loss": 0.8637,
+      "step": 10741
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 7.024905380086846e-05,
+      "loss": 0.7951,
+      "step": 10742
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 7.023239311751661e-05,
+      "loss": 0.8034,
+      "step": 10743
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 7.021573334073645e-05,
+      "loss": 0.7666,
+      "step": 10744
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 7.019907447103536e-05,
+      "loss": 0.8085,
+      "step": 10745
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 7.018241650892063e-05,
+      "loss": 0.8223,
+      "step": 10746
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 7.016575945489969e-05,
+      "loss": 0.8491,
+      "step": 10747
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 7.014910330947973e-05,
+      "loss": 0.8017,
+      "step": 10748
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 7.0132448073168e-05,
+      "loss": 0.8554,
+      "step": 10749
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 7.011579374647187e-05,
+      "loss": 0.9582,
+      "step": 10750
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 7.009914032989837e-05,
+      "loss": 0.9326,
+      "step": 10751
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 7.008248782395483e-05,
+      "loss": 0.9053,
+      "step": 10752
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 7.00658362291483e-05,
+      "loss": 0.9062,
+      "step": 10753
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 7.004918554598596e-05,
+      "loss": 0.967,
+      "step": 10754
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 7.00325357749749e-05,
+      "loss": 0.8962,
+      "step": 10755
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 7.001588691662215e-05,
+      "loss": 0.891,
+      "step": 10756
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 6.99992389714348e-05,
+      "loss": 1.0049,
+      "step": 10757
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 6.998259193991982e-05,
+      "loss": 0.7778,
+      "step": 10758
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 6.996594582258423e-05,
+      "loss": 0.8614,
+      "step": 10759
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 6.994930061993498e-05,
+      "loss": 0.8634,
+      "step": 10760
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 6.993265633247896e-05,
+      "loss": 0.7754,
+      "step": 10761
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 6.991601296072314e-05,
+      "loss": 0.6335,
+      "step": 10762
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 6.989937050517431e-05,
+      "loss": 0.8859,
+      "step": 10763
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 6.98827289663394e-05,
+      "loss": 0.9265,
+      "step": 10764
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 6.986608834472517e-05,
+      "loss": 0.9234,
+      "step": 10765
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 6.984944864083843e-05,
+      "loss": 0.8733,
+      "step": 10766
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 6.983280985518596e-05,
+      "loss": 0.8987,
+      "step": 10767
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 6.981617198827444e-05,
+      "loss": 0.9684,
+      "step": 10768
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 6.979953504061063e-05,
+      "loss": 0.7734,
+      "step": 10769
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 6.978289901270116e-05,
+      "loss": 0.8672,
+      "step": 10770
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 6.976626390505277e-05,
+      "loss": 0.7829,
+      "step": 10771
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 6.974962971817199e-05,
+      "loss": 0.8918,
+      "step": 10772
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 6.973299645256539e-05,
+      "loss": 0.8587,
+      "step": 10773
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 6.971636410873967e-05,
+      "loss": 0.7875,
+      "step": 10774
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 6.969973268720122e-05,
+      "loss": 0.6855,
+      "step": 10775
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 6.968310218845665e-05,
+      "loss": 0.8476,
+      "step": 10776
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 6.966647261301244e-05,
+      "loss": 0.9406,
+      "step": 10777
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 6.964984396137493e-05,
+      "loss": 0.7854,
+      "step": 10778
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 6.963321623405067e-05,
+      "loss": 0.8775,
+      "step": 10779
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 6.9616589431546e-05,
+      "loss": 0.7832,
+      "step": 10780
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 6.959996355436733e-05,
+      "loss": 0.8236,
+      "step": 10781
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 6.958333860302095e-05,
+      "loss": 0.8682,
+      "step": 10782
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 6.956671457801321e-05,
+      "loss": 0.7858,
+      "step": 10783
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 6.955009147985037e-05,
+      "loss": 0.7907,
+      "step": 10784
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 6.953346930903868e-05,
+      "loss": 0.803,
+      "step": 10785
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 6.95168480660844e-05,
+      "loss": 0.9189,
+      "step": 10786
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 6.95002277514937e-05,
+      "loss": 0.8427,
+      "step": 10787
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 6.948360836577278e-05,
+      "loss": 0.8841,
+      "step": 10788
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 6.946698990942777e-05,
+      "loss": 0.8459,
+      "step": 10789
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 6.945037238296475e-05,
+      "loss": 0.8643,
+      "step": 10790
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 6.943375578688987e-05,
+      "loss": 0.8351,
+      "step": 10791
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 6.94171401217091e-05,
+      "loss": 0.885,
+      "step": 10792
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 6.94005253879286e-05,
+      "loss": 0.8019,
+      "step": 10793
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 6.938391158605426e-05,
+      "loss": 0.8705,
+      "step": 10794
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 6.936729871659207e-05,
+      "loss": 0.9107,
+      "step": 10795
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 6.935068678004804e-05,
+      "loss": 0.8374,
+      "step": 10796
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 6.933407577692798e-05,
+      "loss": 0.8934,
+      "step": 10797
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 6.931746570773788e-05,
+      "loss": 0.9227,
+      "step": 10798
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 6.930085657298353e-05,
+      "loss": 0.9537,
+      "step": 10799
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 6.928424837317082e-05,
+      "loss": 0.7402,
+      "step": 10800
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 6.926764110880552e-05,
+      "loss": 0.7188,
+      "step": 10801
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 6.925103478039336e-05,
+      "loss": 0.9039,
+      "step": 10802
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 6.923442938844018e-05,
+      "loss": 0.7575,
+      "step": 10803
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 6.921782493345161e-05,
+      "loss": 0.8579,
+      "step": 10804
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 6.920122141593339e-05,
+      "loss": 0.8028,
+      "step": 10805
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 6.918461883639117e-05,
+      "loss": 0.9128,
+      "step": 10806
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 6.916801719533055e-05,
+      "loss": 0.8903,
+      "step": 10807
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 6.915141649325718e-05,
+      "loss": 0.8341,
+      "step": 10808
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 6.913481673067658e-05,
+      "loss": 0.9606,
+      "step": 10809
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 6.911821790809436e-05,
+      "loss": 0.7874,
+      "step": 10810
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 6.910162002601599e-05,
+      "loss": 0.8707,
+      "step": 10811
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 6.908502308494698e-05,
+      "loss": 0.7791,
+      "step": 10812
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 6.906842708539278e-05,
+      "loss": 0.8672,
+      "step": 10813
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 6.905183202785881e-05,
+      "loss": 0.8256,
+      "step": 10814
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 6.903523791285049e-05,
+      "loss": 0.9171,
+      "step": 10815
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 6.901864474087315e-05,
+      "loss": 0.8171,
+      "step": 10816
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 6.900205251243223e-05,
+      "loss": 0.8994,
+      "step": 10817
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 6.898546122803297e-05,
+      "loss": 0.763,
+      "step": 10818
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 6.896887088818064e-05,
+      "loss": 0.8894,
+      "step": 10819
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 6.89522814933806e-05,
+      "loss": 0.9172,
+      "step": 10820
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 6.893569304413794e-05,
+      "loss": 0.9856,
+      "step": 10821
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 6.891910554095797e-05,
+      "loss": 0.9647,
+      "step": 10822
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 6.890251898434585e-05,
+      "loss": 0.9336,
+      "step": 10823
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 6.888593337480663e-05,
+      "loss": 0.9048,
+      "step": 10824
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 6.886934871284554e-05,
+      "loss": 0.7975,
+      "step": 10825
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 6.88527649989676e-05,
+      "loss": 0.7738,
+      "step": 10826
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 6.88361822336779e-05,
+      "loss": 0.8342,
+      "step": 10827
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 6.881960041748143e-05,
+      "loss": 0.8759,
+      "step": 10828
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 6.880301955088324e-05,
+      "loss": 0.8624,
+      "step": 10829
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 6.878643963438826e-05,
+      "loss": 0.8996,
+      "step": 10830
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 6.876986066850143e-05,
+      "loss": 0.9383,
+      "step": 10831
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 6.875328265372769e-05,
+      "loss": 0.6604,
+      "step": 10832
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 6.873670559057188e-05,
+      "loss": 0.8462,
+      "step": 10833
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 6.872012947953891e-05,
+      "loss": 0.9363,
+      "step": 10834
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 6.870355432113357e-05,
+      "loss": 0.685,
+      "step": 10835
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 6.868698011586065e-05,
+      "loss": 0.8634,
+      "step": 10836
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 6.867040686422495e-05,
+      "loss": 0.912,
+      "step": 10837
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 6.865383456673113e-05,
+      "loss": 0.7687,
+      "step": 10838
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 6.863726322388404e-05,
+      "loss": 0.8692,
+      "step": 10839
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 6.862069283618823e-05,
+      "loss": 0.8327,
+      "step": 10840
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 6.860412340414837e-05,
+      "loss": 0.8568,
+      "step": 10841
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 6.858755492826916e-05,
+      "loss": 0.8506,
+      "step": 10842
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 6.85709874090551e-05,
+      "loss": 0.7923,
+      "step": 10843
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 6.85544208470108e-05,
+      "loss": 0.9311,
+      "step": 10844
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 6.85378552426408e-05,
+      "loss": 0.9913,
+      "step": 10845
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 6.852129059644959e-05,
+      "loss": 0.9874,
+      "step": 10846
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 6.850472690894167e-05,
+      "loss": 0.8079,
+      "step": 10847
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 6.848816418062142e-05,
+      "loss": 0.8116,
+      "step": 10848
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 6.847160241199335e-05,
+      "loss": 0.8881,
+      "step": 10849
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 6.845504160356176e-05,
+      "loss": 1.0085,
+      "step": 10850
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 6.843848175583108e-05,
+      "loss": 0.9248,
+      "step": 10851
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 6.842192286930561e-05,
+      "loss": 0.9746,
+      "step": 10852
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 6.840536494448962e-05,
+      "loss": 0.9221,
+      "step": 10853
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 6.838880798188745e-05,
+      "loss": 0.8619,
+      "step": 10854
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 6.837225198200327e-05,
+      "loss": 0.9194,
+      "step": 10855
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 6.835569694534134e-05,
+      "loss": 0.822,
+      "step": 10856
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 6.833914287240581e-05,
+      "loss": 0.7553,
+      "step": 10857
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 6.832258976370089e-05,
+      "loss": 0.82,
+      "step": 10858
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 6.830603761973066e-05,
+      "loss": 0.9027,
+      "step": 10859
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 6.82894864409992e-05,
+      "loss": 0.8337,
+      "step": 10860
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 6.82729362280106e-05,
+      "loss": 0.8692,
+      "step": 10861
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 6.825638698126888e-05,
+      "loss": 0.9245,
+      "step": 10862
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 6.823983870127813e-05,
+      "loss": 0.8485,
+      "step": 10863
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 6.822329138854222e-05,
+      "loss": 0.858,
+      "step": 10864
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 6.82067450435651e-05,
+      "loss": 0.7923,
+      "step": 10865
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 6.819019966685077e-05,
+      "loss": 0.7684,
+      "step": 10866
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 6.817365525890302e-05,
+      "loss": 0.815,
+      "step": 10867
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 6.815711182022581e-05,
+      "loss": 0.8693,
+      "step": 10868
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 6.814056935132296e-05,
+      "loss": 0.7676,
+      "step": 10869
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 6.812402785269816e-05,
+      "loss": 0.9264,
+      "step": 10870
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 6.81074873248553e-05,
+      "loss": 0.8144,
+      "step": 10871
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 6.809094776829805e-05,
+      "loss": 0.8209,
+      "step": 10872
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 6.807440918353016e-05,
+      "loss": 0.8917,
+      "step": 10873
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 6.805787157105529e-05,
+      "loss": 0.8434,
+      "step": 10874
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 6.804133493137712e-05,
+      "loss": 0.7705,
+      "step": 10875
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 6.802479926499928e-05,
+      "loss": 0.7285,
+      "step": 10876
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 6.800826457242528e-05,
+      "loss": 0.8304,
+      "step": 10877
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 6.799173085415879e-05,
+      "loss": 0.9082,
+      "step": 10878
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 6.797519811070327e-05,
+      "loss": 0.7968,
+      "step": 10879
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 6.795866634256227e-05,
+      "loss": 0.9628,
+      "step": 10880
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 6.794213555023924e-05,
+      "loss": 0.9024,
+      "step": 10881
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 6.79256057342376e-05,
+      "loss": 0.7771,
+      "step": 10882
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 6.790907689506084e-05,
+      "loss": 0.8398,
+      "step": 10883
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 6.789254903321225e-05,
+      "loss": 0.7972,
+      "step": 10884
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 6.78760221491953e-05,
+      "loss": 0.904,
+      "step": 10885
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 6.785949624351322e-05,
+      "loss": 0.8561,
+      "step": 10886
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 6.784297131666931e-05,
+      "loss": 0.8461,
+      "step": 10887
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 6.782644736916692e-05,
+      "loss": 0.8276,
+      "step": 10888
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 6.780992440150917e-05,
+      "loss": 0.8883,
+      "step": 10889
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 6.779340241419938e-05,
+      "loss": 0.8681,
+      "step": 10890
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 6.777688140774064e-05,
+      "loss": 0.8445,
+      "step": 10891
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 6.776036138263617e-05,
+      "loss": 0.936,
+      "step": 10892
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 6.774384233938903e-05,
+      "loss": 0.6878,
+      "step": 10893
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 6.772732427850231e-05,
+      "loss": 0.8381,
+      "step": 10894
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 6.77108072004791e-05,
+      "loss": 0.9123,
+      "step": 10895
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 6.76942911058224e-05,
+      "loss": 0.8003,
+      "step": 10896
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 6.767777599503522e-05,
+      "loss": 0.9063,
+      "step": 10897
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 6.766126186862051e-05,
+      "loss": 0.7787,
+      "step": 10898
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 6.764474872708125e-05,
+      "loss": 0.8459,
+      "step": 10899
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 6.76282365709203e-05,
+      "loss": 0.8345,
+      "step": 10900
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 6.761172540064053e-05,
+      "loss": 0.831,
+      "step": 10901
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 6.759521521674486e-05,
+      "loss": 0.9021,
+      "step": 10902
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 6.757870601973601e-05,
+      "loss": 0.8101,
+      "step": 10903
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 6.756219781011688e-05,
+      "loss": 0.8861,
+      "step": 10904
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 6.754569058839013e-05,
+      "loss": 0.8753,
+      "step": 10905
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 6.752918435505849e-05,
+      "loss": 0.7972,
+      "step": 10906
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 6.751267911062475e-05,
+      "loss": 0.847,
+      "step": 10907
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 6.749617485559145e-05,
+      "loss": 0.9027,
+      "step": 10908
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 6.747967159046134e-05,
+      "loss": 0.9123,
+      "step": 10909
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 6.7463169315737e-05,
+      "loss": 0.9146,
+      "step": 10910
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 6.744666803192091e-05,
+      "loss": 0.8275,
+      "step": 10911
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 6.743016773951576e-05,
+      "loss": 0.9081,
+      "step": 10912
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 6.741366843902395e-05,
+      "loss": 0.9588,
+      "step": 10913
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 6.739717013094803e-05,
+      "loss": 0.8186,
+      "step": 10914
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 6.738067281579046e-05,
+      "loss": 0.8439,
+      "step": 10915
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 6.73641764940536e-05,
+      "loss": 0.8734,
+      "step": 10916
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 6.734768116623991e-05,
+      "loss": 0.7624,
+      "step": 10917
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 6.733118683285172e-05,
+      "loss": 0.8689,
+      "step": 10918
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 6.73146934943914e-05,
+      "loss": 0.9096,
+      "step": 10919
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 6.729820115136121e-05,
+      "loss": 0.8984,
+      "step": 10920
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 6.728170980426347e-05,
+      "loss": 0.83,
+      "step": 10921
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 6.72652194536004e-05,
+      "loss": 0.7949,
+      "step": 10922
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 6.724873009987418e-05,
+      "loss": 1.0372,
+      "step": 10923
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 6.723224174358706e-05,
+      "loss": 0.8613,
+      "step": 10924
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 6.721575438524112e-05,
+      "loss": 0.8623,
+      "step": 10925
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 6.719926802533855e-05,
+      "loss": 0.8657,
+      "step": 10926
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 6.718278266438142e-05,
+      "loss": 0.9347,
+      "step": 10927
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 6.716629830287176e-05,
+      "loss": 0.9228,
+      "step": 10928
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 6.714981494131163e-05,
+      "loss": 0.7438,
+      "step": 10929
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 6.713333258020298e-05,
+      "loss": 0.848,
+      "step": 10930
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 6.71168512200479e-05,
+      "loss": 0.8418,
+      "step": 10931
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 6.710037086134823e-05,
+      "loss": 0.7961,
+      "step": 10932
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 6.708389150460585e-05,
+      "loss": 0.8238,
+      "step": 10933
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 6.706741315032276e-05,
+      "loss": 0.8833,
+      "step": 10934
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 6.705093579900068e-05,
+      "loss": 0.8528,
+      "step": 10935
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 6.703445945114153e-05,
+      "loss": 0.8111,
+      "step": 10936
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 6.701798410724703e-05,
+      "loss": 0.8893,
+      "step": 10937
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 6.700150976781897e-05,
+      "loss": 0.7434,
+      "step": 10938
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 6.698503643335907e-05,
+      "loss": 0.8995,
+      "step": 10939
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 6.6968564104369e-05,
+      "loss": 0.8236,
+      "step": 10940
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 6.695209278135047e-05,
+      "loss": 0.7949,
+      "step": 10941
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 6.693562246480505e-05,
+      "loss": 0.9041,
+      "step": 10942
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 6.69191531552344e-05,
+      "loss": 0.9084,
+      "step": 10943
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 6.690268485314009e-05,
+      "loss": 0.8757,
+      "step": 10944
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 6.68862175590236e-05,
+      "loss": 0.7549,
+      "step": 10945
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 6.686975127338653e-05,
+      "loss": 0.9045,
+      "step": 10946
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 6.685328599673028e-05,
+      "loss": 0.893,
+      "step": 10947
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 6.683682172955636e-05,
+      "loss": 0.8398,
+      "step": 10948
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 6.682035847236612e-05,
+      "loss": 0.8498,
+      "step": 10949
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 6.680389622566106e-05,
+      "loss": 0.9705,
+      "step": 10950
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 6.678743498994245e-05,
+      "loss": 0.9835,
+      "step": 10951
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 6.677097476571157e-05,
+      "loss": 0.9404,
+      "step": 10952
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 6.675451555346986e-05,
+      "loss": 0.8555,
+      "step": 10953
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 6.673805735371845e-05,
+      "loss": 0.8514,
+      "step": 10954
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 6.672160016695868e-05,
+      "loss": 1.0028,
+      "step": 10955
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 6.67051439936917e-05,
+      "loss": 0.8332,
+      "step": 10956
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 6.668868883441863e-05,
+      "loss": 1.0423,
+      "step": 10957
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 6.667223468964073e-05,
+      "loss": 0.9456,
+      "step": 10958
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 6.6655781559859e-05,
+      "loss": 0.8621,
+      "step": 10959
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 6.66393294455746e-05,
+      "loss": 0.828,
+      "step": 10960
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 6.662287834728855e-05,
+      "loss": 0.9517,
+      "step": 10961
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 6.660642826550183e-05,
+      "loss": 0.8306,
+      "step": 10962
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 6.65899792007155e-05,
+      "loss": 0.9164,
+      "step": 10963
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 6.657353115343044e-05,
+      "loss": 0.8863,
+      "step": 10964
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 6.655708412414765e-05,
+      "loss": 0.8838,
+      "step": 10965
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 6.654063811336794e-05,
+      "loss": 0.8908,
+      "step": 10966
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 6.652419312159226e-05,
+      "loss": 0.7741,
+      "step": 10967
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 6.650774914932139e-05,
+      "loss": 0.8612,
+      "step": 10968
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 6.649130619705612e-05,
+      "loss": 0.7637,
+      "step": 10969
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 6.647486426529726e-05,
+      "loss": 0.8628,
+      "step": 10970
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 6.64584233545455e-05,
+      "loss": 0.8017,
+      "step": 10971
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 6.644198346530161e-05,
+      "loss": 0.775,
+      "step": 10972
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 6.642554459806624e-05,
+      "loss": 0.7891,
+      "step": 10973
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 6.640910675334e-05,
+      "loss": 0.8971,
+      "step": 10974
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 6.639266993162355e-05,
+      "loss": 0.7623,
+      "step": 10975
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 6.637623413341741e-05,
+      "loss": 0.9027,
+      "step": 10976
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 6.635979935922225e-05,
+      "loss": 0.8293,
+      "step": 10977
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 6.634336560953849e-05,
+      "loss": 0.8753,
+      "step": 10978
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 6.632693288486659e-05,
+      "loss": 0.9138,
+      "step": 10979
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 6.631050118570715e-05,
+      "loss": 0.9135,
+      "step": 10980
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 6.629407051256044e-05,
+      "loss": 0.8665,
+      "step": 10981
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 6.627764086592698e-05,
+      "loss": 0.8858,
+      "step": 10982
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 6.626121224630706e-05,
+      "loss": 0.9411,
+      "step": 10983
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 6.624478465420104e-05,
+      "loss": 0.8699,
+      "step": 10984
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 6.622835809010923e-05,
+      "loss": 0.9719,
+      "step": 10985
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 6.621193255453187e-05,
+      "loss": 0.9913,
+      "step": 10986
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 6.619550804796924e-05,
+      "loss": 0.7562,
+      "step": 10987
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 6.617908457092148e-05,
+      "loss": 0.8051,
+      "step": 10988
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 6.616266212388886e-05,
+      "loss": 0.8346,
+      "step": 10989
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 6.614624070737146e-05,
+      "loss": 0.9133,
+      "step": 10990
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 6.612982032186941e-05,
+      "loss": 0.9004,
+      "step": 10991
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 6.61134009678828e-05,
+      "loss": 0.8174,
+      "step": 10992
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 6.609698264591165e-05,
+      "loss": 0.8522,
+      "step": 10993
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 6.608056535645603e-05,
+      "loss": 0.8155,
+      "step": 10994
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 6.606414910001585e-05,
+      "loss": 0.8907,
+      "step": 10995
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 6.604773387709119e-05,
+      "loss": 0.8762,
+      "step": 10996
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 6.603131968818188e-05,
+      "loss": 0.699,
+      "step": 10997
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 6.60149065337878e-05,
+      "loss": 0.9032,
+      "step": 10998
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 6.599849441440891e-05,
+      "loss": 0.9151,
+      "step": 10999
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 6.598208333054492e-05,
+      "loss": 0.8818,
+      "step": 11000
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 6.596567328269573e-05,
+      "loss": 0.914,
+      "step": 11001
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 6.594926427136108e-05,
+      "loss": 0.7849,
+      "step": 11002
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 6.593285629704064e-05,
+      "loss": 0.7549,
+      "step": 11003
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 6.59164493602342e-05,
+      "loss": 0.8287,
+      "step": 11004
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 6.590004346144138e-05,
+      "loss": 0.8588,
+      "step": 11005
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 6.588363860116188e-05,
+      "loss": 0.9219,
+      "step": 11006
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 6.586723477989524e-05,
+      "loss": 0.8183,
+      "step": 11007
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 6.585083199814108e-05,
+      "loss": 0.8358,
+      "step": 11008
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 6.583443025639894e-05,
+      "loss": 0.7771,
+      "step": 11009
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 6.581802955516832e-05,
+      "loss": 0.8186,
+      "step": 11010
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 6.580162989494872e-05,
+      "loss": 0.8162,
+      "step": 11011
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 6.578523127623957e-05,
+      "loss": 0.7982,
+      "step": 11012
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 6.576883369954032e-05,
+      "loss": 0.8593,
+      "step": 11013
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 6.575243716535035e-05,
+      "loss": 0.9452,
+      "step": 11014
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 6.573604167416898e-05,
+      "loss": 0.8452,
+      "step": 11015
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 6.571964722649557e-05,
+      "loss": 0.8489,
+      "step": 11016
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 6.57032538228294e-05,
+      "loss": 0.7747,
+      "step": 11017
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 6.568686146366976e-05,
+      "loss": 0.8802,
+      "step": 11018
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 6.567047014951584e-05,
+      "loss": 0.9378,
+      "step": 11019
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 6.565407988086682e-05,
+      "loss": 0.8376,
+      "step": 11020
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 6.563769065822193e-05,
+      "loss": 0.902,
+      "step": 11021
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 6.562130248208024e-05,
+      "loss": 0.8141,
+      "step": 11022
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 6.560491535294094e-05,
+      "loss": 0.7372,
+      "step": 11023
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 6.558852927130297e-05,
+      "loss": 0.9929,
+      "step": 11024
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 6.557214423766547e-05,
+      "loss": 0.8097,
+      "step": 11025
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 6.555576025252744e-05,
+      "loss": 0.9439,
+      "step": 11026
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 6.553937731638777e-05,
+      "loss": 0.8755,
+      "step": 11027
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 6.55229954297455e-05,
+      "loss": 0.7939,
+      "step": 11028
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 6.550661459309948e-05,
+      "loss": 0.974,
+      "step": 11029
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 6.549023480694862e-05,
+      "loss": 0.959,
+      "step": 11030
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 6.547385607179176e-05,
+      "loss": 0.8635,
+      "step": 11031
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 6.545747838812768e-05,
+      "loss": 0.7889,
+      "step": 11032
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 6.544110175645522e-05,
+      "loss": 0.8667,
+      "step": 11033
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 6.542472617727306e-05,
+      "loss": 0.7921,
+      "step": 11034
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 6.540835165108001e-05,
+      "loss": 0.838,
+      "step": 11035
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 6.539197817837468e-05,
+      "loss": 0.8554,
+      "step": 11036
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 6.537560575965572e-05,
+      "loss": 1.0207,
+      "step": 11037
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 6.535923439542182e-05,
+      "loss": 0.9144,
+      "step": 11038
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 6.534286408617148e-05,
+      "loss": 0.8499,
+      "step": 11039
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 6.532649483240334e-05,
+      "loss": 1.0015,
+      "step": 11040
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 6.531012663461585e-05,
+      "loss": 0.9122,
+      "step": 11041
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 6.52937594933076e-05,
+      "loss": 0.9609,
+      "step": 11042
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 6.527739340897698e-05,
+      "loss": 0.894,
+      "step": 11043
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 6.526102838212237e-05,
+      "loss": 0.7869,
+      "step": 11044
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 6.524466441324233e-05,
+      "loss": 0.8527,
+      "step": 11045
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 6.522830150283503e-05,
+      "loss": 0.8391,
+      "step": 11046
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 6.521193965139895e-05,
+      "loss": 0.8809,
+      "step": 11047
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 6.519557885943235e-05,
+      "loss": 0.8604,
+      "step": 11048
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 6.517921912743341e-05,
+      "loss": 0.9075,
+      "step": 11049
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 6.51628604559005e-05,
+      "loss": 0.7572,
+      "step": 11050
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 6.514650284533173e-05,
+      "loss": 0.7943,
+      "step": 11051
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 6.513014629622532e-05,
+      "loss": 0.9443,
+      "step": 11052
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 6.511379080907939e-05,
+      "loss": 0.8265,
+      "step": 11053
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 6.509743638439203e-05,
+      "loss": 0.7969,
+      "step": 11054
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 6.508108302266133e-05,
+      "loss": 0.7788,
+      "step": 11055
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 6.506473072438532e-05,
+      "loss": 1.0015,
+      "step": 11056
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 6.504837949006206e-05,
+      "loss": 0.9032,
+      "step": 11057
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 6.503202932018943e-05,
+      "loss": 0.8671,
+      "step": 11058
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 6.501568021526548e-05,
+      "loss": 0.9176,
+      "step": 11059
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 6.499933217578806e-05,
+      "loss": 0.9103,
+      "step": 11060
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 6.498298520225503e-05,
+      "loss": 0.8854,
+      "step": 11061
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 6.49666392951643e-05,
+      "loss": 0.9525,
+      "step": 11062
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 6.49502944550136e-05,
+      "loss": 0.8219,
+      "step": 11063
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 6.493395068230084e-05,
+      "loss": 0.8949,
+      "step": 11064
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 6.491760797752366e-05,
+      "loss": 1.0088,
+      "step": 11065
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 6.490126634117979e-05,
+      "loss": 0.9777,
+      "step": 11066
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 6.488492577376694e-05,
+      "loss": 0.8497,
+      "step": 11067
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 6.486858627578273e-05,
+      "loss": 0.9091,
+      "step": 11068
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 6.485224784772486e-05,
+      "loss": 0.8899,
+      "step": 11069
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 6.483591049009079e-05,
+      "loss": 0.8547,
+      "step": 11070
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 6.481957420337818e-05,
+      "loss": 0.8224,
+      "step": 11071
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 6.480323898808455e-05,
+      "loss": 0.9178,
+      "step": 11072
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 6.478690484470729e-05,
+      "loss": 0.9581,
+      "step": 11073
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 6.477057177374396e-05,
+      "loss": 0.8751,
+      "step": 11074
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 6.475423977569192e-05,
+      "loss": 0.8897,
+      "step": 11075
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 6.47379088510486e-05,
+      "loss": 0.8753,
+      "step": 11076
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 6.472157900031134e-05,
+      "loss": 0.9267,
+      "step": 11077
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 6.470525022397744e-05,
+      "loss": 0.9103,
+      "step": 11078
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 6.468892252254425e-05,
+      "loss": 0.9074,
+      "step": 11079
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 6.467259589650897e-05,
+      "loss": 0.8054,
+      "step": 11080
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 6.465627034636888e-05,
+      "loss": 0.7352,
+      "step": 11081
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 6.463994587262114e-05,
+      "loss": 0.8611,
+      "step": 11082
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 6.46236224757629e-05,
+      "loss": 0.8498,
+      "step": 11083
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 6.460730015629134e-05,
+      "loss": 0.8784,
+      "step": 11084
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 6.459097891470349e-05,
+      "loss": 0.854,
+      "step": 11085
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 6.457465875149649e-05,
+      "loss": 0.756,
+      "step": 11086
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 6.455833966716726e-05,
+      "loss": 0.8588,
+      "step": 11087
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 6.454202166221295e-05,
+      "loss": 0.7156,
+      "step": 11088
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 6.452570473713042e-05,
+      "loss": 0.8905,
+      "step": 11089
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 6.450938889241656e-05,
+      "loss": 0.8548,
+      "step": 11090
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 6.44930741285684e-05,
+      "loss": 0.7948,
+      "step": 11091
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 6.447676044608269e-05,
+      "loss": 0.9165,
+      "step": 11092
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 6.446044784545633e-05,
+      "loss": 0.8777,
+      "step": 11093
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 6.444413632718612e-05,
+      "loss": 0.908,
+      "step": 11094
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 6.442782589176875e-05,
+      "loss": 0.7307,
+      "step": 11095
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 6.441151653970104e-05,
+      "loss": 0.8472,
+      "step": 11096
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 6.439520827147965e-05,
+      "loss": 1.0116,
+      "step": 11097
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 6.437890108760128e-05,
+      "loss": 0.7209,
+      "step": 11098
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 6.436259498856255e-05,
+      "loss": 0.7454,
+      "step": 11099
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 6.434628997486001e-05,
+      "loss": 0.7676,
+      "step": 11100
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 6.43299860469903e-05,
+      "loss": 0.8737,
+      "step": 11101
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 6.43136832054499e-05,
+      "loss": 0.8608,
+      "step": 11102
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 6.429738145073539e-05,
+      "loss": 0.7779,
+      "step": 11103
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 6.428108078334315e-05,
+      "loss": 0.9449,
+      "step": 11104
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 6.426478120376969e-05,
+      "loss": 0.7115,
+      "step": 11105
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 6.424848271251136e-05,
+      "loss": 0.8325,
+      "step": 11106
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 6.423218531006454e-05,
+      "loss": 0.8716,
+      "step": 11107
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 6.421588899692562e-05,
+      "loss": 0.8413,
+      "step": 11108
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 6.419959377359079e-05,
+      "loss": 0.8584,
+      "step": 11109
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 6.418329964055646e-05,
+      "loss": 0.9717,
+      "step": 11110
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 6.416700659831879e-05,
+      "loss": 0.8273,
+      "step": 11111
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 6.415071464737394e-05,
+      "loss": 0.8975,
+      "step": 11112
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 6.413442378821822e-05,
+      "loss": 0.8364,
+      "step": 11113
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 6.41181340213476e-05,
+      "loss": 1.0005,
+      "step": 11114
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 6.410184534725829e-05,
+      "loss": 0.8787,
+      "step": 11115
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 6.408555776644635e-05,
+      "loss": 0.8569,
+      "step": 11116
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 6.40692712794078e-05,
+      "loss": 0.7121,
+      "step": 11117
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 6.405298588663867e-05,
+      "loss": 0.7735,
+      "step": 11118
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 6.403670158863488e-05,
+      "loss": 0.8751,
+      "step": 11119
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 6.402041838589243e-05,
+      "loss": 0.8553,
+      "step": 11120
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 6.400413627890718e-05,
+      "loss": 0.8292,
+      "step": 11121
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 6.398785526817502e-05,
+      "loss": 0.7829,
+      "step": 11122
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 6.397157535419178e-05,
+      "loss": 0.894,
+      "step": 11123
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 6.395529653745325e-05,
+      "loss": 0.8321,
+      "step": 11124
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 6.393901881845524e-05,
+      "loss": 1.0038,
+      "step": 11125
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 6.392274219769346e-05,
+      "loss": 0.945,
+      "step": 11126
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 6.390646667566362e-05,
+      "loss": 0.8887,
+      "step": 11127
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 6.389019225286139e-05,
+      "loss": 0.8429,
+      "step": 11128
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 6.387391892978238e-05,
+      "loss": 0.8423,
+      "step": 11129
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 6.385764670692228e-05,
+      "loss": 0.8263,
+      "step": 11130
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 6.384137558477654e-05,
+      "loss": 0.7141,
+      "step": 11131
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 6.382510556384078e-05,
+      "loss": 0.8845,
+      "step": 11132
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 6.380883664461045e-05,
+      "loss": 0.8737,
+      "step": 11133
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 6.379256882758112e-05,
+      "loss": 0.9,
+      "step": 11134
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 6.377630211324814e-05,
+      "loss": 0.7769,
+      "step": 11135
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 6.376003650210686e-05,
+      "loss": 0.9281,
+      "step": 11136
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 6.37437719946528e-05,
+      "loss": 0.843,
+      "step": 11137
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 6.372750859138115e-05,
+      "loss": 0.8566,
+      "step": 11138
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 6.371124629278731e-05,
+      "loss": 0.9924,
+      "step": 11139
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 6.369498509936653e-05,
+      "loss": 0.8752,
+      "step": 11140
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 6.367872501161397e-05,
+      "loss": 0.8712,
+      "step": 11141
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 6.366246603002493e-05,
+      "loss": 0.8409,
+      "step": 11142
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 6.36462081550945e-05,
+      "loss": 0.8657,
+      "step": 11143
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 6.362995138731788e-05,
+      "loss": 0.8481,
+      "step": 11144
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 6.361369572719012e-05,
+      "loss": 0.8151,
+      "step": 11145
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 6.359744117520629e-05,
+      "loss": 0.783,
+      "step": 11146
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 6.358118773186146e-05,
+      "loss": 0.8472,
+      "step": 11147
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 6.356493539765059e-05,
+      "loss": 0.8556,
+      "step": 11148
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 6.354868417306864e-05,
+      "loss": 0.8891,
+      "step": 11149
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 6.353243405861056e-05,
+      "loss": 0.8609,
+      "step": 11150
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 6.351618505477128e-05,
+      "loss": 0.9116,
+      "step": 11151
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 6.349993716204561e-05,
+      "loss": 0.9086,
+      "step": 11152
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 6.348369038092836e-05,
+      "loss": 0.8563,
+      "step": 11153
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 6.346744471191441e-05,
+      "loss": 0.9429,
+      "step": 11154
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 6.345120015549842e-05,
+      "loss": 0.901,
+      "step": 11155
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 6.343495671217524e-05,
+      "loss": 0.915,
+      "step": 11156
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 6.341871438243944e-05,
+      "loss": 0.8741,
+      "step": 11157
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 6.340247316678571e-05,
+      "loss": 0.8342,
+      "step": 11158
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 6.338623306570877e-05,
+      "loss": 0.827,
+      "step": 11159
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 6.336999407970307e-05,
+      "loss": 0.9299,
+      "step": 11160
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 6.335375620926329e-05,
+      "loss": 0.8462,
+      "step": 11161
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 6.333751945488389e-05,
+      "loss": 0.8309,
+      "step": 11162
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 6.332128381705937e-05,
+      "loss": 0.9742,
+      "step": 11163
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 6.330504929628421e-05,
+      "loss": 0.9492,
+      "step": 11164
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 6.328881589305278e-05,
+      "loss": 0.7871,
+      "step": 11165
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 6.327258360785951e-05,
+      "loss": 0.8219,
+      "step": 11166
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 6.325635244119874e-05,
+      "loss": 0.8649,
+      "step": 11167
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 6.32401223935648e-05,
+      "loss": 0.9612,
+      "step": 11168
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 6.322389346545196e-05,
+      "loss": 0.8677,
+      "step": 11169
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 6.320766565735448e-05,
+      "loss": 0.9092,
+      "step": 11170
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 6.319143896976658e-05,
+      "loss": 0.8579,
+      "step": 11171
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 6.317521340318243e-05,
+      "loss": 0.8202,
+      "step": 11172
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 6.31589889580962e-05,
+      "loss": 0.8004,
+      "step": 11173
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 6.314276563500201e-05,
+      "loss": 0.8246,
+      "step": 11174
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 6.31265434343939e-05,
+      "loss": 1.0147,
+      "step": 11175
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 6.311032235676595e-05,
+      "loss": 0.8834,
+      "step": 11176
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 6.309410240261215e-05,
+      "loss": 0.8947,
+      "step": 11177
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 6.307788357242652e-05,
+      "loss": 0.7552,
+      "step": 11178
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 6.306166586670292e-05,
+      "loss": 0.8153,
+      "step": 11179
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 6.304544928593539e-05,
+      "loss": 0.9307,
+      "step": 11180
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 6.302923383061772e-05,
+      "loss": 0.8692,
+      "step": 11181
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 6.301301950124369e-05,
+      "loss": 0.7556,
+      "step": 11182
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 6.299680629830727e-05,
+      "loss": 0.8705,
+      "step": 11183
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 6.298059422230208e-05,
+      "loss": 0.9111,
+      "step": 11184
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 6.296438327372197e-05,
+      "loss": 0.8824,
+      "step": 11185
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 6.294817345306058e-05,
+      "loss": 0.9651,
+      "step": 11186
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 6.293196476081157e-05,
+      "loss": 1.0311,
+      "step": 11187
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 6.291575719746863e-05,
+      "loss": 0.8677,
+      "step": 11188
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 6.28995507635253e-05,
+      "loss": 0.9412,
+      "step": 11189
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 6.288334545947522e-05,
+      "loss": 0.7838,
+      "step": 11190
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 6.286714128581186e-05,
+      "loss": 0.8892,
+      "step": 11191
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 6.285093824302872e-05,
+      "loss": 0.7541,
+      "step": 11192
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 6.283473633161931e-05,
+      "loss": 0.961,
+      "step": 11193
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 6.2818535552077e-05,
+      "loss": 0.7735,
+      "step": 11194
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 6.280233590489523e-05,
+      "loss": 0.919,
+      "step": 11195
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 6.278613739056734e-05,
+      "loss": 0.7275,
+      "step": 11196
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 6.276994000958668e-05,
+      "loss": 0.9073,
+      "step": 11197
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 6.275374376244652e-05,
+      "loss": 0.9152,
+      "step": 11198
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 6.273754864964008e-05,
+      "loss": 0.8777,
+      "step": 11199
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 6.272135467166065e-05,
+      "loss": 0.7931,
+      "step": 11200
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 6.270516182900132e-05,
+      "loss": 0.8578,
+      "step": 11201
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 6.26889701221554e-05,
+      "loss": 0.8598,
+      "step": 11202
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 6.267277955161589e-05,
+      "loss": 0.9487,
+      "step": 11203
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 6.265659011787585e-05,
+      "loss": 0.8312,
+      "step": 11204
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 6.264040182142844e-05,
+      "loss": 0.8054,
+      "step": 11205
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 6.262421466276655e-05,
+      "loss": 0.9985,
+      "step": 11206
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 6.260802864238326e-05,
+      "loss": 0.7984,
+      "step": 11207
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 6.259184376077147e-05,
+      "loss": 0.7824,
+      "step": 11208
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 6.25756600184241e-05,
+      "loss": 0.9287,
+      "step": 11209
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 6.255947741583403e-05,
+      "loss": 0.8756,
+      "step": 11210
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 6.254329595349407e-05,
+      "loss": 0.8602,
+      "step": 11211
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 6.252711563189708e-05,
+      "loss": 0.6865,
+      "step": 11212
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 6.251093645153576e-05,
+      "loss": 0.8152,
+      "step": 11213
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 6.249475841290292e-05,
+      "loss": 0.8898,
+      "step": 11214
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 6.247858151649122e-05,
+      "loss": 0.8939,
+      "step": 11215
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 6.246240576279332e-05,
+      "loss": 0.7973,
+      "step": 11216
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 6.244623115230188e-05,
+      "loss": 0.849,
+      "step": 11217
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 6.243005768550947e-05,
+      "loss": 0.9129,
+      "step": 11218
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 6.241388536290869e-05,
+      "loss": 0.8793,
+      "step": 11219
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 6.239771418499205e-05,
+      "loss": 0.8689,
+      "step": 11220
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 6.238154415225201e-05,
+      "loss": 0.9128,
+      "step": 11221
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 6.236537526518107e-05,
+      "loss": 0.7782,
+      "step": 11222
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 6.234920752427162e-05,
+      "loss": 0.7836,
+      "step": 11223
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 6.233304093001609e-05,
+      "loss": 0.9301,
+      "step": 11224
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 6.231687548290678e-05,
+      "loss": 0.8563,
+      "step": 11225
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 6.230071118343609e-05,
+      "loss": 0.9502,
+      "step": 11226
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 6.228454803209623e-05,
+      "loss": 0.7558,
+      "step": 11227
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 6.226838602937944e-05,
+      "loss": 0.9499,
+      "step": 11228
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 6.225222517577801e-05,
+      "loss": 0.8709,
+      "step": 11229
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 6.2236065471784e-05,
+      "loss": 0.9065,
+      "step": 11230
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 6.221990691788969e-05,
+      "loss": 0.8714,
+      "step": 11231
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 6.220374951458713e-05,
+      "loss": 0.7884,
+      "step": 11232
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 6.218759326236833e-05,
+      "loss": 0.7675,
+      "step": 11233
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 6.217143816172542e-05,
+      "loss": 0.9571,
+      "step": 11234
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 6.215528421315036e-05,
+      "loss": 0.7545,
+      "step": 11235
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 6.213913141713513e-05,
+      "loss": 0.8955,
+      "step": 11236
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 6.212297977417168e-05,
+      "loss": 0.822,
+      "step": 11237
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 6.210682928475185e-05,
+      "loss": 0.8478,
+      "step": 11238
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 6.209067994936756e-05,
+      "loss": 0.8108,
+      "step": 11239
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 6.207453176851059e-05,
+      "loss": 0.8688,
+      "step": 11240
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 6.20583847426728e-05,
+      "loss": 0.9381,
+      "step": 11241
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 6.204223887234588e-05,
+      "loss": 0.9207,
+      "step": 11242
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 6.202609415802158e-05,
+      "loss": 0.7617,
+      "step": 11243
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 6.200995060019162e-05,
+      "loss": 0.729,
+      "step": 11244
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 6.199380819934758e-05,
+      "loss": 0.8672,
+      "step": 11245
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 6.197766695598113e-05,
+      "loss": 0.8055,
+      "step": 11246
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 6.196152687058378e-05,
+      "loss": 0.78,
+      "step": 11247
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 6.194538794364721e-05,
+      "loss": 0.7534,
+      "step": 11248
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 6.192925017566284e-05,
+      "loss": 0.859,
+      "step": 11249
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 6.191311356712211e-05,
+      "loss": 0.8781,
+      "step": 11250
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 6.189697811851655e-05,
+      "loss": 0.8212,
+      "step": 11251
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 6.188084383033748e-05,
+      "loss": 0.7888,
+      "step": 11252
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 6.186471070307636e-05,
+      "loss": 0.9019,
+      "step": 11253
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 6.184857873722443e-05,
+      "loss": 0.7906,
+      "step": 11254
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 6.183244793327306e-05,
+      "loss": 0.8247,
+      "step": 11255
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 6.18163182917135e-05,
+      "loss": 0.8021,
+      "step": 11256
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 6.180018981303692e-05,
+      "loss": 0.8667,
+      "step": 11257
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 6.178406249773461e-05,
+      "loss": 0.8389,
+      "step": 11258
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 6.176793634629762e-05,
+      "loss": 0.8511,
+      "step": 11259
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 6.175181135921716e-05,
+      "loss": 0.7542,
+      "step": 11260
+    },
+    {
+      "epoch": 2.47,
+      "eval_loss": 0.5662412047386169,
+      "eval_runtime": 321.8066,
+      "eval_samples_per_second": 56.366,
+      "eval_steps_per_second": 28.185,
+      "step": 11260
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 6.173568753698429e-05,
+      "loss": 0.8288,
+      "step": 11261
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 6.171956488009e-05,
+      "loss": 0.8497,
+      "step": 11262
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 6.170344338902541e-05,
+      "loss": 0.8777,
+      "step": 11263
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 6.168732306428142e-05,
+      "loss": 0.9069,
+      "step": 11264
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 6.167120390634901e-05,
+      "loss": 0.7924,
+      "step": 11265
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 6.16550859157191e-05,
+      "loss": 0.854,
+      "step": 11266
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 6.163896909288249e-05,
+      "loss": 0.9035,
+      "step": 11267
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 6.162285343833013e-05,
+      "loss": 0.9587,
+      "step": 11268
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 6.160673895255268e-05,
+      "loss": 0.7864,
+      "step": 11269
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 6.159062563604106e-05,
+      "loss": 0.9069,
+      "step": 11270
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 6.157451348928589e-05,
+      "loss": 0.8184,
+      "step": 11271
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 6.155840251277792e-05,
+      "loss": 0.9522,
+      "step": 11272
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 6.154229270700782e-05,
+      "loss": 0.8178,
+      "step": 11273
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 6.152618407246613e-05,
+      "loss": 0.9127,
+      "step": 11274
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 6.151007660964351e-05,
+      "loss": 0.8838,
+      "step": 11275
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 6.14939703190305e-05,
+      "loss": 0.8204,
+      "step": 11276
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 6.147786520111764e-05,
+      "loss": 0.9236,
+      "step": 11277
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 6.146176125639539e-05,
+      "loss": 0.9866,
+      "step": 11278
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 6.144565848535413e-05,
+      "loss": 1.0138,
+      "step": 11279
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 6.142955688848438e-05,
+      "loss": 0.9054,
+      "step": 11280
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 6.141345646627643e-05,
+      "loss": 0.9241,
+      "step": 11281
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 6.139735721922067e-05,
+      "loss": 0.8067,
+      "step": 11282
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 6.13812591478074e-05,
+      "loss": 0.8196,
+      "step": 11283
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 6.136516225252682e-05,
+      "loss": 0.8351,
+      "step": 11284
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 6.134906653386925e-05,
+      "loss": 0.7787,
+      "step": 11285
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 6.133297199232482e-05,
+      "loss": 0.8517,
+      "step": 11286
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 6.131687862838373e-05,
+      "loss": 0.942,
+      "step": 11287
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 6.130078644253608e-05,
+      "loss": 0.7255,
+      "step": 11288
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 6.128469543527197e-05,
+      "loss": 1.0504,
+      "step": 11289
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 6.126860560708145e-05,
+      "loss": 0.8989,
+      "step": 11290
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 6.125251695845451e-05,
+      "loss": 0.7723,
+      "step": 11291
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 6.123642948988115e-05,
+      "loss": 0.8657,
+      "step": 11292
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 6.12203432018513e-05,
+      "loss": 0.8648,
+      "step": 11293
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 6.120425809485493e-05,
+      "loss": 0.9202,
+      "step": 11294
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 6.118817416938182e-05,
+      "loss": 0.787,
+      "step": 11295
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 6.117209142592184e-05,
+      "loss": 0.8233,
+      "step": 11296
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 6.115600986496485e-05,
+      "loss": 0.8417,
+      "step": 11297
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 6.11399294870005e-05,
+      "loss": 0.8648,
+      "step": 11298
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 6.11238502925186e-05,
+      "loss": 0.7705,
+      "step": 11299
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 6.11077722820088e-05,
+      "loss": 0.8147,
+      "step": 11300
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 6.10916954559608e-05,
+      "loss": 0.8653,
+      "step": 11301
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 6.107561981486421e-05,
+      "loss": 0.882,
+      "step": 11302
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 6.105954535920857e-05,
+      "loss": 0.8654,
+      "step": 11303
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 6.104347208948346e-05,
+      "loss": 0.7496,
+      "step": 11304
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 6.102740000617838e-05,
+      "loss": 0.8285,
+      "step": 11305
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 6.1011329109782825e-05,
+      "loss": 0.9248,
+      "step": 11306
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 6.099525940078621e-05,
+      "loss": 0.7036,
+      "step": 11307
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 6.0979190879677936e-05,
+      "loss": 0.8811,
+      "step": 11308
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 6.09631235469474e-05,
+      "loss": 0.8432,
+      "step": 11309
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 6.0947057403083884e-05,
+      "loss": 0.8619,
+      "step": 11310
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 6.093099244857674e-05,
+      "loss": 0.9146,
+      "step": 11311
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 6.0914928683915196e-05,
+      "loss": 0.8685,
+      "step": 11312
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 6.089886610958844e-05,
+      "loss": 0.854,
+      "step": 11313
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 6.088280472608572e-05,
+      "loss": 0.8781,
+      "step": 11314
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 6.0866744533896113e-05,
+      "loss": 0.8085,
+      "step": 11315
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 6.0850685533508834e-05,
+      "loss": 0.7759,
+      "step": 11316
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 6.083462772541284e-05,
+      "loss": 0.838,
+      "step": 11317
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 6.0818571110097276e-05,
+      "loss": 0.8379,
+      "step": 11318
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 6.080251568805111e-05,
+      "loss": 0.7797,
+      "step": 11319
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 6.078646145976324e-05,
+      "loss": 0.8853,
+      "step": 11320
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 6.07704084257227e-05,
+      "loss": 0.7704,
+      "step": 11321
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 6.0754356586418326e-05,
+      "loss": 0.8294,
+      "step": 11322
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 6.0738305942339015e-05,
+      "loss": 1.0199,
+      "step": 11323
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 6.072225649397356e-05,
+      "loss": 0.7684,
+      "step": 11324
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 6.0706208241810736e-05,
+      "loss": 0.8091,
+      "step": 11325
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 6.0690161186339324e-05,
+      "loss": 0.8581,
+      "step": 11326
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 6.067411532804801e-05,
+      "loss": 0.9242,
+      "step": 11327
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 6.0658070667425485e-05,
+      "loss": 0.8887,
+      "step": 11328
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 6.064202720496039e-05,
+      "loss": 0.8488,
+      "step": 11329
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 6.062598494114131e-05,
+      "loss": 0.8703,
+      "step": 11330
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 6.0609943876456854e-05,
+      "loss": 0.8448,
+      "step": 11331
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 6.059390401139549e-05,
+      "loss": 0.8726,
+      "step": 11332
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 6.057786534644577e-05,
+      "loss": 0.8031,
+      "step": 11333
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 6.05618278820961e-05,
+      "loss": 0.8744,
+      "step": 11334
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 6.0545791618834954e-05,
+      "loss": 0.8327,
+      "step": 11335
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 6.0529756557150684e-05,
+      "loss": 0.9309,
+      "step": 11336
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 6.051372269753163e-05,
+      "loss": 0.6432,
+      "step": 11337
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 6.0497690040466126e-05,
+      "loss": 0.9306,
+      "step": 11338
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 6.048165858644239e-05,
+      "loss": 0.7217,
+      "step": 11339
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 6.0465628335948774e-05,
+      "loss": 0.8317,
+      "step": 11340
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 6.04495992894734e-05,
+      "loss": 0.8774,
+      "step": 11341
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 6.043357144750439e-05,
+      "loss": 0.9105,
+      "step": 11342
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 6.041754481052997e-05,
+      "loss": 0.7995,
+      "step": 11343
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 6.040151937903814e-05,
+      "loss": 0.8035,
+      "step": 11344
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 6.038549515351704e-05,
+      "loss": 0.7236,
+      "step": 11345
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 6.036947213445462e-05,
+      "loss": 0.8777,
+      "step": 11346
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 6.03534503223389e-05,
+      "loss": 0.8183,
+      "step": 11347
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 6.0337429717657815e-05,
+      "loss": 0.7274,
+      "step": 11348
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 6.0321410320899254e-05,
+      "loss": 0.8273,
+      "step": 11349
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 6.030539213255111e-05,
+      "loss": 0.8177,
+      "step": 11350
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 6.028937515310119e-05,
+      "loss": 0.8706,
+      "step": 11351
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 6.027335938303733e-05,
+      "loss": 0.8932,
+      "step": 11352
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 6.025734482284726e-05,
+      "loss": 0.9744,
+      "step": 11353
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 6.024133147301869e-05,
+      "loss": 0.8704,
+      "step": 11354
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 6.022531933403937e-05,
+      "loss": 0.9057,
+      "step": 11355
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 6.0209308406396855e-05,
+      "loss": 0.8341,
+      "step": 11356
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 6.019329869057884e-05,
+      "loss": 0.886,
+      "step": 11357
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 6.0177290187072874e-05,
+      "loss": 0.8141,
+      "step": 11358
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 6.0161282896366464e-05,
+      "loss": 0.7701,
+      "step": 11359
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 6.014527681894716e-05,
+      "loss": 0.8481,
+      "step": 11360
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 6.012927195530236e-05,
+      "loss": 0.8795,
+      "step": 11361
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 6.0113268305919615e-05,
+      "loss": 0.8773,
+      "step": 11362
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 6.0097265871286146e-05,
+      "loss": 0.7348,
+      "step": 11363
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 6.0081264651889457e-05,
+      "loss": 0.88,
+      "step": 11364
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 6.006526464821682e-05,
+      "loss": 0.8585,
+      "step": 11365
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 6.004926586075544e-05,
+      "loss": 0.8196,
+      "step": 11366
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 6.003326828999267e-05,
+      "loss": 0.7999,
+      "step": 11367
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 6.001727193641563e-05,
+      "loss": 0.9549,
+      "step": 11368
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 6.000127680051155e-05,
+      "loss": 0.8053,
+      "step": 11369
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 5.998528288276755e-05,
+      "loss": 0.7927,
+      "step": 11370
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 5.996929018367066e-05,
+      "loss": 0.7929,
+      "step": 11371
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 5.995329870370803e-05,
+      "loss": 0.9395,
+      "step": 11372
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 5.9937308443366604e-05,
+      "loss": 0.8181,
+      "step": 11373
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 5.992131940313342e-05,
+      "loss": 0.7575,
+      "step": 11374
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 5.9905331583495405e-05,
+      "loss": 0.7862,
+      "step": 11375
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 5.988934498493944e-05,
+      "loss": 0.8352,
+      "step": 11376
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 5.987335960795245e-05,
+      "loss": 0.809,
+      "step": 11377
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 5.985737545302121e-05,
+      "loss": 0.8779,
+      "step": 11378
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 5.9841392520632566e-05,
+      "loss": 0.8415,
+      "step": 11379
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 5.9825410811273255e-05,
+      "loss": 0.8681,
+      "step": 11380
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 5.980943032543e-05,
+      "loss": 0.9079,
+      "step": 11381
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 5.979345106358951e-05,
+      "loss": 0.8748,
+      "step": 11382
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 5.9777473026238385e-05,
+      "loss": 0.8292,
+      "step": 11383
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 5.9761496213863286e-05,
+      "loss": 0.8625,
+      "step": 11384
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 5.9745520626950715e-05,
+      "loss": 0.9947,
+      "step": 11385
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 5.972954626598732e-05,
+      "loss": 0.9397,
+      "step": 11386
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 5.971357313145952e-05,
+      "loss": 0.8301,
+      "step": 11387
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 5.969760122385375e-05,
+      "loss": 1.0268,
+      "step": 11388
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 5.968163054365654e-05,
+      "loss": 0.8366,
+      "step": 11389
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 5.966566109135415e-05,
+      "loss": 0.9894,
+      "step": 11390
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 5.9649692867433035e-05,
+      "loss": 1.0381,
+      "step": 11391
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 5.963372587237945e-05,
+      "loss": 0.9813,
+      "step": 11392
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 5.96177601066797e-05,
+      "loss": 0.8677,
+      "step": 11393
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 5.960179557082e-05,
+      "loss": 0.8322,
+      "step": 11394
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 5.9585832265286534e-05,
+      "loss": 0.9856,
+      "step": 11395
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 5.9569870190565525e-05,
+      "loss": 0.9208,
+      "step": 11396
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 5.955390934714302e-05,
+      "loss": 0.8587,
+      "step": 11397
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 5.9537949735505175e-05,
+      "loss": 0.8121,
+      "step": 11398
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 5.9521991356138e-05,
+      "loss": 0.9514,
+      "step": 11399
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 5.950603420952749e-05,
+      "loss": 0.8664,
+      "step": 11400
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 5.949007829615968e-05,
+      "loss": 0.9282,
+      "step": 11401
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 5.947412361652045e-05,
+      "loss": 0.7451,
+      "step": 11402
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 5.945817017109574e-05,
+      "loss": 0.8385,
+      "step": 11403
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 5.9442217960371395e-05,
+      "loss": 0.7562,
+      "step": 11404
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 5.94262669848332e-05,
+      "loss": 0.9293,
+      "step": 11405
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 5.941031724496702e-05,
+      "loss": 0.7799,
+      "step": 11406
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 5.939436874125852e-05,
+      "loss": 0.8611,
+      "step": 11407
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 5.9378421474193504e-05,
+      "loss": 0.9065,
+      "step": 11408
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 5.936247544425755e-05,
+      "loss": 0.8612,
+      "step": 11409
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 5.9346530651936364e-05,
+      "loss": 0.9208,
+      "step": 11410
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 5.933058709771555e-05,
+      "loss": 0.8736,
+      "step": 11411
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 5.9314644782080574e-05,
+      "loss": 0.7856,
+      "step": 11412
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 5.929870370551707e-05,
+      "loss": 0.9932,
+      "step": 11413
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 5.928276386851045e-05,
+      "loss": 0.7423,
+      "step": 11414
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 5.926682527154621e-05,
+      "loss": 0.894,
+      "step": 11415
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 5.9250887915109734e-05,
+      "loss": 0.9383,
+      "step": 11416
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 5.923495179968638e-05,
+      "loss": 0.9484,
+      "step": 11417
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 5.9219016925761516e-05,
+      "loss": 0.7307,
+      "step": 11418
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 5.9203083293820404e-05,
+      "loss": 0.9478,
+      "step": 11419
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 5.9187150904348344e-05,
+      "loss": 0.9733,
+      "step": 11420
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 5.917121975783053e-05,
+      "loss": 0.8564,
+      "step": 11421
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 5.915528985475212e-05,
+      "loss": 0.7117,
+      "step": 11422
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 5.913936119559831e-05,
+      "loss": 0.9749,
+      "step": 11423
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 5.912343378085416e-05,
+      "loss": 0.9987,
+      "step": 11424
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 5.9107507611004786e-05,
+      "loss": 0.7417,
+      "step": 11425
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 5.909158268653515e-05,
+      "loss": 0.8019,
+      "step": 11426
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 5.907565900793035e-05,
+      "loss": 0.8573,
+      "step": 11427
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 5.9059736575675275e-05,
+      "loss": 0.9677,
+      "step": 11428
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 5.904381539025481e-05,
+      "loss": 0.9022,
+      "step": 11429
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 5.902789545215389e-05,
+      "loss": 0.8359,
+      "step": 11430
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 5.9011976761857304e-05,
+      "loss": 0.822,
+      "step": 11431
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 5.899605931984995e-05,
+      "loss": 0.7368,
+      "step": 11432
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 5.898014312661653e-05,
+      "loss": 0.9396,
+      "step": 11433
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 5.896422818264171e-05,
+      "loss": 0.8853,
+      "step": 11434
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 5.8948314488410315e-05,
+      "loss": 0.7913,
+      "step": 11435
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 5.8932402044406865e-05,
+      "loss": 0.9412,
+      "step": 11436
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 5.891649085111607e-05,
+      "loss": 0.9299,
+      "step": 11437
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 5.890058090902244e-05,
+      "loss": 0.8411,
+      "step": 11438
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 5.888467221861058e-05,
+      "loss": 0.8278,
+      "step": 11439
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 5.886876478036494e-05,
+      "loss": 0.8446,
+      "step": 11440
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 5.8852858594769966e-05,
+      "loss": 0.9096,
+      "step": 11441
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 5.883695366231012e-05,
+      "loss": 0.7728,
+      "step": 11442
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 5.8821049983469754e-05,
+      "loss": 0.8197,
+      "step": 11443
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 5.8805147558733255e-05,
+      "loss": 0.9235,
+      "step": 11444
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 5.87892463885849e-05,
+      "loss": 0.8818,
+      "step": 11445
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 5.877334647350894e-05,
+      "loss": 0.9743,
+      "step": 11446
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 5.875744781398965e-05,
+      "loss": 0.8704,
+      "step": 11447
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 5.8741550410511194e-05,
+      "loss": 0.8597,
+      "step": 11448
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 5.872565426355775e-05,
+      "loss": 0.8731,
+      "step": 11449
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 5.8709759373613425e-05,
+      "loss": 0.8689,
+      "step": 11450
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 5.869386574116228e-05,
+      "loss": 0.8097,
+      "step": 11451
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 5.8677973366688385e-05,
+      "loss": 0.8681,
+      "step": 11452
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 5.866208225067569e-05,
+      "loss": 0.8616,
+      "step": 11453
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 5.864619239360825e-05,
+      "loss": 0.8907,
+      "step": 11454
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 5.863030379596988e-05,
+      "loss": 0.8014,
+      "step": 11455
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 5.861441645824455e-05,
+      "loss": 0.7982,
+      "step": 11456
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 5.859853038091612e-05,
+      "loss": 0.8863,
+      "step": 11457
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 5.8582645564468286e-05,
+      "loss": 0.806,
+      "step": 11458
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 5.8566762009384934e-05,
+      "loss": 0.7439,
+      "step": 11459
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 5.855087971614974e-05,
+      "loss": 0.8575,
+      "step": 11460
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 5.853499868524643e-05,
+      "loss": 0.9563,
+      "step": 11461
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 5.8519118917158665e-05,
+      "loss": 0.8374,
+      "step": 11462
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 5.850324041237001e-05,
+      "loss": 0.7602,
+      "step": 11463
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 5.848736317136411e-05,
+      "loss": 0.9858,
+      "step": 11464
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 5.847148719462445e-05,
+      "loss": 0.814,
+      "step": 11465
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 5.845561248263458e-05,
+      "loss": 0.8257,
+      "step": 11466
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 5.843973903587794e-05,
+      "loss": 0.7534,
+      "step": 11467
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 5.842386685483792e-05,
+      "loss": 0.8849,
+      "step": 11468
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 5.8407995939997985e-05,
+      "loss": 0.9685,
+      "step": 11469
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 5.839212629184141e-05,
+      "loss": 0.9241,
+      "step": 11470
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 5.8376257910851575e-05,
+      "loss": 0.8179,
+      "step": 11471
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 5.836039079751166e-05,
+      "loss": 0.8138,
+      "step": 11472
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 5.834452495230498e-05,
+      "loss": 0.8749,
+      "step": 11473
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 5.832866037571472e-05,
+      "loss": 0.8762,
+      "step": 11474
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 5.831279706822399e-05,
+      "loss": 0.9582,
+      "step": 11475
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 5.829693503031595e-05,
+      "loss": 0.9048,
+      "step": 11476
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 5.828107426247361e-05,
+      "loss": 0.8813,
+      "step": 11477
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 5.82652147651801e-05,
+      "loss": 1.0423,
+      "step": 11478
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 5.824935653891839e-05,
+      "loss": 0.7661,
+      "step": 11479
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 5.823349958417141e-05,
+      "loss": 0.8447,
+      "step": 11480
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 5.8217643901422126e-05,
+      "loss": 0.8773,
+      "step": 11481
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 5.820178949115335e-05,
+      "loss": 0.8537,
+      "step": 11482
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 5.8185936353848014e-05,
+      "loss": 0.8286,
+      "step": 11483
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 5.8170084489988865e-05,
+      "loss": 0.9285,
+      "step": 11484
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 5.8154233900058765e-05,
+      "loss": 0.9394,
+      "step": 11485
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 5.8138384584540325e-05,
+      "loss": 0.9267,
+      "step": 11486
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 5.812253654391624e-05,
+      "loss": 0.8898,
+      "step": 11487
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 5.8106689778669264e-05,
+      "loss": 0.7763,
+      "step": 11488
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 5.809084428928191e-05,
+      "loss": 0.8283,
+      "step": 11489
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 5.8075000076236816e-05,
+      "loss": 0.7786,
+      "step": 11490
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 5.805915714001655e-05,
+      "loss": 0.8125,
+      "step": 11491
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 5.804331548110347e-05,
+      "loss": 0.663,
+      "step": 11492
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 5.802747509998014e-05,
+      "loss": 0.832,
+      "step": 11493
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 5.8011635997128935e-05,
+      "loss": 0.8268,
+      "step": 11494
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 5.799579817303229e-05,
+      "loss": 0.8589,
+      "step": 11495
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 5.79799616281725e-05,
+      "loss": 0.825,
+      "step": 11496
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 5.796412636303188e-05,
+      "loss": 0.8695,
+      "step": 11497
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 5.794829237809269e-05,
+      "loss": 0.8342,
+      "step": 11498
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 5.7932459673837114e-05,
+      "loss": 0.9192,
+      "step": 11499
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 5.7916628250747404e-05,
+      "loss": 0.8919,
+      "step": 11500
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 5.7900798109305686e-05,
+      "loss": 0.838,
+      "step": 11501
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 5.788496924999405e-05,
+      "loss": 0.9695,
+      "step": 11502
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 5.786914167329458e-05,
+      "loss": 0.8143,
+      "step": 11503
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 5.785331537968925e-05,
+      "loss": 0.8994,
+      "step": 11504
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 5.783749036966013e-05,
+      "loss": 0.8111,
+      "step": 11505
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 5.782166664368913e-05,
+      "loss": 0.8312,
+      "step": 11506
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 5.780584420225819e-05,
+      "loss": 0.8494,
+      "step": 11507
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 5.779002304584913e-05,
+      "loss": 0.8481,
+      "step": 11508
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 5.7774203174943774e-05,
+      "loss": 0.9601,
+      "step": 11509
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 5.7758384590024003e-05,
+      "loss": 0.7943,
+      "step": 11510
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 5.774256729157148e-05,
+      "loss": 0.7923,
+      "step": 11511
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 5.772675128006804e-05,
+      "loss": 0.8883,
+      "step": 11512
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 5.771093655599522e-05,
+      "loss": 0.9069,
+      "step": 11513
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 5.769512311983469e-05,
+      "loss": 0.7972,
+      "step": 11514
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 5.767931097206813e-05,
+      "loss": 0.9182,
+      "step": 11515
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 5.766350011317699e-05,
+      "loss": 0.8643,
+      "step": 11516
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 5.764769054364288e-05,
+      "loss": 0.8898,
+      "step": 11517
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 5.7631882263947254e-05,
+      "loss": 0.8313,
+      "step": 11518
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 5.7616075274571534e-05,
+      "loss": 0.7933,
+      "step": 11519
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 5.7600269575997136e-05,
+      "loss": 0.8326,
+      "step": 11520
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 5.758446516870537e-05,
+      "loss": 0.8231,
+      "step": 11521
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 5.7568662053177636e-05,
+      "loss": 0.8657,
+      "step": 11522
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 5.75528602298952e-05,
+      "loss": 0.8891,
+      "step": 11523
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 5.753705969933929e-05,
+      "loss": 0.9337,
+      "step": 11524
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 5.7521260461991114e-05,
+      "loss": 0.9726,
+      "step": 11525
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 5.750546251833178e-05,
+      "loss": 0.8602,
+      "step": 11526
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 5.7489665868842525e-05,
+      "loss": 0.8559,
+      "step": 11527
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 5.747387051400437e-05,
+      "loss": 0.891,
+      "step": 11528
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 5.745807645429837e-05,
+      "loss": 0.9236,
+      "step": 11529
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 5.74422836902055e-05,
+      "loss": 0.9377,
+      "step": 11530
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 5.7426492222206816e-05,
+      "loss": 0.9113,
+      "step": 11531
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 5.7410702050783184e-05,
+      "loss": 1.0398,
+      "step": 11532
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 5.739491317641545e-05,
+      "loss": 0.7707,
+      "step": 11533
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 5.7379125599584614e-05,
+      "loss": 0.9165,
+      "step": 11534
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 5.73633393207713e-05,
+      "loss": 0.8155,
+      "step": 11535
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 5.734755434045642e-05,
+      "loss": 0.8831,
+      "step": 11536
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 5.733177065912064e-05,
+      "loss": 0.9389,
+      "step": 11537
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 5.731598827724462e-05,
+      "loss": 1.0216,
+      "step": 11538
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 5.730020719530912e-05,
+      "loss": 0.8544,
+      "step": 11539
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 5.7284427413794674e-05,
+      "loss": 0.9162,
+      "step": 11540
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 5.726864893318187e-05,
+      "loss": 1.0271,
+      "step": 11541
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 5.7252871753951246e-05,
+      "loss": 0.7977,
+      "step": 11542
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 5.7237095876583244e-05,
+      "loss": 0.7838,
+      "step": 11543
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 5.722132130155842e-05,
+      "loss": 0.81,
+      "step": 11544
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 5.7205548029357116e-05,
+      "loss": 0.9806,
+      "step": 11545
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 5.718977606045972e-05,
+      "loss": 1.0009,
+      "step": 11546
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 5.717400539534654e-05,
+      "loss": 0.8179,
+      "step": 11547
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 5.715823603449795e-05,
+      "loss": 0.9491,
+      "step": 11548
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 5.7142467978394144e-05,
+      "loss": 0.8989,
+      "step": 11549
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 5.712670122751535e-05,
+      "loss": 0.8869,
+      "step": 11550
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 5.7110935782341746e-05,
+      "loss": 0.919,
+      "step": 11551
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 5.709517164335343e-05,
+      "loss": 0.8109,
+      "step": 11552
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 5.7079408811030574e-05,
+      "loss": 0.8584,
+      "step": 11553
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 5.7063647285853205e-05,
+      "loss": 0.935,
+      "step": 11554
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 5.7047887068301285e-05,
+      "loss": 0.8582,
+      "step": 11555
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 5.7032128158854926e-05,
+      "loss": 0.9317,
+      "step": 11556
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 5.701637055799388e-05,
+      "loss": 0.878,
+      "step": 11557
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 5.7000614266198185e-05,
+      "loss": 0.8529,
+      "step": 11558
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 5.6984859283947655e-05,
+      "loss": 0.9086,
+      "step": 11559
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 5.696910561172206e-05,
+      "loss": 0.8926,
+      "step": 11560
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 5.695335325000129e-05,
+      "loss": 0.8207,
+      "step": 11561
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 5.693760219926498e-05,
+      "loss": 0.8047,
+      "step": 11562
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 5.692185245999289e-05,
+      "loss": 0.8506,
+      "step": 11563
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 5.690610403266461e-05,
+      "loss": 0.6818,
+      "step": 11564
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 5.689035691775983e-05,
+      "loss": 0.8898,
+      "step": 11565
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 5.6874611115758114e-05,
+      "loss": 0.8493,
+      "step": 11566
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 5.685886662713897e-05,
+      "loss": 0.845,
+      "step": 11567
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 5.684312345238192e-05,
+      "loss": 0.8985,
+      "step": 11568
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 5.682738159196637e-05,
+      "loss": 0.8122,
+      "step": 11569
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 5.6811641046371824e-05,
+      "loss": 0.7397,
+      "step": 11570
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 5.679590181607761e-05,
+      "loss": 0.8591,
+      "step": 11571
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 5.6780163901563066e-05,
+      "loss": 0.8435,
+      "step": 11572
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 5.676442730330751e-05,
+      "loss": 1.0012,
+      "step": 11573
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 5.674869202179014e-05,
+      "loss": 0.9499,
+      "step": 11574
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 5.673295805749026e-05,
+      "loss": 0.7882,
+      "step": 11575
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 5.671722541088696e-05,
+      "loss": 0.9181,
+      "step": 11576
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 5.670149408245951e-05,
+      "loss": 0.8039,
+      "step": 11577
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 5.6685764072686886e-05,
+      "loss": 0.8601,
+      "step": 11578
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 5.667003538204814e-05,
+      "loss": 0.7977,
+      "step": 11579
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 5.6654308011022364e-05,
+      "loss": 0.9753,
+      "step": 11580
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 5.663858196008845e-05,
+      "loss": 0.829,
+      "step": 11581
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 5.6622857229725444e-05,
+      "loss": 0.8777,
+      "step": 11582
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 5.6607133820412205e-05,
+      "loss": 0.8667,
+      "step": 11583
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 5.65914117326275e-05,
+      "loss": 0.8619,
+      "step": 11584
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 5.657569096685025e-05,
+      "loss": 0.8176,
+      "step": 11585
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 5.655997152355914e-05,
+      "loss": 0.8995,
+      "step": 11586
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 5.6544253403233014e-05,
+      "loss": 0.8145,
+      "step": 11587
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 5.652853660635049e-05,
+      "loss": 0.9935,
+      "step": 11588
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 5.651282113339026e-05,
+      "loss": 0.8006,
+      "step": 11589
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 5.649710698483092e-05,
+      "loss": 0.9043,
+      "step": 11590
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 5.6481394161150994e-05,
+      "loss": 0.7914,
+      "step": 11591
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 5.6465682662829114e-05,
+      "loss": 0.9258,
+      "step": 11592
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 5.6449972490343714e-05,
+      "loss": 0.9,
+      "step": 11593
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 5.643426364417328e-05,
+      "loss": 0.795,
+      "step": 11594
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 5.6418556124796185e-05,
+      "loss": 0.8561,
+      "step": 11595
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 5.6402849932690796e-05,
+      "loss": 0.8548,
+      "step": 11596
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 5.638714506833549e-05,
+      "loss": 0.9277,
+      "step": 11597
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 5.637144153220851e-05,
+      "loss": 1.0391,
+      "step": 11598
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 5.635573932478822e-05,
+      "loss": 1.036,
+      "step": 11599
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 5.6340038446552676e-05,
+      "loss": 0.7265,
+      "step": 11600
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 5.63243388979801e-05,
+      "loss": 0.9865,
+      "step": 11601
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 5.630864067954866e-05,
+      "loss": 0.864,
+      "step": 11602
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 5.6292943791736396e-05,
+      "loss": 0.8356,
+      "step": 11603
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 5.627724823502141e-05,
+      "loss": 0.844,
+      "step": 11604
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 5.626155400988172e-05,
+      "loss": 0.8817,
+      "step": 11605
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 5.624586111679519e-05,
+      "loss": 0.8712,
+      "step": 11606
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 5.623016955623983e-05,
+      "loss": 0.7256,
+      "step": 11607
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 5.6214479328693484e-05,
+      "loss": 0.9398,
+      "step": 11608
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 5.619879043463405e-05,
+      "loss": 0.7601,
+      "step": 11609
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 5.618310287453932e-05,
+      "loss": 0.8465,
+      "step": 11610
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 5.616741664888703e-05,
+      "loss": 0.8545,
+      "step": 11611
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 5.615173175815491e-05,
+      "loss": 0.9412,
+      "step": 11612
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 5.613604820282063e-05,
+      "loss": 0.8807,
+      "step": 11613
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 5.6120365983361867e-05,
+      "loss": 0.8663,
+      "step": 11614
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 5.610468510025621e-05,
+      "loss": 0.9155,
+      "step": 11615
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 5.608900555398122e-05,
+      "loss": 0.8039,
+      "step": 11616
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 5.607332734501443e-05,
+      "loss": 0.8315,
+      "step": 11617
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 5.6057650473833246e-05,
+      "loss": 0.94,
+      "step": 11618
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 5.60419749409152e-05,
+      "loss": 0.835,
+      "step": 11619
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 5.6026300746737656e-05,
+      "loss": 0.7268,
+      "step": 11620
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 5.601062789177798e-05,
+      "loss": 0.8088,
+      "step": 11621
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 5.599495637651342e-05,
+      "loss": 0.8177,
+      "step": 11622
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 5.5979286201421355e-05,
+      "loss": 0.7829,
+      "step": 11623
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 5.596361736697897e-05,
+      "loss": 0.7923,
+      "step": 11624
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 5.594794987366343e-05,
+      "loss": 0.7937,
+      "step": 11625
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 5.5932283721952005e-05,
+      "loss": 0.9915,
+      "step": 11626
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 5.5916618912321625e-05,
+      "loss": 0.8743,
+      "step": 11627
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 5.590095544524951e-05,
+      "loss": 0.8152,
+      "step": 11628
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 5.588529332121264e-05,
+      "loss": 0.7089,
+      "step": 11629
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 5.586963254068796e-05,
+      "loss": 0.8637,
+      "step": 11630
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 5.585397310415251e-05,
+      "loss": 0.9737,
+      "step": 11631
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 5.583831501208314e-05,
+      "loss": 0.883,
+      "step": 11632
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 5.582265826495674e-05,
+      "loss": 0.7002,
+      "step": 11633
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 5.580700286325011e-05,
+      "loss": 0.8355,
+      "step": 11634
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 5.5791348807440025e-05,
+      "loss": 0.9837,
+      "step": 11635
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 5.577569609800328e-05,
+      "loss": 0.8912,
+      "step": 11636
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 5.576004473541655e-05,
+      "loss": 0.9319,
+      "step": 11637
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 5.574439472015649e-05,
+      "loss": 0.7798,
+      "step": 11638
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 5.57287460526997e-05,
+      "loss": 0.8835,
+      "step": 11639
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 5.5713098733522816e-05,
+      "loss": 0.7148,
+      "step": 11640
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 5.5697452763102365e-05,
+      "loss": 0.812,
+      "step": 11641
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 5.5681808141914815e-05,
+      "loss": 0.8561,
+      "step": 11642
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 5.566616487043663e-05,
+      "loss": 0.7862,
+      "step": 11643
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 5.565052294914419e-05,
+      "loss": 0.7831,
+      "step": 11644
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 5.5634882378513954e-05,
+      "loss": 0.8548,
+      "step": 11645
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 5.561924315902222e-05,
+      "loss": 0.8285,
+      "step": 11646
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 5.560360529114521e-05,
+      "loss": 0.9304,
+      "step": 11647
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 5.558796877535933e-05,
+      "loss": 0.9161,
+      "step": 11648
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 5.557233361214063e-05,
+      "loss": 0.8181,
+      "step": 11649
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 5.555669980196536e-05,
+      "loss": 0.7404,
+      "step": 11650
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 5.5541067345309636e-05,
+      "loss": 0.8099,
+      "step": 11651
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 5.5525436242649495e-05,
+      "loss": 0.8627,
+      "step": 11652
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 5.550980649446107e-05,
+      "loss": 0.9523,
+      "step": 11653
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 5.549417810122033e-05,
+      "loss": 0.8763,
+      "step": 11654
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 5.5478551063403215e-05,
+      "loss": 0.9363,
+      "step": 11655
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 5.546292538148563e-05,
+      "loss": 0.8664,
+      "step": 11656
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 5.5447301055943514e-05,
+      "loss": 0.8718,
+      "step": 11657
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 5.543167808725268e-05,
+      "loss": 0.8778,
+      "step": 11658
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 5.541605647588892e-05,
+      "loss": 0.8757,
+      "step": 11659
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 5.540043622232799e-05,
+      "loss": 0.9228,
+      "step": 11660
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 5.5384817327045566e-05,
+      "loss": 0.7913,
+      "step": 11661
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 5.536919979051739e-05,
+      "loss": 0.8296,
+      "step": 11662
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 5.5353583613219075e-05,
+      "loss": 0.8999,
+      "step": 11663
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 5.533796879562618e-05,
+      "loss": 0.8542,
+      "step": 11664
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 5.5322355338214294e-05,
+      "loss": 0.932,
+      "step": 11665
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 5.530674324145886e-05,
+      "loss": 0.8252,
+      "step": 11666
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 5.529113250583542e-05,
+      "loss": 0.8343,
+      "step": 11667
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 5.527552313181933e-05,
+      "loss": 0.9659,
+      "step": 11668
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 5.5259915119886086e-05,
+      "loss": 0.8574,
+      "step": 11669
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 5.5244308470510916e-05,
+      "loss": 0.9192,
+      "step": 11670
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 5.522870318416912e-05,
+      "loss": 0.857,
+      "step": 11671
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 5.521309926133602e-05,
+      "loss": 0.8511,
+      "step": 11672
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 5.5197496702486785e-05,
+      "loss": 1.0279,
+      "step": 11673
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 5.518189550809665e-05,
+      "loss": 0.7646,
+      "step": 11674
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 5.516629567864073e-05,
+      "loss": 0.8778,
+      "step": 11675
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 5.515069721459404e-05,
+      "loss": 0.9036,
+      "step": 11676
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 5.5135100116431726e-05,
+      "loss": 0.8531,
+      "step": 11677
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 5.511950438462872e-05,
+      "loss": 0.9068,
+      "step": 11678
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 5.510391001966007e-05,
+      "loss": 0.8994,
+      "step": 11679
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 5.508831702200064e-05,
+      "loss": 0.8311,
+      "step": 11680
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 5.507272539212536e-05,
+      "loss": 0.882,
+      "step": 11681
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 5.5057135130509055e-05,
+      "loss": 0.8325,
+      "step": 11682
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 5.504154623762645e-05,
+      "loss": 0.7859,
+      "step": 11683
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 5.502595871395244e-05,
+      "loss": 0.7543,
+      "step": 11684
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 5.5010372559961665e-05,
+      "loss": 0.9609,
+      "step": 11685
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 5.499478777612883e-05,
+      "loss": 0.7914,
+      "step": 11686
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 5.497920436292854e-05,
+      "loss": 0.9883,
+      "step": 11687
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 5.496362232083535e-05,
+      "loss": 1.0191,
+      "step": 11688
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 5.494804165032391e-05,
+      "loss": 0.8432,
+      "step": 11689
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 5.493246235186864e-05,
+      "loss": 0.8019,
+      "step": 11690
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 5.491688442594412e-05,
+      "loss": 0.8284,
+      "step": 11691
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 5.4901307873024677e-05,
+      "loss": 0.8396,
+      "step": 11692
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 5.488573269358466e-05,
+      "loss": 0.7358,
+      "step": 11693
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 5.487015888809852e-05,
+      "loss": 0.9663,
+      "step": 11694
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 5.485458645704047e-05,
+      "loss": 0.9359,
+      "step": 11695
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 5.483901540088483e-05,
+      "loss": 0.9599,
+      "step": 11696
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 5.482344572010585e-05,
+      "loss": 0.8379,
+      "step": 11697
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 5.480787741517755e-05,
+      "loss": 0.7994,
+      "step": 11698
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 5.479231048657421e-05,
+      "loss": 0.8877,
+      "step": 11699
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 5.4776744934769806e-05,
+      "loss": 0.8908,
+      "step": 11700
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 5.47611807602385e-05,
+      "loss": 0.7574,
+      "step": 11701
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 5.474561796345426e-05,
+      "loss": 0.9434,
+      "step": 11702
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 5.4730056544891016e-05,
+      "loss": 0.81,
+      "step": 11703
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 5.4714496505022715e-05,
+      "loss": 0.809,
+      "step": 11704
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 5.469893784432321e-05,
+      "loss": 0.8481,
+      "step": 11705
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 5.46833805632664e-05,
+      "loss": 0.8337,
+      "step": 11706
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 5.4667824662326027e-05,
+      "loss": 0.8356,
+      "step": 11707
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 5.465227014197588e-05,
+      "loss": 0.8603,
+      "step": 11708
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 5.4636717002689666e-05,
+      "loss": 0.7416,
+      "step": 11709
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 5.4621165244940985e-05,
+      "loss": 0.8653,
+      "step": 11710
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 5.460561486920357e-05,
+      "loss": 0.9575,
+      "step": 11711
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 5.4590065875950925e-05,
+      "loss": 0.9314,
+      "step": 11712
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 5.457451826565672e-05,
+      "loss": 0.9894,
+      "step": 11713
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 5.455897203879428e-05,
+      "loss": 0.9093,
+      "step": 11714
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 5.454342719583718e-05,
+      "loss": 0.9114,
+      "step": 11715
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 5.4527883737258836e-05,
+      "loss": 1.0184,
+      "step": 11716
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 5.451234166353254e-05,
+      "loss": 0.829,
+      "step": 11717
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 5.4496800975131776e-05,
+      "loss": 0.9145,
+      "step": 11718
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 5.448126167252967e-05,
+      "loss": 0.8158,
+      "step": 11719
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 5.446572375619957e-05,
+      "loss": 0.828,
+      "step": 11720
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 5.4450187226614666e-05,
+      "loss": 0.8977,
+      "step": 11721
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 5.443465208424807e-05,
+      "loss": 0.8492,
+      "step": 11722
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 5.4419118329572996e-05,
+      "loss": 0.8998,
+      "step": 11723
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 5.4403585963062476e-05,
+      "loss": 0.8482,
+      "step": 11724
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 5.438805498518954e-05,
+      "loss": 0.8099,
+      "step": 11725
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 5.43725253964272e-05,
+      "loss": 0.6966,
+      "step": 11726
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 5.435699719724836e-05,
+      "loss": 0.7915,
+      "step": 11727
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 5.434147038812601e-05,
+      "loss": 0.8909,
+      "step": 11728
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 5.432594496953298e-05,
+      "loss": 0.8364,
+      "step": 11729
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 5.431042094194211e-05,
+      "loss": 0.7232,
+      "step": 11730
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 5.4294898305826127e-05,
+      "loss": 0.8337,
+      "step": 11731
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 5.4279377061657844e-05,
+      "loss": 0.8124,
+      "step": 11732
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 5.426385720990994e-05,
+      "loss": 0.8543,
+      "step": 11733
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 5.424833875105506e-05,
+      "loss": 0.8131,
+      "step": 11734
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 5.4232821685565826e-05,
+      "loss": 0.8739,
+      "step": 11735
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 5.421730601391477e-05,
+      "loss": 0.9928,
+      "step": 11736
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 5.420179173657449e-05,
+      "loss": 0.8813,
+      "step": 11737
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 5.4186278854017445e-05,
+      "loss": 0.8018,
+      "step": 11738
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 5.417076736671603e-05,
+      "loss": 0.9716,
+      "step": 11739
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 5.415525727514279e-05,
+      "loss": 0.7456,
+      "step": 11740
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 5.4139748579769886e-05,
+      "loss": 0.9121,
+      "step": 11741
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 5.412424128106979e-05,
+      "loss": 0.8012,
+      "step": 11742
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 5.4108735379514706e-05,
+      "loss": 0.8393,
+      "step": 11743
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 5.409323087557685e-05,
+      "loss": 0.8852,
+      "step": 11744
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 5.407772776972848e-05,
+      "loss": 0.7751,
+      "step": 11745
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 5.4062226062441714e-05,
+      "loss": 0.8369,
+      "step": 11746
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 5.404672575418863e-05,
+      "loss": 0.8232,
+      "step": 11747
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 5.403122684544129e-05,
+      "loss": 0.8843,
+      "step": 11748
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 5.401572933667176e-05,
+      "loss": 0.8679,
+      "step": 11749
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 5.400023322835198e-05,
+      "loss": 0.8633,
+      "step": 11750
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 5.3984738520953895e-05,
+      "loss": 0.8643,
+      "step": 11751
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 5.396924521494939e-05,
+      "loss": 0.8757,
+      "step": 11752
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 5.395375331081027e-05,
+      "loss": 0.816,
+      "step": 11753
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 5.393826280900843e-05,
+      "loss": 0.7952,
+      "step": 11754
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 5.3922773710015594e-05,
+      "loss": 0.8104,
+      "step": 11755
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 5.390728601430346e-05,
+      "loss": 0.8378,
+      "step": 11756
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 5.389179972234373e-05,
+      "loss": 0.8264,
+      "step": 11757
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 5.387631483460798e-05,
+      "loss": 0.7417,
+      "step": 11758
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 5.38608313515679e-05,
+      "loss": 0.8716,
+      "step": 11759
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 5.384534927369496e-05,
+      "loss": 0.8685,
+      "step": 11760
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 5.382986860146071e-05,
+      "loss": 0.8675,
+      "step": 11761
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 5.381438933533666e-05,
+      "loss": 0.9249,
+      "step": 11762
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 5.379891147579408e-05,
+      "loss": 0.8783,
+      "step": 11763
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 5.3783435023304485e-05,
+      "loss": 0.8406,
+      "step": 11764
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 5.376795997833912e-05,
+      "loss": 0.8639,
+      "step": 11765
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 5.375248634136937e-05,
+      "loss": 0.8268,
+      "step": 11766
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 5.3737014112866424e-05,
+      "loss": 0.8585,
+      "step": 11767
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 5.3721543293301504e-05,
+      "loss": 0.8365,
+      "step": 11768
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 5.370607388314577e-05,
+      "loss": 0.8788,
+      "step": 11769
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 5.369060588287029e-05,
+      "loss": 0.8683,
+      "step": 11770
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 5.367513929294624e-05,
+      "loss": 0.792,
+      "step": 11771
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 5.36596741138446e-05,
+      "loss": 0.9802,
+      "step": 11772
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 5.364421034603637e-05,
+      "loss": 0.9812,
+      "step": 11773
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 5.362874798999251e-05,
+      "loss": 0.7742,
+      "step": 11774
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 5.361328704618387e-05,
+      "loss": 1.0092,
+      "step": 11775
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 5.359782751508139e-05,
+      "loss": 0.8813,
+      "step": 11776
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 5.358236939715585e-05,
+      "loss": 0.8778,
+      "step": 11777
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 5.3566912692878036e-05,
+      "loss": 0.7758,
+      "step": 11778
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 5.355145740271868e-05,
+      "loss": 0.803,
+      "step": 11779
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 5.3536003527148426e-05,
+      "loss": 0.8806,
+      "step": 11780
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 5.3520551066638016e-05,
+      "loss": 0.8396,
+      "step": 11781
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 5.350510002165795e-05,
+      "loss": 0.7321,
+      "step": 11782
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 5.3489650392678925e-05,
+      "loss": 0.904,
+      "step": 11783
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 5.347420218017134e-05,
+      "loss": 0.8051,
+      "step": 11784
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 5.3458755384605665e-05,
+      "loss": 0.7532,
+      "step": 11785
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 5.344331000645242e-05,
+      "loss": 0.7967,
+      "step": 11786
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 5.3427866046181894e-05,
+      "loss": 0.879,
+      "step": 11787
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 5.341242350426453e-05,
+      "loss": 0.9234,
+      "step": 11788
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 5.3396982381170635e-05,
+      "loss": 0.8384,
+      "step": 11789
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 5.338154267737032e-05,
+      "loss": 0.9539,
+      "step": 11790
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 5.336610439333396e-05,
+      "loss": 0.839,
+      "step": 11791
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 5.3350667529531606e-05,
+      "loss": 0.8886,
+      "step": 11792
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 5.3335232086433496e-05,
+      "loss": 0.8417,
+      "step": 11793
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 5.3319798064509675e-05,
+      "loss": 0.8084,
+      "step": 11794
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 5.3304365464230164e-05,
+      "loss": 0.9386,
+      "step": 11795
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 5.3288934286065e-05,
+      "loss": 0.8595,
+      "step": 11796
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 5.327350453048405e-05,
+      "loss": 0.7486,
+      "step": 11797
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 5.325807619795735e-05,
+      "loss": 0.8536,
+      "step": 11798
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 5.324264928895472e-05,
+      "loss": 0.9161,
+      "step": 11799
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 5.322722380394596e-05,
+      "loss": 1.0159,
+      "step": 11800
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 5.321179974340088e-05,
+      "loss": 0.7747,
+      "step": 11801
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 5.319637710778917e-05,
+      "loss": 0.9875,
+      "step": 11802
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 5.318095589758061e-05,
+      "loss": 0.8434,
+      "step": 11803
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 5.3165536113244764e-05,
+      "loss": 0.8457,
+      "step": 11804
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 5.315011775525136e-05,
+      "loss": 0.8134,
+      "step": 11805
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 5.313470082406983e-05,
+      "loss": 0.8556,
+      "step": 11806
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 5.3119285320169784e-05,
+      "loss": 0.8304,
+      "step": 11807
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 5.310387124402068e-05,
+      "loss": 0.7783,
+      "step": 11808
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 5.3088458596091885e-05,
+      "loss": 0.8469,
+      "step": 11809
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 5.3073047376852914e-05,
+      "loss": 0.9052,
+      "step": 11810
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 5.305763758677305e-05,
+      "loss": 0.872,
+      "step": 11811
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 5.304222922632159e-05,
+      "loss": 0.8168,
+      "step": 11812
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 5.3026822295967825e-05,
+      "loss": 0.864,
+      "step": 11813
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 5.301141679618089e-05,
+      "loss": 0.8982,
+      "step": 11814
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 5.299601272743009e-05,
+      "loss": 0.8463,
+      "step": 11815
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 5.2980610090184465e-05,
+      "loss": 0.8431,
+      "step": 11816
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 5.296520888491314e-05,
+      "loss": 0.7879,
+      "step": 11817
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 5.2949809112085136e-05,
+      "loss": 0.7931,
+      "step": 11818
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 5.293441077216943e-05,
+      "loss": 0.9362,
+      "step": 11819
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 5.2919013865635046e-05,
+      "loss": 0.9299,
+      "step": 11820
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 5.290361839295086e-05,
+      "loss": 0.8217,
+      "step": 11821
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 5.288822435458574e-05,
+      "loss": 0.8292,
+      "step": 11822
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 5.287283175100846e-05,
+      "loss": 0.8065,
+      "step": 11823
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 5.285744058268789e-05,
+      "loss": 0.818,
+      "step": 11824
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 5.284205085009275e-05,
+      "loss": 0.9071,
+      "step": 11825
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 5.28266625536917e-05,
+      "loss": 0.8219,
+      "step": 11826
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 5.281127569395341e-05,
+      "loss": 0.9103,
+      "step": 11827
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 5.2795890271346435e-05,
+      "loss": 0.8703,
+      "step": 11828
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 5.278050628633942e-05,
+      "loss": 0.8196,
+      "step": 11829
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 5.2765123739400854e-05,
+      "loss": 0.792,
+      "step": 11830
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 5.2749742630999155e-05,
+      "loss": 0.7987,
+      "step": 11831
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 5.273436296160289e-05,
+      "loss": 0.907,
+      "step": 11832
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 5.2718984731680285e-05,
+      "loss": 1.0158,
+      "step": 11833
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 5.2703607941699796e-05,
+      "loss": 0.9277,
+      "step": 11834
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 5.268823259212968e-05,
+      "loss": 0.7634,
+      "step": 11835
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 5.267285868343816e-05,
+      "loss": 0.9562,
+      "step": 11836
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 5.2657486216093545e-05,
+      "loss": 0.7648,
+      "step": 11837
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 5.2642115190563924e-05,
+      "loss": 0.8595,
+      "step": 11838
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 5.262674560731744e-05,
+      "loss": 0.8383,
+      "step": 11839
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 5.2611377466822144e-05,
+      "loss": 0.9448,
+      "step": 11840
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 5.2596010769546146e-05,
+      "loss": 0.8755,
+      "step": 11841
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 5.2580645515957386e-05,
+      "loss": 0.7976,
+      "step": 11842
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 5.256528170652384e-05,
+      "loss": 0.7558,
+      "step": 11843
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 5.254991934171338e-05,
+      "loss": 0.9246,
+      "step": 11844
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 5.253455842199384e-05,
+      "loss": 0.7982,
+      "step": 11845
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 5.2519198947833125e-05,
+      "loss": 0.953,
+      "step": 11846
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 5.2503840919698955e-05,
+      "loss": 0.7976,
+      "step": 11847
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 5.248848433805906e-05,
+      "loss": 0.7117,
+      "step": 11848
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 5.247312920338113e-05,
+      "loss": 0.8945,
+      "step": 11849
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 5.2457775516132766e-05,
+      "loss": 0.8397,
+      "step": 11850
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 5.244242327678164e-05,
+      "loss": 0.8152,
+      "step": 11851
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 5.242707248579522e-05,
+      "loss": 0.8304,
+      "step": 11852
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 5.241172314364111e-05,
+      "loss": 0.9357,
+      "step": 11853
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 5.2396375250786754e-05,
+      "loss": 0.9322,
+      "step": 11854
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 5.238102880769946e-05,
+      "loss": 0.8256,
+      "step": 11855
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 5.2365683814846724e-05,
+      "loss": 0.8633,
+      "step": 11856
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 5.235034027269581e-05,
+      "loss": 0.9228,
+      "step": 11857
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 5.2334998181714053e-05,
+      "loss": 0.847,
+      "step": 11858
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 5.23196575423687e-05,
+      "loss": 0.9827,
+      "step": 11859
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 5.23043183551269e-05,
+      "loss": 0.9147,
+      "step": 11860
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 5.228898062045584e-05,
+      "loss": 0.868,
+      "step": 11861
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 5.227364433882258e-05,
+      "loss": 0.6637,
+      "step": 11862
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 5.225830951069428e-05,
+      "loss": 0.8977,
+      "step": 11863
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 5.2242976136537905e-05,
+      "loss": 0.8447,
+      "step": 11864
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 5.2227644216820426e-05,
+      "loss": 0.904,
+      "step": 11865
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 5.221231375200881e-05,
+      "loss": 0.8495,
+      "step": 11866
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 5.219698474256986e-05,
+      "loss": 0.8608,
+      "step": 11867
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 5.218165718897055e-05,
+      "loss": 0.8346,
+      "step": 11868
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 5.216633109167757e-05,
+      "loss": 0.9701,
+      "step": 11869
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 5.2151006451157794e-05,
+      "loss": 0.8499,
+      "step": 11870
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 5.213568326787781e-05,
+      "loss": 0.8533,
+      "step": 11871
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 5.2120361542304316e-05,
+      "loss": 0.8427,
+      "step": 11872
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 5.210504127490399e-05,
+      "loss": 0.798,
+      "step": 11873
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 5.208972246614333e-05,
+      "loss": 0.9796,
+      "step": 11874
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 5.2074405116489e-05,
+      "loss": 0.8545,
+      "step": 11875
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 5.205908922640737e-05,
+      "loss": 0.8437,
+      "step": 11876
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 5.2043774796364866e-05,
+      "loss": 0.889,
+      "step": 11877
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 5.202846182682798e-05,
+      "loss": 0.9597,
+      "step": 11878
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 5.2013150318262996e-05,
+      "loss": 0.9044,
+      "step": 11879
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 5.19978402711363e-05,
+      "loss": 0.801,
+      "step": 11880
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 5.198253168591416e-05,
+      "loss": 0.9258,
+      "step": 11881
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 5.196722456306266e-05,
+      "loss": 0.8443,
+      "step": 11882
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 5.195191890304813e-05,
+      "loss": 0.7647,
+      "step": 11883
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 5.193661470633662e-05,
+      "loss": 0.7835,
+      "step": 11884
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 5.192131197339427e-05,
+      "loss": 0.8755,
+      "step": 11885
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 5.1906010704687124e-05,
+      "loss": 0.877,
+      "step": 11886
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 5.1890710900681136e-05,
+      "loss": 0.8787,
+      "step": 11887
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 5.1875412561842295e-05,
+      "loss": 0.8534,
+      "step": 11888
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 5.186011568863647e-05,
+      "loss": 0.9044,
+      "step": 11889
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 5.184482028152957e-05,
+      "loss": 0.6906,
+      "step": 11890
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 5.182952634098743e-05,
+      "loss": 0.8156,
+      "step": 11891
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 5.181423386747579e-05,
+      "loss": 0.8573,
+      "step": 11892
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 5.1798942861460396e-05,
+      "loss": 0.7702,
+      "step": 11893
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 5.17836533234069e-05,
+      "loss": 0.7222,
+      "step": 11894
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 5.1768365253781007e-05,
+      "loss": 0.7257,
+      "step": 11895
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 5.175307865304825e-05,
+      "loss": 0.7979,
+      "step": 11896
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 5.1737793521674296e-05,
+      "loss": 0.8156,
+      "step": 11897
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 5.172250986012448e-05,
+      "loss": 0.8656,
+      "step": 11898
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 5.170722766886441e-05,
+      "loss": 1.0053,
+      "step": 11899
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 5.169194694835946e-05,
+      "loss": 0.8333,
+      "step": 11900
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 5.167666769907494e-05,
+      "loss": 0.8287,
+      "step": 11901
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 5.166138992147629e-05,
+      "loss": 0.8171,
+      "step": 11902
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 5.1646113616028735e-05,
+      "loss": 1.0129,
+      "step": 11903
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 5.163083878319753e-05,
+      "loss": 0.8886,
+      "step": 11904
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 5.1615565423447856e-05,
+      "loss": 0.9226,
+      "step": 11905
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 5.1600293537244825e-05,
+      "loss": 0.8479,
+      "step": 11906
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 5.1585023125053625e-05,
+      "loss": 0.8481,
+      "step": 11907
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 5.156975418733927e-05,
+      "loss": 0.9551,
+      "step": 11908
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 5.15544867245668e-05,
+      "loss": 0.89,
+      "step": 11909
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 5.1539220737201144e-05,
+      "loss": 0.8849,
+      "step": 11910
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 5.1523956225707206e-05,
+      "loss": 0.814,
+      "step": 11911
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 5.150869319054996e-05,
+      "loss": 0.8878,
+      "step": 11912
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 5.149343163219419e-05,
+      "loss": 0.7797,
+      "step": 11913
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 5.147817155110468e-05,
+      "loss": 0.9163,
+      "step": 11914
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 5.1462912947746145e-05,
+      "loss": 0.8464,
+      "step": 11915
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 5.144765582258337e-05,
+      "loss": 0.9095,
+      "step": 11916
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 5.143240017608095e-05,
+      "loss": 0.9612,
+      "step": 11917
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 5.1417146008703486e-05,
+      "loss": 0.8943,
+      "step": 11918
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 5.140189332091563e-05,
+      "loss": 0.8126,
+      "step": 11919
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 5.138664211318176e-05,
+      "loss": 0.9241,
+      "step": 11920
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 5.1371392385966465e-05,
+      "loss": 0.8348,
+      "step": 11921
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 5.135614413973415e-05,
+      "loss": 0.9575,
+      "step": 11922
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 5.1340897374949136e-05,
+      "loss": 0.8745,
+      "step": 11923
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 5.132565209207586e-05,
+      "loss": 0.9146,
+      "step": 11924
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 5.1310408291578574e-05,
+      "loss": 0.9224,
+      "step": 11925
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 5.129516597392152e-05,
+      "loss": 0.9564,
+      "step": 11926
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 5.127992513956892e-05,
+      "loss": 0.8592,
+      "step": 11927
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 5.126468578898487e-05,
+      "loss": 1.0708,
+      "step": 11928
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 5.124944792263358e-05,
+      "loss": 0.7201,
+      "step": 11929
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 5.1234211540979074e-05,
+      "loss": 0.8668,
+      "step": 11930
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 5.121897664448536e-05,
+      "loss": 0.7793,
+      "step": 11931
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 5.1203743233616407e-05,
+      "loss": 0.7924,
+      "step": 11932
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 5.118851130883622e-05,
+      "loss": 1.0349,
+      "step": 11933
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 5.117328087060863e-05,
+      "loss": 0.8349,
+      "step": 11934
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 5.1158051919397486e-05,
+      "loss": 0.9524,
+      "step": 11935
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 5.1142824455666605e-05,
+      "loss": 0.8763,
+      "step": 11936
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 5.112759847987967e-05,
+      "loss": 0.8176,
+      "step": 11937
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 5.111237399250048e-05,
+      "loss": 0.8065,
+      "step": 11938
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 5.109715099399266e-05,
+      "loss": 0.8613,
+      "step": 11939
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 5.108192948481982e-05,
+      "loss": 0.792,
+      "step": 11940
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 5.106670946544554e-05,
+      "loss": 0.8468,
+      "step": 11941
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 5.105149093633329e-05,
+      "loss": 0.8003,
+      "step": 11942
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 5.103627389794665e-05,
+      "loss": 0.8765,
+      "step": 11943
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 5.1021058350748954e-05,
+      "loss": 0.9473,
+      "step": 11944
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 5.100584429520367e-05,
+      "loss": 0.8769,
+      "step": 11945
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 5.0990631731774176e-05,
+      "loss": 0.7531,
+      "step": 11946
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 5.097542066092362e-05,
+      "loss": 0.8255,
+      "step": 11947
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 5.096021108311537e-05,
+      "loss": 0.635,
+      "step": 11948
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 5.094500299881258e-05,
+      "loss": 0.7727,
+      "step": 11949
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 5.092979640847846e-05,
+      "loss": 0.8306,
+      "step": 11950
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 5.091459131257611e-05,
+      "loss": 0.7599,
+      "step": 11951
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 5.0899387711568595e-05,
+      "loss": 0.822,
+      "step": 11952
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 5.088418560591894e-05,
+      "loss": 0.9276,
+      "step": 11953
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 5.086898499609007e-05,
+      "loss": 0.7801,
+      "step": 11954
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 5.085378588254504e-05,
+      "loss": 0.9099,
+      "step": 11955
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 5.083858826574663e-05,
+      "loss": 0.8838,
+      "step": 11956
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 5.082339214615777e-05,
+      "loss": 0.949,
+      "step": 11957
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 5.080819752424118e-05,
+      "loss": 0.8622,
+      "step": 11958
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 5.079300440045961e-05,
+      "loss": 0.9363,
+      "step": 11959
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 5.0777812775275846e-05,
+      "loss": 0.8739,
+      "step": 11960
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 5.076262264915246e-05,
+      "loss": 0.7876,
+      "step": 11961
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 5.074743402255218e-05,
+      "loss": 0.9729,
+      "step": 11962
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 5.073224689593747e-05,
+      "loss": 0.9566,
+      "step": 11963
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 5.071706126977085e-05,
+      "loss": 0.774,
+      "step": 11964
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 5.0701877144514866e-05,
+      "loss": 0.8447,
+      "step": 11965
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 5.068669452063189e-05,
+      "loss": 1.0405,
+      "step": 11966
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 5.067151339858438e-05,
+      "loss": 0.8094,
+      "step": 11967
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 5.0656333778834674e-05,
+      "loss": 0.922,
+      "step": 11968
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 5.0641155661844944e-05,
+      "loss": 0.8724,
+      "step": 11969
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 5.062597904807755e-05,
+      "loss": 0.812,
+      "step": 11970
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 5.0610803937994645e-05,
+      "loss": 0.7783,
+      "step": 11971
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 5.059563033205845e-05,
+      "loss": 0.7682,
+      "step": 11972
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 5.058045823073103e-05,
+      "loss": 0.8264,
+      "step": 11973
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 5.0565287634474454e-05,
+      "loss": 0.8305,
+      "step": 11974
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 5.055011854375076e-05,
+      "loss": 0.9549,
+      "step": 11975
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 5.0534950959021856e-05,
+      "loss": 0.8062,
+      "step": 11976
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 5.0519784880749764e-05,
+      "loss": 0.8741,
+      "step": 11977
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 5.0504620309396333e-05,
+      "loss": 0.8823,
+      "step": 11978
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 5.048945724542338e-05,
+      "loss": 0.8766,
+      "step": 11979
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 5.047429568929272e-05,
+      "loss": 0.8952,
+      "step": 11980
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 5.045913564146603e-05,
+      "loss": 0.8859,
+      "step": 11981
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 5.0443977102405115e-05,
+      "loss": 0.8815,
+      "step": 11982
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 5.042882007257157e-05,
+      "loss": 0.8245,
+      "step": 11983
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 5.041366455242701e-05,
+      "loss": 0.9637,
+      "step": 11984
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 5.0398510542432984e-05,
+      "loss": 0.8925,
+      "step": 11985
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 5.0383358043051e-05,
+      "loss": 0.7977,
+      "step": 11986
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 5.036820705474257e-05,
+      "loss": 0.8772,
+      "step": 11987
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 5.0353057577969055e-05,
+      "loss": 0.8813,
+      "step": 11988
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 5.033790961319195e-05,
+      "loss": 0.8846,
+      "step": 11989
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 5.032276316087242e-05,
+      "loss": 0.9315,
+      "step": 11990
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 5.030761822147188e-05,
+      "loss": 0.8787,
+      "step": 11991
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 5.0292474795451516e-05,
+      "loss": 0.8944,
+      "step": 11992
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 5.02773328832725e-05,
+      "loss": 0.8475,
+      "step": 11993
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 5.026219248539603e-05,
+      "loss": 1.1312,
+      "step": 11994
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 5.024705360228319e-05,
+      "loss": 0.8687,
+      "step": 11995
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 5.023191623439503e-05,
+      "loss": 0.9114,
+      "step": 11996
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 5.021678038219255e-05,
+      "loss": 0.8018,
+      "step": 11997
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 5.0201646046136675e-05,
+      "loss": 0.8529,
+      "step": 11998
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 5.0186513226688406e-05,
+      "loss": 0.9655,
+      "step": 11999
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 5.017138192430857e-05,
+      "loss": 0.7413,
+      "step": 12000
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 5.015625213945798e-05,
+      "loss": 0.9784,
+      "step": 12001
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 5.014112387259744e-05,
+      "loss": 0.7196,
+      "step": 12002
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 5.0125997124187606e-05,
+      "loss": 0.8077,
+      "step": 12003
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 5.011087189468927e-05,
+      "loss": 0.8697,
+      "step": 12004
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 5.009574818456302e-05,
+      "loss": 0.8096,
+      "step": 12005
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 5.008062599426946e-05,
+      "loss": 0.7878,
+      "step": 12006
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 5.006550532426907e-05,
+      "loss": 0.8767,
+      "step": 12007
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 5.0050386175022444e-05,
+      "loss": 0.7428,
+      "step": 12008
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 5.003526854699001e-05,
+      "loss": 0.829,
+      "step": 12009
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 5.0020152440632104e-05,
+      "loss": 0.9544,
+      "step": 12010
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 5.000503785640923e-05,
+      "loss": 0.8326,
+      "step": 12011
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 4.998992479478154e-05,
+      "loss": 0.7887,
+      "step": 12012
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 4.997481325620942e-05,
+      "loss": 0.9133,
+      "step": 12013
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 4.995970324115304e-05,
+      "loss": 0.8145,
+      "step": 12014
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 4.994459475007255e-05,
+      "loss": 0.8644,
+      "step": 12015
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 4.9929487783428154e-05,
+      "loss": 0.8234,
+      "step": 12016
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 4.99143823416799e-05,
+      "loss": 0.8215,
+      "step": 12017
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 4.98992784252878e-05,
+      "loss": 0.9621,
+      "step": 12018
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 4.988417603471184e-05,
+      "loss": 0.9199,
+      "step": 12019
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 4.9869075170412e-05,
+      "loss": 0.9219,
+      "step": 12020
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 4.985397583284817e-05,
+      "loss": 0.7491,
+      "step": 12021
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 4.983887802248021e-05,
+      "loss": 0.7633,
+      "step": 12022
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 4.982378173976788e-05,
+      "loss": 0.898,
+      "step": 12023
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 4.980868698517093e-05,
+      "loss": 0.9151,
+      "step": 12024
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 4.9793593759149137e-05,
+      "loss": 0.8122,
+      "step": 12025
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 4.9778502062162125e-05,
+      "loss": 0.865,
+      "step": 12026
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 4.976341189466952e-05,
+      "loss": 0.8712,
+      "step": 12027
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 4.974832325713088e-05,
+      "loss": 0.9721,
+      "step": 12028
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 4.9733236150005705e-05,
+      "loss": 0.8195,
+      "step": 12029
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 4.971815057375354e-05,
+      "loss": 1.1148,
+      "step": 12030
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 4.9703066528833774e-05,
+      "loss": 0.8413,
+      "step": 12031
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 4.96879840157058e-05,
+      "loss": 0.9028,
+      "step": 12032
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 4.9672903034828956e-05,
+      "loss": 0.9212,
+      "step": 12033
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 4.965782358666249e-05,
+      "loss": 0.8659,
+      "step": 12034
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 4.964274567166573e-05,
+      "loss": 0.797,
+      "step": 12035
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 4.962766929029778e-05,
+      "loss": 0.9195,
+      "step": 12036
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 4.961259444301788e-05,
+      "loss": 0.9469,
+      "step": 12037
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 4.959752113028513e-05,
+      "loss": 0.7916,
+      "step": 12038
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 4.958244935255848e-05,
+      "loss": 0.8709,
+      "step": 12039
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 4.956737911029705e-05,
+      "loss": 0.7836,
+      "step": 12040
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 4.955231040395972e-05,
+      "loss": 0.9271,
+      "step": 12041
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 4.9537243234005505e-05,
+      "loss": 0.876,
+      "step": 12042
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 4.9522177600893205e-05,
+      "loss": 0.7676,
+      "step": 12043
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 4.950711350508167e-05,
+      "loss": 0.8919,
+      "step": 12044
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 4.9492050947029676e-05,
+      "loss": 0.8712,
+      "step": 12045
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 4.947698992719589e-05,
+      "loss": 0.891,
+      "step": 12046
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 4.9461930446039104e-05,
+      "loss": 0.8246,
+      "step": 12047
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 4.9446872504017895e-05,
+      "loss": 0.8529,
+      "step": 12048
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 4.943181610159084e-05,
+      "loss": 0.9455,
+      "step": 12049
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 4.941676123921652e-05,
+      "loss": 0.8611,
+      "step": 12050
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 4.940170791735337e-05,
+      "loss": 0.8151,
+      "step": 12051
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 4.938665613645991e-05,
+      "loss": 0.7774,
+      "step": 12052
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 4.937160589699447e-05,
+      "loss": 0.8268,
+      "step": 12053
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 4.9356557199415507e-05,
+      "loss": 0.7497,
+      "step": 12054
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 4.934151004418125e-05,
+      "loss": 0.9108,
+      "step": 12055
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 4.9326464431749916e-05,
+      "loss": 0.8782,
+      "step": 12056
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 4.931142036257983e-05,
+      "loss": 0.9183,
+      "step": 12057
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 4.9296377837129074e-05,
+      "loss": 0.8275,
+      "step": 12058
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 4.928133685585582e-05,
+      "loss": 0.957,
+      "step": 12059
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 4.9266297419218166e-05,
+      "loss": 0.7621,
+      "step": 12060
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 4.925125952767402e-05,
+      "loss": 0.9165,
+      "step": 12061
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 4.923622318168148e-05,
+      "loss": 0.8832,
+      "step": 12062
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 4.922118838169837e-05,
+      "loss": 0.8621,
+      "step": 12063
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 4.920615512818268e-05,
+      "loss": 0.783,
+      "step": 12064
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 4.919112342159221e-05,
+      "loss": 0.9502,
+      "step": 12065
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 4.917609326238475e-05,
+      "loss": 0.781,
+      "step": 12066
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 4.916106465101802e-05,
+      "loss": 0.7914,
+      "step": 12067
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 4.91460375879497e-05,
+      "loss": 0.8405,
+      "step": 12068
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 4.913101207363751e-05,
+      "loss": 0.8395,
+      "step": 12069
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 4.9115988108539014e-05,
+      "loss": 0.9133,
+      "step": 12070
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 4.9100965693111754e-05,
+      "loss": 0.8266,
+      "step": 12071
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 4.908594482781326e-05,
+      "loss": 0.7802,
+      "step": 12072
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 4.907092551310093e-05,
+      "loss": 0.7787,
+      "step": 12073
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 4.905590774943227e-05,
+      "loss": 0.8078,
+      "step": 12074
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 4.9040891537264564e-05,
+      "loss": 0.8638,
+      "step": 12075
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 4.902587687705525e-05,
+      "loss": 0.9311,
+      "step": 12076
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 4.9010863769261486e-05,
+      "loss": 0.8225,
+      "step": 12077
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 4.899585221434048e-05,
+      "loss": 0.9038,
+      "step": 12078
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 4.898084221274949e-05,
+      "loss": 1.0026,
+      "step": 12079
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 4.8965833764945577e-05,
+      "loss": 0.7982,
+      "step": 12080
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 4.895082687138589e-05,
+      "loss": 0.9061,
+      "step": 12081
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 4.8935821532527436e-05,
+      "loss": 0.9183,
+      "step": 12082
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 4.892081774882721e-05,
+      "loss": 0.9763,
+      "step": 12083
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 4.890581552074212e-05,
+      "loss": 1.0313,
+      "step": 12084
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 4.889081484872905e-05,
+      "loss": 0.9176,
+      "step": 12085
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 4.88758157332449e-05,
+      "loss": 0.837,
+      "step": 12086
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 4.886081817474645e-05,
+      "loss": 0.9504,
+      "step": 12087
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 4.8845822173690434e-05,
+      "loss": 0.9158,
+      "step": 12088
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 4.8830827730533554e-05,
+      "loss": 0.8123,
+      "step": 12089
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 4.8815834845732454e-05,
+      "loss": 1.0182,
+      "step": 12090
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 4.880084351974379e-05,
+      "loss": 0.8025,
+      "step": 12091
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 4.878585375302409e-05,
+      "loss": 0.859,
+      "step": 12092
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 4.8770865546029864e-05,
+      "loss": 0.9161,
+      "step": 12093
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 4.875587889921759e-05,
+      "loss": 0.8886,
+      "step": 12094
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 4.874089381304364e-05,
+      "loss": 0.7068,
+      "step": 12095
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 4.872591028796447e-05,
+      "loss": 0.8612,
+      "step": 12096
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 4.8710928324436344e-05,
+      "loss": 0.8386,
+      "step": 12097
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 4.869594792291555e-05,
+      "loss": 0.8442,
+      "step": 12098
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 4.868096908385829e-05,
+      "loss": 0.8234,
+      "step": 12099
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 4.86659918077208e-05,
+      "loss": 0.9824,
+      "step": 12100
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 4.865101609495918e-05,
+      "loss": 0.7886,
+      "step": 12101
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 4.8636041946029474e-05,
+      "loss": 0.9022,
+      "step": 12102
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 4.862106936138785e-05,
+      "loss": 0.8214,
+      "step": 12103
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 4.860609834149014e-05,
+      "loss": 0.8538,
+      "step": 12104
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 4.859112888679238e-05,
+      "loss": 0.7784,
+      "step": 12105
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 4.857616099775044e-05,
+      "loss": 0.8061,
+      "step": 12106
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 4.856119467482013e-05,
+      "loss": 0.7683,
+      "step": 12107
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 4.854622991845733e-05,
+      "loss": 0.8309,
+      "step": 12108
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 4.853126672911774e-05,
+      "loss": 0.9422,
+      "step": 12109
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 4.8516305107257074e-05,
+      "loss": 0.9438,
+      "step": 12110
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 4.850134505333095e-05,
+      "loss": 0.7618,
+      "step": 12111
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 4.848638656779505e-05,
+      "loss": 0.833,
+      "step": 12112
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 4.847142965110489e-05,
+      "loss": 0.8922,
+      "step": 12113
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 4.845647430371599e-05,
+      "loss": 0.7876,
+      "step": 12114
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 4.844152052608382e-05,
+      "loss": 0.919,
+      "step": 12115
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 4.8426568318663736e-05,
+      "loss": 0.7766,
+      "step": 12116
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 4.84116176819112e-05,
+      "loss": 0.8705,
+      "step": 12117
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 4.8396668616281504e-05,
+      "loss": 0.8728,
+      "step": 12118
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 4.8381721122229906e-05,
+      "loss": 0.9126,
+      "step": 12119
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 4.836677520021164e-05,
+      "loss": 0.8672,
+      "step": 12120
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 4.835183085068183e-05,
+      "loss": 0.8591,
+      "step": 12121
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 4.83368880740957e-05,
+      "loss": 0.7462,
+      "step": 12122
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 4.8321946870908294e-05,
+      "loss": 0.8904,
+      "step": 12123
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 4.83070072415746e-05,
+      "loss": 0.8391,
+      "step": 12124
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 4.8292069186549716e-05,
+      "loss": 0.8963,
+      "step": 12125
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 4.827713270628843e-05,
+      "loss": 0.9387,
+      "step": 12126
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 4.8262197801245746e-05,
+      "loss": 0.8908,
+      "step": 12127
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 4.824726447187644e-05,
+      "loss": 0.6817,
+      "step": 12128
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 4.823233271863537e-05,
+      "loss": 0.7188,
+      "step": 12129
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 4.821740254197725e-05,
+      "loss": 0.9349,
+      "step": 12130
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 4.820247394235677e-05,
+      "loss": 0.736,
+      "step": 12131
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 4.81875469202286e-05,
+      "loss": 0.8013,
+      "step": 12132
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 4.8172621476047286e-05,
+      "loss": 0.7503,
+      "step": 12133
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 4.815769761026745e-05,
+      "loss": 0.9464,
+      "step": 12134
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 4.814277532334358e-05,
+      "loss": 0.7756,
+      "step": 12135
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 4.812785461573014e-05,
+      "loss": 0.8269,
+      "step": 12136
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 4.8112935487881505e-05,
+      "loss": 0.8727,
+      "step": 12137
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 4.809801794025204e-05,
+      "loss": 0.8422,
+      "step": 12138
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 4.808310197329609e-05,
+      "loss": 0.9299,
+      "step": 12139
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 4.8068187587467926e-05,
+      "loss": 0.9378,
+      "step": 12140
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 4.8053274783221746e-05,
+      "loss": 0.8511,
+      "step": 12141
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 4.803836356101171e-05,
+      "loss": 0.7777,
+      "step": 12142
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 4.8023453921291916e-05,
+      "loss": 0.9053,
+      "step": 12143
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 4.8008545864516496e-05,
+      "loss": 0.9034,
+      "step": 12144
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 4.799363939113942e-05,
+      "loss": 0.9119,
+      "step": 12145
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 4.797873450161476e-05,
+      "loss": 0.8263,
+      "step": 12146
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 4.796383119639635e-05,
+      "loss": 0.821,
+      "step": 12147
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 4.794892947593805e-05,
+      "loss": 0.853,
+      "step": 12148
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 4.793402934069377e-05,
+      "loss": 0.8954,
+      "step": 12149
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 4.791913079111723e-05,
+      "loss": 0.8578,
+      "step": 12150
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 4.7904233827662223e-05,
+      "loss": 0.9089,
+      "step": 12151
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 4.788933845078245e-05,
+      "loss": 0.8627,
+      "step": 12152
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 4.787444466093144e-05,
+      "loss": 0.9967,
+      "step": 12153
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 4.7859552458562876e-05,
+      "loss": 0.8917,
+      "step": 12154
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 4.784466184413023e-05,
+      "loss": 0.8618,
+      "step": 12155
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 4.782977281808709e-05,
+      "loss": 0.7547,
+      "step": 12156
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 4.781488538088685e-05,
+      "loss": 0.9459,
+      "step": 12157
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 4.77999995329829e-05,
+      "loss": 0.9331,
+      "step": 12158
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 4.778511527482859e-05,
+      "loss": 0.9664,
+      "step": 12159
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 4.77702326068772e-05,
+      "loss": 0.9197,
+      "step": 12160
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 4.775535152958203e-05,
+      "loss": 0.7466,
+      "step": 12161
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 4.774047204339627e-05,
+      "loss": 0.8661,
+      "step": 12162
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 4.772559414877304e-05,
+      "loss": 0.7967,
+      "step": 12163
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 4.771071784616549e-05,
+      "loss": 0.9382,
+      "step": 12164
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 4.769584313602661e-05,
+      "loss": 0.7957,
+      "step": 12165
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 4.768097001880949e-05,
+      "loss": 0.8458,
+      "step": 12166
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 4.7666098494967025e-05,
+      "loss": 0.943,
+      "step": 12167
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 4.7651228564952225e-05,
+      "loss": 0.9001,
+      "step": 12168
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 4.763636022921785e-05,
+      "loss": 0.9157,
+      "step": 12169
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 4.7621493488216705e-05,
+      "loss": 0.8564,
+      "step": 12170
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 4.760662834240165e-05,
+      "loss": 1.0093,
+      "step": 12171
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 4.759176479222531e-05,
+      "loss": 0.7345,
+      "step": 12172
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 4.757690283814044e-05,
+      "loss": 0.8625,
+      "step": 12173
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 4.7562042480599614e-05,
+      "loss": 0.9648,
+      "step": 12174
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 4.754718372005541e-05,
+      "loss": 0.9444,
+      "step": 12175
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 4.753232655696035e-05,
+      "loss": 0.9049,
+      "step": 12176
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 4.751747099176687e-05,
+      "loss": 0.8186,
+      "step": 12177
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 4.750261702492746e-05,
+      "loss": 0.8598,
+      "step": 12178
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 4.7487764656894486e-05,
+      "loss": 0.9037,
+      "step": 12179
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 4.747291388812025e-05,
+      "loss": 0.8366,
+      "step": 12180
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 4.745806471905705e-05,
+      "loss": 0.9294,
+      "step": 12181
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 4.744321715015706e-05,
+      "loss": 0.9692,
+      "step": 12182
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 4.7428371181872565e-05,
+      "loss": 0.9428,
+      "step": 12183
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 4.741352681465564e-05,
+      "loss": 0.8564,
+      "step": 12184
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 4.7398684048958376e-05,
+      "loss": 0.702,
+      "step": 12185
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 4.738384288523281e-05,
+      "loss": 0.7372,
+      "step": 12186
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 4.7369003323930886e-05,
+      "loss": 1.0171,
+      "step": 12187
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 4.735416536550463e-05,
+      "loss": 0.935,
+      "step": 12188
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 4.7339329010405876e-05,
+      "loss": 0.6985,
+      "step": 12189
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 4.732449425908648e-05,
+      "loss": 0.904,
+      "step": 12190
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 4.730966111199819e-05,
+      "loss": 0.7548,
+      "step": 12191
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 4.729482956959283e-05,
+      "loss": 0.9418,
+      "step": 12192
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 4.727999963232205e-05,
+      "loss": 0.9436,
+      "step": 12193
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 4.7265171300637455e-05,
+      "loss": 0.9574,
+      "step": 12194
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 4.725034457499077e-05,
+      "loss": 0.8149,
+      "step": 12195
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 4.7235519455833366e-05,
+      "loss": 0.8153,
+      "step": 12196
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 4.7220695943616875e-05,
+      "loss": 0.9374,
+      "step": 12197
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 4.720587403879271e-05,
+      "loss": 0.9178,
+      "step": 12198
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 4.719105374181223e-05,
+      "loss": 0.7987,
+      "step": 12199
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 4.717623505312685e-05,
+      "loss": 0.8574,
+      "step": 12200
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 4.716141797318785e-05,
+      "loss": 0.8524,
+      "step": 12201
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 4.714660250244648e-05,
+      "loss": 0.721,
+      "step": 12202
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 4.71317886413539e-05,
+      "loss": 0.8327,
+      "step": 12203
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 4.711697639036136e-05,
+      "loss": 0.9197,
+      "step": 12204
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 4.7102165749919904e-05,
+      "loss": 0.8629,
+      "step": 12205
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 4.70873567204806e-05,
+      "loss": 0.9457,
+      "step": 12206
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 4.707254930249446e-05,
+      "loss": 0.8964,
+      "step": 12207
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 4.705774349641241e-05,
+      "loss": 0.9447,
+      "step": 12208
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 4.704293930268543e-05,
+      "loss": 0.8472,
+      "step": 12209
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 4.702813672176434e-05,
+      "loss": 0.8549,
+      "step": 12210
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 4.701333575409995e-05,
+      "loss": 0.7434,
+      "step": 12211
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 4.699853640014303e-05,
+      "loss": 0.6776,
+      "step": 12212
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 4.698373866034427e-05,
+      "loss": 0.8668,
+      "step": 12213
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 4.696894253515438e-05,
+      "loss": 0.7766,
+      "step": 12214
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 4.695414802502396e-05,
+      "loss": 0.8253,
+      "step": 12215
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 4.693935513040354e-05,
+      "loss": 0.8264,
+      "step": 12216
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 4.692456385174373e-05,
+      "loss": 0.7434,
+      "step": 12217
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 4.690977418949486e-05,
+      "loss": 0.7797,
+      "step": 12218
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 4.689498614410747e-05,
+      "loss": 0.7768,
+      "step": 12219
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 4.688019971603184e-05,
+      "loss": 0.8374,
+      "step": 12220
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 4.686541490571836e-05,
+      "loss": 0.8521,
+      "step": 12221
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 4.685063171361729e-05,
+      "loss": 0.9186,
+      "step": 12222
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 4.683585014017883e-05,
+      "loss": 0.7641,
+      "step": 12223
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 4.682107018585315e-05,
+      "loss": 0.9304,
+      "step": 12224
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 4.6806291851090365e-05,
+      "loss": 0.8852,
+      "step": 12225
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 4.6791515136340593e-05,
+      "loss": 0.8254,
+      "step": 12226
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 4.677674004205382e-05,
+      "loss": 0.8623,
+      "step": 12227
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 4.676196656868005e-05,
+      "loss": 0.8517,
+      "step": 12228
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 4.6747194716669185e-05,
+      "loss": 0.8879,
+      "step": 12229
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 4.673242448647106e-05,
+      "loss": 0.7737,
+      "step": 12230
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 4.6717655878535605e-05,
+      "loss": 0.9056,
+      "step": 12231
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 4.670288889331253e-05,
+      "loss": 0.9177,
+      "step": 12232
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 4.668812353125157e-05,
+      "loss": 0.8347,
+      "step": 12233
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 4.6673359792802416e-05,
+      "loss": 0.9478,
+      "step": 12234
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 4.665859767841465e-05,
+      "loss": 0.8355,
+      "step": 12235
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 4.664383718853793e-05,
+      "loss": 0.8859,
+      "step": 12236
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 4.662907832362171e-05,
+      "loss": 0.8936,
+      "step": 12237
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 4.6614321084115585e-05,
+      "loss": 0.9523,
+      "step": 12238
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 4.6599565470468876e-05,
+      "loss": 0.8282,
+      "step": 12239
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 4.6584811483130956e-05,
+      "loss": 0.8628,
+      "step": 12240
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 4.657005912255123e-05,
+      "loss": 0.9913,
+      "step": 12241
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 4.655530838917893e-05,
+      "loss": 0.8191,
+      "step": 12242
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 4.654055928346334e-05,
+      "loss": 0.8689,
+      "step": 12243
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 4.6525811805853645e-05,
+      "loss": 0.6699,
+      "step": 12244
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 4.651106595679888e-05,
+      "loss": 0.8767,
+      "step": 12245
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 4.6496321736748236e-05,
+      "loss": 0.9185,
+      "step": 12246
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 4.648157914615068e-05,
+      "loss": 0.7301,
+      "step": 12247
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 4.646683818545525e-05,
+      "loss": 0.836,
+      "step": 12248
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 4.645209885511086e-05,
+      "loss": 0.878,
+      "step": 12249
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 4.643736115556639e-05,
+      "loss": 0.9151,
+      "step": 12250
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 4.64226250872707e-05,
+      "loss": 0.8051,
+      "step": 12251
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 4.64078906506725e-05,
+      "loss": 0.9984,
+      "step": 12252
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 4.6393157846220626e-05,
+      "loss": 0.95,
+      "step": 12253
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 4.637842667436373e-05,
+      "loss": 0.8094,
+      "step": 12254
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 4.6363697135550435e-05,
+      "loss": 0.7949,
+      "step": 12255
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 4.634896923022933e-05,
+      "loss": 0.8473,
+      "step": 12256
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 4.6334242958848936e-05,
+      "loss": 0.7986,
+      "step": 12257
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 4.6319518321857794e-05,
+      "loss": 0.8519,
+      "step": 12258
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 4.6304795319704275e-05,
+      "loss": 0.75,
+      "step": 12259
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 4.6290073952836875e-05,
+      "loss": 0.9053,
+      "step": 12260
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 4.6275354221703835e-05,
+      "loss": 0.839,
+      "step": 12261
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 4.626063612675343e-05,
+      "loss": 0.9329,
+      "step": 12262
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 4.624591966843398e-05,
+      "loss": 0.759,
+      "step": 12263
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 4.62312048471936e-05,
+      "loss": 0.9091,
+      "step": 12264
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 4.62164916634805e-05,
+      "loss": 0.9944,
+      "step": 12265
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 4.620178011774273e-05,
+      "loss": 0.7866,
+      "step": 12266
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 4.618707021042834e-05,
+      "loss": 0.9289,
+      "step": 12267
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 4.617236194198531e-05,
+      "loss": 0.9518,
+      "step": 12268
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 4.615765531286156e-05,
+      "loss": 0.9372,
+      "step": 12269
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 4.614295032350503e-05,
+      "loss": 0.8202,
+      "step": 12270
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 4.612824697436353e-05,
+      "loss": 0.746,
+      "step": 12271
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 4.611354526588486e-05,
+      "loss": 0.9381,
+      "step": 12272
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 4.6098845198516763e-05,
+      "loss": 1.003,
+      "step": 12273
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 4.608414677270687e-05,
+      "loss": 0.8125,
+      "step": 12274
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 4.6069449988902926e-05,
+      "loss": 0.7846,
+      "step": 12275
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 4.605475484755245e-05,
+      "loss": 0.8973,
+      "step": 12276
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 4.6040061349103e-05,
+      "loss": 0.853,
+      "step": 12277
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 4.602536949400208e-05,
+      "loss": 0.8505,
+      "step": 12278
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 4.601067928269707e-05,
+      "loss": 0.8077,
+      "step": 12279
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 4.599599071563544e-05,
+      "loss": 0.8154,
+      "step": 12280
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 4.598130379326445e-05,
+      "loss": 0.7434,
+      "step": 12281
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 4.5966618516031523e-05,
+      "loss": 0.9368,
+      "step": 12282
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 4.595193488438374e-05,
+      "loss": 0.8122,
+      "step": 12283
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 4.593725289876839e-05,
+      "loss": 0.8489,
+      "step": 12284
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 4.5922572559632584e-05,
+      "loss": 0.8414,
+      "step": 12285
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 4.590789386742338e-05,
+      "loss": 0.8221,
+      "step": 12286
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 4.589321682258788e-05,
+      "loss": 0.7606,
+      "step": 12287
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 4.587854142557304e-05,
+      "loss": 0.9273,
+      "step": 12288
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 4.58638676768258e-05,
+      "loss": 0.8642,
+      "step": 12289
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 4.5849195576793056e-05,
+      "loss": 0.8506,
+      "step": 12290
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 4.58345251259216e-05,
+      "loss": 0.8691,
+      "step": 12291
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 4.5819856324658294e-05,
+      "loss": 0.7937,
+      "step": 12292
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 4.580518917344985e-05,
+      "loss": 0.8028,
+      "step": 12293
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 4.579052367274293e-05,
+      "loss": 0.9343,
+      "step": 12294
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 4.577585982298417e-05,
+      "loss": 0.7507,
+      "step": 12295
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 4.57611976246202e-05,
+      "loss": 0.8288,
+      "step": 12296
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 4.574653707809754e-05,
+      "loss": 0.8151,
+      "step": 12297
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 4.5731878183862665e-05,
+      "loss": 0.8756,
+      "step": 12298
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 4.571722094236202e-05,
+      "loss": 0.8764,
+      "step": 12299
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 4.5702565354041946e-05,
+      "loss": 0.9123,
+      "step": 12300
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 4.568791141934886e-05,
+      "loss": 0.8064,
+      "step": 12301
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 4.5673259138729e-05,
+      "loss": 0.8145,
+      "step": 12302
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 4.565860851262861e-05,
+      "loss": 0.804,
+      "step": 12303
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 4.564395954149387e-05,
+      "loss": 0.9004,
+      "step": 12304
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 4.562931222577088e-05,
+      "loss": 0.9446,
+      "step": 12305
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 4.56146665659058e-05,
+      "loss": 0.8877,
+      "step": 12306
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 4.5600022562344615e-05,
+      "loss": 0.8009,
+      "step": 12307
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 4.558538021553328e-05,
+      "loss": 0.9049,
+      "step": 12308
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 4.557073952591784e-05,
+      "loss": 0.9781,
+      "step": 12309
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 4.555610049394403e-05,
+      "loss": 0.8427,
+      "step": 12310
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 4.554146312005778e-05,
+      "loss": 0.788,
+      "step": 12311
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 4.55268274047048e-05,
+      "loss": 0.792,
+      "step": 12312
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 4.551219334833091e-05,
+      "loss": 1.021,
+      "step": 12313
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 4.5497560951381735e-05,
+      "loss": 0.9627,
+      "step": 12314
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 4.5482930214302913e-05,
+      "loss": 0.7595,
+      "step": 12315
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 4.546830113754001e-05,
+      "loss": 0.8549,
+      "step": 12316
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 4.545367372153854e-05,
+      "loss": 0.7922,
+      "step": 12317
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 4.5439047966744044e-05,
+      "loss": 0.8871,
+      "step": 12318
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 4.54244238736019e-05,
+      "loss": 0.7998,
+      "step": 12319
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 4.54098014425575e-05,
+      "loss": 0.8254,
+      "step": 12320
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 4.539518067405616e-05,
+      "loss": 0.9117,
+      "step": 12321
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 4.538056156854312e-05,
+      "loss": 0.8923,
+      "step": 12322
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 4.536594412646369e-05,
+      "loss": 0.8755,
+      "step": 12323
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 4.5351328348263e-05,
+      "loss": 0.9515,
+      "step": 12324
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 4.533671423438616e-05,
+      "loss": 0.8709,
+      "step": 12325
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.532210178527826e-05,
+      "loss": 0.728,
+      "step": 12326
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.530749100138427e-05,
+      "loss": 0.8249,
+      "step": 12327
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.529288188314925e-05,
+      "loss": 0.9578,
+      "step": 12328
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.527827443101804e-05,
+      "loss": 0.7326,
+      "step": 12329
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.5263668645435586e-05,
+      "loss": 0.8878,
+      "step": 12330
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.5249064526846695e-05,
+      "loss": 0.8924,
+      "step": 12331
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.523446207569604e-05,
+      "loss": 0.8715,
+      "step": 12332
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.521986129242843e-05,
+      "loss": 0.8641,
+      "step": 12333
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.5205262177488485e-05,
+      "loss": 0.832,
+      "step": 12334
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.5190664731320876e-05,
+      "loss": 0.8678,
+      "step": 12335
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.517606895437013e-05,
+      "loss": 0.848,
+      "step": 12336
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.516147484708077e-05,
+      "loss": 0.8753,
+      "step": 12337
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.514688240989725e-05,
+      "loss": 0.9061,
+      "step": 12338
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.513229164326395e-05,
+      "loss": 0.8776,
+      "step": 12339
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.5117702547625295e-05,
+      "loss": 0.8314,
+      "step": 12340
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.510311512342559e-05,
+      "loss": 0.8604,
+      "step": 12341
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.5088529371109054e-05,
+      "loss": 0.8314,
+      "step": 12342
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.5073945291119924e-05,
+      "loss": 0.7973,
+      "step": 12343
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.505936288390231e-05,
+      "loss": 0.7238,
+      "step": 12344
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.5044782149900386e-05,
+      "loss": 0.9032,
+      "step": 12345
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.5030203089558186e-05,
+      "loss": 0.9631,
+      "step": 12346
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.501562570331972e-05,
+      "loss": 0.9117,
+      "step": 12347
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.500104999162892e-05,
+      "loss": 0.9595,
+      "step": 12348
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.498647595492965e-05,
+      "loss": 0.7509,
+      "step": 12349
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.497190359366587e-05,
+      "loss": 0.8947,
+      "step": 12350
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.495733290828127e-05,
+      "loss": 0.7583,
+      "step": 12351
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.494276389921973e-05,
+      "loss": 0.9213,
+      "step": 12352
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.492819656692483e-05,
+      "loss": 0.7552,
+      "step": 12353
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.491363091184022e-05,
+      "loss": 0.9903,
+      "step": 12354
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.4899066934409585e-05,
+      "loss": 0.9566,
+      "step": 12355
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.4884504635076374e-05,
+      "loss": 0.8154,
+      "step": 12356
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.486994401428416e-05,
+      "loss": 1.0027,
+      "step": 12357
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.4855385072476364e-05,
+      "loss": 0.944,
+      "step": 12358
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.484082781009636e-05,
+      "loss": 0.8764,
+      "step": 12359
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.48262722275875e-05,
+      "loss": 0.8623,
+      "step": 12360
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.481171832539304e-05,
+      "loss": 0.8119,
+      "step": 12361
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.479716610395628e-05,
+      "loss": 0.7761,
+      "step": 12362
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.478261556372038e-05,
+      "loss": 0.83,
+      "step": 12363
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.4768066705128463e-05,
+      "loss": 0.8446,
+      "step": 12364
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.4753519528623636e-05,
+      "loss": 0.8876,
+      "step": 12365
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.473897403464888e-05,
+      "loss": 0.9185,
+      "step": 12366
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.472443022364726e-05,
+      "loss": 1.0161,
+      "step": 12367
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.470988809606167e-05,
+      "loss": 0.7887,
+      "step": 12368
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.4695347652334985e-05,
+      "loss": 0.7413,
+      "step": 12369
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.468080889291003e-05,
+      "loss": 0.88,
+      "step": 12370
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.4666271818229555e-05,
+      "loss": 0.8854,
+      "step": 12371
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.4651736428736355e-05,
+      "loss": 0.8699,
+      "step": 12372
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.463720272487304e-05,
+      "loss": 0.8417,
+      "step": 12373
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.462267070708233e-05,
+      "loss": 0.8332,
+      "step": 12374
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.460814037580666e-05,
+      "loss": 0.9236,
+      "step": 12375
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.459361173148865e-05,
+      "loss": 0.7544,
+      "step": 12376
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.457908477457075e-05,
+      "loss": 0.8593,
+      "step": 12377
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.456455950549533e-05,
+      "loss": 0.8155,
+      "step": 12378
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.4550035924704826e-05,
+      "loss": 0.9644,
+      "step": 12379
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.4535514032641525e-05,
+      "loss": 0.9207,
+      "step": 12380
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.452099382974769e-05,
+      "loss": 0.9254,
+      "step": 12381
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.4506475316465526e-05,
+      "loss": 0.8878,
+      "step": 12382
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.449195849323717e-05,
+      "loss": 0.9667,
+      "step": 12383
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.44774433605048e-05,
+      "loss": 0.9767,
+      "step": 12384
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.446292991871043e-05,
+      "loss": 0.7371,
+      "step": 12385
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.444841816829607e-05,
+      "loss": 0.8325,
+      "step": 12386
+    },
+    {
+      "epoch": 2.72,
+      "eval_loss": 0.5639159679412842,
+      "eval_runtime": 321.8716,
+      "eval_samples_per_second": 56.355,
+      "eval_steps_per_second": 28.179,
+      "step": 12386
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.443390810970364e-05,
+      "loss": 0.6339,
+      "step": 12387
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.441939974337512e-05,
+      "loss": 0.809,
+      "step": 12388
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.4404893069752316e-05,
+      "loss": 0.7738,
+      "step": 12389
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.4390388089277044e-05,
+      "loss": 0.8347,
+      "step": 12390
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.4375884802391044e-05,
+      "loss": 0.9912,
+      "step": 12391
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.4361383209535967e-05,
+      "loss": 0.8529,
+      "step": 12392
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.434688331115354e-05,
+      "loss": 0.9278,
+      "step": 12393
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.433238510768534e-05,
+      "loss": 0.9337,
+      "step": 12394
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.4317888599572867e-05,
+      "loss": 0.7559,
+      "step": 12395
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.430339378725764e-05,
+      "loss": 0.9546,
+      "step": 12396
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.428890067118107e-05,
+      "loss": 0.8682,
+      "step": 12397
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.4274409251784584e-05,
+      "loss": 0.8878,
+      "step": 12398
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.425991952950952e-05,
+      "loss": 0.8416,
+      "step": 12399
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.42454315047971e-05,
+      "loss": 0.6962,
+      "step": 12400
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.423094517808868e-05,
+      "loss": 0.8971,
+      "step": 12401
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.421646054982528e-05,
+      "loss": 0.8113,
+      "step": 12402
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.420197762044814e-05,
+      "loss": 0.8847,
+      "step": 12403
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.4187496390398284e-05,
+      "loss": 0.8069,
+      "step": 12404
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.417301686011679e-05,
+      "loss": 0.8963,
+      "step": 12405
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.41585390300446e-05,
+      "loss": 0.8041,
+      "step": 12406
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.414406290062264e-05,
+      "loss": 0.914,
+      "step": 12407
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.412958847229178e-05,
+      "loss": 0.8207,
+      "step": 12408
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.411511574549281e-05,
+      "loss": 0.9352,
+      "step": 12409
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.410064472066654e-05,
+      "loss": 0.7441,
+      "step": 12410
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.408617539825368e-05,
+      "loss": 0.8614,
+      "step": 12411
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.4071707778694884e-05,
+      "loss": 0.8769,
+      "step": 12412
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.4057241862430754e-05,
+      "loss": 1.0414,
+      "step": 12413
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.404277764990181e-05,
+      "loss": 0.8338,
+      "step": 12414
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.4028315141548646e-05,
+      "loss": 0.8286,
+      "step": 12415
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.401385433781169e-05,
+      "loss": 0.8616,
+      "step": 12416
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.399939523913132e-05,
+      "loss": 0.749,
+      "step": 12417
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.398493784594789e-05,
+      "loss": 0.9671,
+      "step": 12418
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.397048215870168e-05,
+      "loss": 0.9593,
+      "step": 12419
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.3956028177832996e-05,
+      "loss": 0.7574,
+      "step": 12420
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.394157590378197e-05,
+      "loss": 0.8419,
+      "step": 12421
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.3927125336988816e-05,
+      "loss": 0.8675,
+      "step": 12422
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.391267647789362e-05,
+      "loss": 0.938,
+      "step": 12423
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.3898229326936325e-05,
+      "loss": 0.7198,
+      "step": 12424
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.388378388455702e-05,
+      "loss": 0.8047,
+      "step": 12425
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.386934015119556e-05,
+      "loss": 0.9289,
+      "step": 12426
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.385489812729191e-05,
+      "loss": 0.7865,
+      "step": 12427
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.384045781328587e-05,
+      "loss": 0.8168,
+      "step": 12428
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.382601920961721e-05,
+      "loss": 0.8187,
+      "step": 12429
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.3811582316725666e-05,
+      "loss": 0.9761,
+      "step": 12430
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.3797147135050855e-05,
+      "loss": 0.9162,
+      "step": 12431
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.37827136650325e-05,
+      "loss": 0.9257,
+      "step": 12432
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.376828190711013e-05,
+      "loss": 0.9224,
+      "step": 12433
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.375385186172325e-05,
+      "loss": 0.7284,
+      "step": 12434
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.373942352931134e-05,
+      "loss": 0.8436,
+      "step": 12435
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.3724996910313764e-05,
+      "loss": 0.8749,
+      "step": 12436
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.371057200516998e-05,
+      "loss": 0.971,
+      "step": 12437
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.36961488143192e-05,
+      "loss": 0.7978,
+      "step": 12438
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.3681727338200814e-05,
+      "loss": 0.9515,
+      "step": 12439
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.36673075772539e-05,
+      "loss": 0.8374,
+      "step": 12440
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.3652889531917616e-05,
+      "loss": 0.9426,
+      "step": 12441
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.363847320263115e-05,
+      "loss": 0.8197,
+      "step": 12442
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.362405858983346e-05,
+      "loss": 0.8377,
+      "step": 12443
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.360964569396362e-05,
+      "loss": 0.7635,
+      "step": 12444
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.3595234515460584e-05,
+      "loss": 0.7465,
+      "step": 12445
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.358082505476312e-05,
+      "loss": 0.8466,
+      "step": 12446
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.356641731231019e-05,
+      "loss": 0.9222,
+      "step": 12447
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.355201128854051e-05,
+      "loss": 0.7958,
+      "step": 12448
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.353760698389286e-05,
+      "loss": 0.8884,
+      "step": 12449
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.352320439880593e-05,
+      "loss": 0.906,
+      "step": 12450
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.3508803533718314e-05,
+      "loss": 0.9663,
+      "step": 12451
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.34944043890686e-05,
+      "loss": 0.8966,
+      "step": 12452
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.3480006965295295e-05,
+      "loss": 0.9576,
+      "step": 12453
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.3465611262836915e-05,
+      "loss": 0.9127,
+      "step": 12454
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.345121728213187e-05,
+      "loss": 0.8547,
+      "step": 12455
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.343682502361852e-05,
+      "loss": 0.8302,
+      "step": 12456
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.342243448773518e-05,
+      "loss": 0.9504,
+      "step": 12457
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.340804567492007e-05,
+      "loss": 0.9066,
+      "step": 12458
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.3393658585611484e-05,
+      "loss": 0.8251,
+      "step": 12459
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.337927322024754e-05,
+      "loss": 0.8502,
+      "step": 12460
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.336488957926634e-05,
+      "loss": 0.8646,
+      "step": 12461
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.335050766310595e-05,
+      "loss": 0.7852,
+      "step": 12462
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.3336127472204314e-05,
+      "loss": 0.9052,
+      "step": 12463
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.332174900699947e-05,
+      "loss": 0.7573,
+      "step": 12464
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.3307372267929235e-05,
+      "loss": 0.9319,
+      "step": 12465
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.3292997255431566e-05,
+      "loss": 0.8279,
+      "step": 12466
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.32786239699441e-05,
+      "loss": 0.77,
+      "step": 12467
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.3264252411904696e-05,
+      "loss": 0.9626,
+      "step": 12468
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.324988258175099e-05,
+      "loss": 0.9757,
+      "step": 12469
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.323551447992058e-05,
+      "loss": 0.8352,
+      "step": 12470
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.322114810685113e-05,
+      "loss": 0.8601,
+      "step": 12471
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.3206783462980126e-05,
+      "loss": 0.8714,
+      "step": 12472
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.319242054874504e-05,
+      "loss": 0.9038,
+      "step": 12473
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.317805936458329e-05,
+      "loss": 0.8015,
+      "step": 12474
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.3163699910932224e-05,
+      "loss": 0.909,
+      "step": 12475
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.3149342188229205e-05,
+      "loss": 0.8636,
+      "step": 12476
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.3134986196911485e-05,
+      "loss": 0.7748,
+      "step": 12477
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.3120631937416275e-05,
+      "loss": 0.8795,
+      "step": 12478
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.3106279410180685e-05,
+      "loss": 0.9443,
+      "step": 12479
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.30919286156419e-05,
+      "loss": 0.8599,
+      "step": 12480
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.307757955423693e-05,
+      "loss": 0.937,
+      "step": 12481
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.3063232226402785e-05,
+      "loss": 0.8938,
+      "step": 12482
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.304888663257641e-05,
+      "loss": 0.729,
+      "step": 12483
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.3034542773194664e-05,
+      "loss": 0.7754,
+      "step": 12484
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.3020200648694455e-05,
+      "loss": 0.8439,
+      "step": 12485
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.300586025951253e-05,
+      "loss": 0.8363,
+      "step": 12486
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.299152160608562e-05,
+      "loss": 0.8315,
+      "step": 12487
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.297718468885048e-05,
+      "loss": 0.9677,
+      "step": 12488
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.296284950824361e-05,
+      "loss": 0.9247,
+      "step": 12489
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.2948516064701704e-05,
+      "loss": 0.7925,
+      "step": 12490
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.2934184358661244e-05,
+      "loss": 0.8666,
+      "step": 12491
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.2919854390558655e-05,
+      "loss": 0.9325,
+      "step": 12492
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.290552616083046e-05,
+      "loss": 1.0334,
+      "step": 12493
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.2891199669912955e-05,
+      "loss": 0.8583,
+      "step": 12494
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.287687491824246e-05,
+      "loss": 0.9766,
+      "step": 12495
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.28625519062552e-05,
+      "loss": 0.933,
+      "step": 12496
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.284823063438748e-05,
+      "loss": 0.8588,
+      "step": 12497
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.283391110307537e-05,
+      "loss": 0.8214,
+      "step": 12498
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.281959331275501e-05,
+      "loss": 0.8351,
+      "step": 12499
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.2805277263862444e-05,
+      "loss": 0.8393,
+      "step": 12500
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.2790962956833605e-05,
+      "loss": 0.9412,
+      "step": 12501
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.277665039210453e-05,
+      "loss": 0.8698,
+      "step": 12502
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.2762339570111065e-05,
+      "loss": 0.7298,
+      "step": 12503
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.274803049128905e-05,
+      "loss": 0.9635,
+      "step": 12504
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.2733723156074254e-05,
+      "loss": 0.8111,
+      "step": 12505
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 4.2719417564902396e-05,
+      "loss": 0.8403,
+      "step": 12506
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 4.27051137182092e-05,
+      "loss": 0.8253,
+      "step": 12507
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 4.269081161643026e-05,
+      "loss": 0.9158,
+      "step": 12508
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 4.267651126000115e-05,
+      "loss": 0.8837,
+      "step": 12509
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 4.266221264935738e-05,
+      "loss": 0.8374,
+      "step": 12510
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 4.264791578493439e-05,
+      "loss": 0.7345,
+      "step": 12511
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 4.263362066716765e-05,
+      "loss": 0.8177,
+      "step": 12512
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 4.261932729649246e-05,
+      "loss": 0.7811,
+      "step": 12513
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 4.260503567334417e-05,
+      "loss": 0.7388,
+      "step": 12514
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 4.259074579815807e-05,
+      "loss": 1.0509,
+      "step": 12515
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 4.257645767136922e-05,
+      "loss": 0.7531,
+      "step": 12516
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 4.256217129341287e-05,
+      "loss": 0.8197,
+      "step": 12517
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 4.2547886664724055e-05,
+      "loss": 0.827,
+      "step": 12518
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 4.2533603785737886e-05,
+      "loss": 0.8931,
+      "step": 12519
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 4.25193226568893e-05,
+      "loss": 0.7975,
+      "step": 12520
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 4.250504327861323e-05,
+      "loss": 0.8903,
+      "step": 12521
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 4.2490765651344566e-05,
+      "loss": 0.8518,
+      "step": 12522
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 4.247648977551809e-05,
+      "loss": 0.7969,
+      "step": 12523
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 4.246221565156864e-05,
+      "loss": 0.7325,
+      "step": 12524
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 4.24479432799309e-05,
+      "loss": 0.777,
+      "step": 12525
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 4.243367266103954e-05,
+      "loss": 0.8645,
+      "step": 12526
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 4.241940379532917e-05,
+      "loss": 0.8054,
+      "step": 12527
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 4.24051366832343e-05,
+      "loss": 0.8944,
+      "step": 12528
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 4.239087132518953e-05,
+      "loss": 0.7964,
+      "step": 12529
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 4.2376607721629214e-05,
+      "loss": 0.7468,
+      "step": 12530
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 4.2362345872987876e-05,
+      "loss": 0.7347,
+      "step": 12531
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 4.234808577969975e-05,
+      "loss": 0.8927,
+      "step": 12532
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 4.233382744219911e-05,
+      "loss": 0.8728,
+      "step": 12533
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 4.231957086092029e-05,
+      "loss": 0.869,
+      "step": 12534
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 4.230531603629738e-05,
+      "loss": 0.8384,
+      "step": 12535
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 4.229106296876461e-05,
+      "loss": 0.8787,
+      "step": 12536
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 4.2276811658756014e-05,
+      "loss": 0.8124,
+      "step": 12537
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 4.226256210670554e-05,
+      "loss": 0.8132,
+      "step": 12538
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 4.224831431304727e-05,
+      "loss": 0.7902,
+      "step": 12539
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 4.223406827821502e-05,
+      "loss": 0.8324,
+      "step": 12540
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 4.221982400264275e-05,
+      "loss": 0.8976,
+      "step": 12541
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 4.220558148676421e-05,
+      "loss": 0.9361,
+      "step": 12542
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 4.2191340731013184e-05,
+      "loss": 0.8204,
+      "step": 12543
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 4.217710173582335e-05,
+      "loss": 0.7878,
+      "step": 12544
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 4.216286450162834e-05,
+      "loss": 0.6961,
+      "step": 12545
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 4.21486290288618e-05,
+      "loss": 0.8307,
+      "step": 12546
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 4.213439531795724e-05,
+      "loss": 0.7741,
+      "step": 12547
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 4.2120163369348164e-05,
+      "loss": 0.7153,
+      "step": 12548
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 4.2105933183468e-05,
+      "loss": 0.8709,
+      "step": 12549
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 4.2091704760750065e-05,
+      "loss": 0.8775,
+      "step": 12550
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 4.207747810162779e-05,
+      "loss": 0.7689,
+      "step": 12551
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 4.2063253206534394e-05,
+      "loss": 0.7916,
+      "step": 12552
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 4.2049030075903096e-05,
+      "loss": 0.9358,
+      "step": 12553
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 4.2034808710167075e-05,
+      "loss": 0.7928,
+      "step": 12554
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 4.202058910975939e-05,
+      "loss": 0.9647,
+      "step": 12555
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 4.200637127511319e-05,
+      "loss": 0.8169,
+      "step": 12556
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 4.1992155206661387e-05,
+      "loss": 0.862,
+      "step": 12557
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 4.197794090483704e-05,
+      "loss": 0.8399,
+      "step": 12558
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 4.196372837007292e-05,
+      "loss": 0.8463,
+      "step": 12559
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 4.1949517602801956e-05,
+      "loss": 0.846,
+      "step": 12560
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 4.1935308603456914e-05,
+      "loss": 0.9293,
+      "step": 12561
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 4.1921101372470484e-05,
+      "loss": 0.8438,
+      "step": 12562
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 4.190689591027543e-05,
+      "loss": 0.8684,
+      "step": 12563
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 4.189269221730434e-05,
+      "loss": 0.975,
+      "step": 12564
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 4.187849029398977e-05,
+      "loss": 0.8197,
+      "step": 12565
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 4.186429014076426e-05,
+      "loss": 0.9212,
+      "step": 12566
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 4.185009175806024e-05,
+      "loss": 0.8287,
+      "step": 12567
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 4.1835895146310166e-05,
+      "loss": 0.806,
+      "step": 12568
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 4.1821700305946396e-05,
+      "loss": 0.8311,
+      "step": 12569
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 4.180750723740121e-05,
+      "loss": 1.0815,
+      "step": 12570
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 4.1793315941106825e-05,
+      "loss": 0.8757,
+      "step": 12571
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 4.177912641749552e-05,
+      "loss": 0.782,
+      "step": 12572
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 4.1764938666999375e-05,
+      "loss": 0.6981,
+      "step": 12573
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 4.175075269005051e-05,
+      "loss": 0.8298,
+      "step": 12574
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 4.173656848708094e-05,
+      "loss": 0.8574,
+      "step": 12575
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 4.172238605852261e-05,
+      "loss": 0.8652,
+      "step": 12576
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 4.1708205404807524e-05,
+      "loss": 0.9359,
+      "step": 12577
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 4.1694026526367506e-05,
+      "loss": 0.8403,
+      "step": 12578
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 4.167984942363434e-05,
+      "loss": 0.927,
+      "step": 12579
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 4.166567409703991e-05,
+      "loss": 0.7272,
+      "step": 12580
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 4.165150054701577e-05,
+      "loss": 1.037,
+      "step": 12581
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 4.163732877399368e-05,
+      "loss": 0.9042,
+      "step": 12582
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 4.1623158778405216e-05,
+      "loss": 0.9334,
+      "step": 12583
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 4.160899056068188e-05,
+      "loss": 0.9558,
+      "step": 12584
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 4.1594824121255236e-05,
+      "loss": 0.9996,
+      "step": 12585
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 4.158065946055669e-05,
+      "loss": 0.7588,
+      "step": 12586
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 4.156649657901765e-05,
+      "loss": 0.8999,
+      "step": 12587
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 4.155233547706936e-05,
+      "loss": 0.8929,
+      "step": 12588
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 4.153817615514321e-05,
+      "loss": 0.8001,
+      "step": 12589
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 4.1524018613670356e-05,
+      "loss": 0.7634,
+      "step": 12590
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 4.150986285308199e-05,
+      "loss": 0.9662,
+      "step": 12591
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 4.1495708873809216e-05,
+      "loss": 0.7734,
+      "step": 12592
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 4.1481556676283064e-05,
+      "loss": 0.8559,
+      "step": 12593
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 4.1467406260934597e-05,
+      "loss": 0.8289,
+      "step": 12594
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 4.145325762819473e-05,
+      "loss": 0.8633,
+      "step": 12595
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 4.143911077849436e-05,
+      "loss": 0.8921,
+      "step": 12596
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 4.142496571226435e-05,
+      "loss": 0.7706,
+      "step": 12597
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 4.1410822429935425e-05,
+      "loss": 0.8252,
+      "step": 12598
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 4.139668093193841e-05,
+      "loss": 0.9052,
+      "step": 12599
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 4.138254121870394e-05,
+      "loss": 0.7705,
+      "step": 12600
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 4.1368403290662625e-05,
+      "loss": 1.0177,
+      "step": 12601
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 4.1354267148245065e-05,
+      "loss": 0.7837,
+      "step": 12602
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 4.134013279188171e-05,
+      "loss": 0.9116,
+      "step": 12603
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 4.132600022200311e-05,
+      "loss": 0.7145,
+      "step": 12604
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 4.131186943903961e-05,
+      "loss": 0.8668,
+      "step": 12605
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 4.1297740443421604e-05,
+      "loss": 0.9761,
+      "step": 12606
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 4.128361323557942e-05,
+      "loss": 0.8119,
+      "step": 12607
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 4.1269487815943185e-05,
+      "loss": 0.6985,
+      "step": 12608
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 4.125536418494318e-05,
+      "loss": 0.8319,
+      "step": 12609
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 4.124124234300949e-05,
+      "loss": 0.8548,
+      "step": 12610
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 4.1227122290572264e-05,
+      "loss": 0.8492,
+      "step": 12611
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 4.1213004028061474e-05,
+      "loss": 0.7636,
+      "step": 12612
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 4.119888755590711e-05,
+      "loss": 0.9291,
+      "step": 12613
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 4.118477287453907e-05,
+      "loss": 0.9816,
+      "step": 12614
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 4.1170659984387203e-05,
+      "loss": 0.8096,
+      "step": 12615
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 4.115654888588138e-05,
+      "loss": 0.7856,
+      "step": 12616
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 4.114243957945131e-05,
+      "loss": 0.9501,
+      "step": 12617
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 4.11283320655267e-05,
+      "loss": 0.8799,
+      "step": 12618
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 4.11142263445372e-05,
+      "loss": 0.8391,
+      "step": 12619
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 4.1100122416912344e-05,
+      "loss": 0.8028,
+      "step": 12620
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 4.108602028308175e-05,
+      "loss": 1.047,
+      "step": 12621
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 4.107191994347483e-05,
+      "loss": 0.8537,
+      "step": 12622
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 4.105782139852111e-05,
+      "loss": 0.8165,
+      "step": 12623
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 4.104372464864986e-05,
+      "loss": 0.8517,
+      "step": 12624
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 4.102962969429039e-05,
+      "loss": 0.7692,
+      "step": 12625
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 4.1015536535872046e-05,
+      "loss": 0.8955,
+      "step": 12626
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 4.100144517382394e-05,
+      "loss": 0.7115,
+      "step": 12627
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 4.098735560857532e-05,
+      "loss": 0.9074,
+      "step": 12628
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 4.0973267840555265e-05,
+      "loss": 0.8825,
+      "step": 12629
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 4.0959181870192706e-05,
+      "loss": 0.8778,
+      "step": 12630
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 4.094509769791677e-05,
+      "loss": 0.7899,
+      "step": 12631
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 4.093101532415628e-05,
+      "loss": 0.7913,
+      "step": 12632
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 4.091693474934022e-05,
+      "loss": 0.7961,
+      "step": 12633
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 4.090285597389735e-05,
+      "loss": 0.7604,
+      "step": 12634
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 4.088877899825645e-05,
+      "loss": 0.7946,
+      "step": 12635
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 4.087470382284624e-05,
+      "loss": 0.8846,
+      "step": 12636
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 4.086063044809534e-05,
+      "loss": 0.8276,
+      "step": 12637
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 4.084655887443242e-05,
+      "loss": 0.8888,
+      "step": 12638
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 4.083248910228599e-05,
+      "loss": 0.813,
+      "step": 12639
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 4.081842113208456e-05,
+      "loss": 0.8321,
+      "step": 12640
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 4.0804354964256553e-05,
+      "loss": 0.776,
+      "step": 12641
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 4.079029059923033e-05,
+      "loss": 0.887,
+      "step": 12642
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 4.07762280374343e-05,
+      "loss": 0.8331,
+      "step": 12643
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 4.0762167279296636e-05,
+      "loss": 0.9201,
+      "step": 12644
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 4.0748108325245695e-05,
+      "loss": 0.8244,
+      "step": 12645
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 4.073405117570952e-05,
+      "loss": 0.7974,
+      "step": 12646
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 4.071999583111623e-05,
+      "loss": 0.8802,
+      "step": 12647
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 4.070594229189394e-05,
+      "loss": 0.9325,
+      "step": 12648
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 4.069189055847059e-05,
+      "loss": 0.9201,
+      "step": 12649
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 4.067784063127419e-05,
+      "loss": 0.9276,
+      "step": 12650
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 4.06637925107326e-05,
+      "loss": 0.8749,
+      "step": 12651
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 4.064974619727365e-05,
+      "loss": 0.8335,
+      "step": 12652
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 4.063570169132512e-05,
+      "loss": 0.9872,
+      "step": 12653
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 4.062165899331471e-05,
+      "loss": 0.6984,
+      "step": 12654
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 4.060761810367014e-05,
+      "loss": 0.83,
+      "step": 12655
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 4.0593579022819006e-05,
+      "loss": 0.8029,
+      "step": 12656
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 4.0579541751188864e-05,
+      "loss": 0.8861,
+      "step": 12657
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 4.056550628920722e-05,
+      "loss": 0.8153,
+      "step": 12658
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 4.055147263730148e-05,
+      "loss": 0.8976,
+      "step": 12659
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 4.0537440795899125e-05,
+      "loss": 0.9,
+      "step": 12660
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 4.052341076542745e-05,
+      "loss": 0.8863,
+      "step": 12661
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 4.050938254631373e-05,
+      "loss": 0.8501,
+      "step": 12662
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 4.0495356138985176e-05,
+      "loss": 0.8536,
+      "step": 12663
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 4.048133154386902e-05,
+      "loss": 0.8706,
+      "step": 12664
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 4.046730876139235e-05,
+      "loss": 0.8438,
+      "step": 12665
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 4.045328779198223e-05,
+      "loss": 1.0017,
+      "step": 12666
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 4.043926863606567e-05,
+      "loss": 0.8489,
+      "step": 12667
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 4.0425251294069585e-05,
+      "loss": 0.7663,
+      "step": 12668
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 4.041123576642094e-05,
+      "loss": 0.808,
+      "step": 12669
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 4.039722205354654e-05,
+      "loss": 0.8065,
+      "step": 12670
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 4.0383210155873155e-05,
+      "loss": 0.802,
+      "step": 12671
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 4.036920007382761e-05,
+      "loss": 0.9469,
+      "step": 12672
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 4.0355191807836435e-05,
+      "loss": 0.8417,
+      "step": 12673
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 4.034118535832636e-05,
+      "loss": 0.9393,
+      "step": 12674
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 4.0327180725723925e-05,
+      "loss": 0.7815,
+      "step": 12675
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 4.03131779104556e-05,
+      "loss": 0.7743,
+      "step": 12676
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 4.029917691294791e-05,
+      "loss": 0.888,
+      "step": 12677
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 4.0285177733627225e-05,
+      "loss": 0.9007,
+      "step": 12678
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 4.027118037291987e-05,
+      "loss": 0.902,
+      "step": 12679
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 4.025718483125213e-05,
+      "loss": 0.8559,
+      "step": 12680
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 4.0243191109050284e-05,
+      "loss": 0.8767,
+      "step": 12681
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 4.022919920674049e-05,
+      "loss": 0.8019,
+      "step": 12682
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 4.0215209124748856e-05,
+      "loss": 0.8766,
+      "step": 12683
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 4.020122086350147e-05,
+      "loss": 0.9816,
+      "step": 12684
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 4.018723442342428e-05,
+      "loss": 0.9188,
+      "step": 12685
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 4.017324980494335e-05,
+      "loss": 0.9332,
+      "step": 12686
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 4.0159267008484516e-05,
+      "loss": 0.9629,
+      "step": 12687
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 4.0145286034473636e-05,
+      "loss": 0.8142,
+      "step": 12688
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 4.0131306883336504e-05,
+      "loss": 0.908,
+      "step": 12689
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 4.011732955549882e-05,
+      "loss": 0.9238,
+      "step": 12690
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 4.010335405138631e-05,
+      "loss": 0.769,
+      "step": 12691
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 4.008938037142459e-05,
+      "loss": 0.7878,
+      "step": 12692
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 4.007540851603919e-05,
+      "loss": 0.9211,
+      "step": 12693
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 4.006143848565571e-05,
+      "loss": 0.9014,
+      "step": 12694
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 4.0047470280699485e-05,
+      "loss": 0.87,
+      "step": 12695
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 4.0033503901596015e-05,
+      "loss": 0.9864,
+      "step": 12696
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 4.0019539348770565e-05,
+      "loss": 0.8484,
+      "step": 12697
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 4.000557662264851e-05,
+      "loss": 0.8798,
+      "step": 12698
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 3.999161572365505e-05,
+      "loss": 0.8969,
+      "step": 12699
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 3.9977656652215344e-05,
+      "loss": 0.9306,
+      "step": 12700
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 3.996369940875454e-05,
+      "loss": 0.8393,
+      "step": 12701
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 3.994974399369766e-05,
+      "loss": 0.864,
+      "step": 12702
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 3.9935790407469784e-05,
+      "loss": 0.9565,
+      "step": 12703
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 3.992183865049584e-05,
+      "loss": 0.9452,
+      "step": 12704
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 3.990788872320071e-05,
+      "loss": 0.8052,
+      "step": 12705
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 3.9893940626009265e-05,
+      "loss": 0.8689,
+      "step": 12706
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 3.987999435934624e-05,
+      "loss": 0.9147,
+      "step": 12707
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 3.986604992363644e-05,
+      "loss": 0.8244,
+      "step": 12708
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 3.985210731930452e-05,
+      "loss": 0.7934,
+      "step": 12709
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 3.983816654677508e-05,
+      "loss": 0.8431,
+      "step": 12710
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 3.982422760647271e-05,
+      "loss": 0.8176,
+      "step": 12711
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 3.981029049882187e-05,
+      "loss": 0.8355,
+      "step": 12712
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 3.9796355224247095e-05,
+      "loss": 0.8725,
+      "step": 12713
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 3.97824217831727e-05,
+      "loss": 0.9356,
+      "step": 12714
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 3.976849017602313e-05,
+      "loss": 0.9538,
+      "step": 12715
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 3.9754560403222585e-05,
+      "loss": 0.8776,
+      "step": 12716
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 3.974063246519529e-05,
+      "loss": 0.9248,
+      "step": 12717
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 3.972670636236548e-05,
+      "loss": 0.996,
+      "step": 12718
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 3.971278209515722e-05,
+      "loss": 0.9012,
+      "step": 12719
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 3.9698859663994624e-05,
+      "loss": 0.839,
+      "step": 12720
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 3.968493906930172e-05,
+      "loss": 0.8104,
+      "step": 12721
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 3.9671020311502324e-05,
+      "loss": 1.0264,
+      "step": 12722
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 3.965710339102048e-05,
+      "loss": 0.88,
+      "step": 12723
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 3.964318830827992e-05,
+      "loss": 0.8282,
+      "step": 12724
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 3.962927506370451e-05,
+      "loss": 0.9125,
+      "step": 12725
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 3.961536365771795e-05,
+      "loss": 0.9047,
+      "step": 12726
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 3.9601454090743904e-05,
+      "loss": 0.9621,
+      "step": 12727
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 3.9587546363205985e-05,
+      "loss": 0.9423,
+      "step": 12728
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 3.957364047552772e-05,
+      "loss": 0.8491,
+      "step": 12729
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 3.9559736428132686e-05,
+      "loss": 0.7651,
+      "step": 12730
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 3.954583422144429e-05,
+      "loss": 0.8061,
+      "step": 12731
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 3.953193385588593e-05,
+      "loss": 0.8225,
+      "step": 12732
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 3.951803533188093e-05,
+      "loss": 0.8643,
+      "step": 12733
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 3.950413864985255e-05,
+      "loss": 0.8983,
+      "step": 12734
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 3.949024381022407e-05,
+      "loss": 0.9727,
+      "step": 12735
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 3.94763508134186e-05,
+      "loss": 0.8751,
+      "step": 12736
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 3.946245965985934e-05,
+      "loss": 0.9472,
+      "step": 12737
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 3.944857034996924e-05,
+      "loss": 0.9891,
+      "step": 12738
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 3.943468288417131e-05,
+      "loss": 0.9536,
+      "step": 12739
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 3.9420797262888566e-05,
+      "loss": 0.7972,
+      "step": 12740
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 3.9406913486543815e-05,
+      "loss": 0.8723,
+      "step": 12741
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 3.939303155555996e-05,
+      "loss": 0.8003,
+      "step": 12742
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 3.937915147035973e-05,
+      "loss": 0.8853,
+      "step": 12743
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 3.936527323136585e-05,
+      "loss": 0.8158,
+      "step": 12744
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 3.935139683900099e-05,
+      "loss": 0.8136,
+      "step": 12745
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 3.933752229368771e-05,
+      "loss": 0.8554,
+      "step": 12746
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 3.932364959584863e-05,
+      "loss": 0.864,
+      "step": 12747
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 3.930977874590621e-05,
+      "loss": 0.8054,
+      "step": 12748
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 3.929590974428288e-05,
+      "loss": 0.915,
+      "step": 12749
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 3.9282042591401036e-05,
+      "loss": 0.9227,
+      "step": 12750
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 3.926817728768295e-05,
+      "loss": 0.893,
+      "step": 12751
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 3.925431383355096e-05,
+      "loss": 1.0089,
+      "step": 12752
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 3.924045222942726e-05,
+      "loss": 0.7856,
+      "step": 12753
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 3.922659247573398e-05,
+      "loss": 0.8769,
+      "step": 12754
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 3.9212734572893205e-05,
+      "loss": 0.8262,
+      "step": 12755
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 3.919887852132704e-05,
+      "loss": 0.7834,
+      "step": 12756
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 3.918502432145743e-05,
+      "loss": 0.8478,
+      "step": 12757
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 3.91711719737063e-05,
+      "loss": 0.7976,
+      "step": 12758
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 3.915732147849555e-05,
+      "loss": 0.8555,
+      "step": 12759
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 3.914347283624692e-05,
+      "loss": 0.9661,
+      "step": 12760
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 3.912962604738228e-05,
+      "loss": 0.8572,
+      "step": 12761
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 3.911578111232327e-05,
+      "loss": 0.9127,
+      "step": 12762
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 3.910193803149152e-05,
+      "loss": 0.9738,
+      "step": 12763
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 3.908809680530872e-05,
+      "loss": 0.7726,
+      "step": 12764
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 3.907425743419626e-05,
+      "loss": 0.9544,
+      "step": 12765
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 3.9060419918575716e-05,
+      "loss": 0.8697,
+      "step": 12766
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 3.9046584258868504e-05,
+      "loss": 0.7017,
+      "step": 12767
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 3.9032750455495925e-05,
+      "loss": 0.8943,
+      "step": 12768
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 3.9018918508879366e-05,
+      "loss": 0.8185,
+      "step": 12769
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 3.900508841944004e-05,
+      "loss": 0.9782,
+      "step": 12770
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 3.899126018759914e-05,
+      "loss": 0.8708,
+      "step": 12771
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 3.8977433813777784e-05,
+      "loss": 0.922,
+      "step": 12772
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 3.896360929839711e-05,
+      "loss": 1.1005,
+      "step": 12773
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 3.894978664187812e-05,
+      "loss": 0.8414,
+      "step": 12774
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 3.8935965844641765e-05,
+      "loss": 0.8411,
+      "step": 12775
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 3.8922146907108966e-05,
+      "loss": 0.8427,
+      "step": 12776
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 3.890832982970056e-05,
+      "loss": 0.8377,
+      "step": 12777
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 3.8894514612837394e-05,
+      "loss": 0.8084,
+      "step": 12778
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 3.888070125694018e-05,
+      "loss": 0.8215,
+      "step": 12779
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 3.8866889762429605e-05,
+      "loss": 0.9843,
+      "step": 12780
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 3.8853080129726295e-05,
+      "loss": 0.9043,
+      "step": 12781
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 3.8839272359250786e-05,
+      "loss": 0.8278,
+      "step": 12782
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 3.882546645142368e-05,
+      "loss": 0.7643,
+      "step": 12783
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 3.8811662406665386e-05,
+      "loss": 0.8355,
+      "step": 12784
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 3.879786022539627e-05,
+      "loss": 0.8651,
+      "step": 12785
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 3.8784059908036777e-05,
+      "loss": 0.7562,
+      "step": 12786
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 3.877026145500706e-05,
+      "loss": 0.8774,
+      "step": 12787
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 3.875646486672747e-05,
+      "loss": 0.8887,
+      "step": 12788
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 3.87426701436181e-05,
+      "loss": 0.8112,
+      "step": 12789
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 3.872887728609913e-05,
+      "loss": 0.7131,
+      "step": 12790
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 3.87150862945906e-05,
+      "loss": 0.8273,
+      "step": 12791
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 3.870129716951251e-05,
+      "loss": 0.8435,
+      "step": 12792
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 3.868750991128479e-05,
+      "loss": 0.8713,
+      "step": 12793
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 3.8673724520327326e-05,
+      "loss": 0.9804,
+      "step": 12794
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 3.865994099706001e-05,
+      "loss": 0.8976,
+      "step": 12795
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 3.8646159341902576e-05,
+      "loss": 0.6998,
+      "step": 12796
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 3.863237955527475e-05,
+      "loss": 0.8012,
+      "step": 12797
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 3.861860163759618e-05,
+      "loss": 0.8456,
+      "step": 12798
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 3.8604825589286475e-05,
+      "loss": 0.8593,
+      "step": 12799
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 3.859105141076521e-05,
+      "loss": 0.8357,
+      "step": 12800
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 3.8577279102451844e-05,
+      "loss": 0.7447,
+      "step": 12801
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 3.856350866476589e-05,
+      "loss": 0.8969,
+      "step": 12802
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 3.854974009812664e-05,
+      "loss": 0.835,
+      "step": 12803
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 3.85359734029534e-05,
+      "loss": 0.9006,
+      "step": 12804
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 3.852220857966551e-05,
+      "loss": 0.8801,
+      "step": 12805
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 3.850844562868212e-05,
+      "loss": 0.9399,
+      "step": 12806
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 3.849468455042243e-05,
+      "loss": 0.9674,
+      "step": 12807
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 3.848092534530555e-05,
+      "loss": 0.8609,
+      "step": 12808
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 3.8467168013750396e-05,
+      "loss": 0.8812,
+      "step": 12809
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 3.845341255617606e-05,
+      "loss": 0.8236,
+      "step": 12810
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 3.84396589730014e-05,
+      "loss": 0.8155,
+      "step": 12811
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 3.842590726464535e-05,
+      "loss": 0.731,
+      "step": 12812
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 3.841215743152671e-05,
+      "loss": 0.8012,
+      "step": 12813
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 3.8398409474064126e-05,
+      "loss": 0.7298,
+      "step": 12814
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 3.838466339267639e-05,
+      "loss": 0.825,
+      "step": 12815
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 3.837091918778208e-05,
+      "loss": 0.7982,
+      "step": 12816
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 3.835717685979986e-05,
+      "loss": 0.8321,
+      "step": 12817
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 3.8343436409148194e-05,
+      "loss": 0.8572,
+      "step": 12818
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 3.832969783624555e-05,
+      "loss": 0.8571,
+      "step": 12819
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 3.8315961141510336e-05,
+      "loss": 0.9517,
+      "step": 12820
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 3.830222632536088e-05,
+      "loss": 0.8658,
+      "step": 12821
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 3.828849338821553e-05,
+      "loss": 0.9251,
+      "step": 12822
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 3.82747623304925e-05,
+      "loss": 0.8166,
+      "step": 12823
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 3.826103315260996e-05,
+      "loss": 0.7816,
+      "step": 12824
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 3.824730585498604e-05,
+      "loss": 0.9751,
+      "step": 12825
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 3.823358043803876e-05,
+      "loss": 0.8745,
+      "step": 12826
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 3.82198569021862e-05,
+      "loss": 0.7703,
+      "step": 12827
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 3.820613524784624e-05,
+      "loss": 0.8265,
+      "step": 12828
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 3.819241547543688e-05,
+      "loss": 0.7864,
+      "step": 12829
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 3.8178697585375857e-05,
+      "loss": 0.9097,
+      "step": 12830
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 3.816498157808094e-05,
+      "loss": 0.8594,
+      "step": 12831
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 3.815126745396992e-05,
+      "loss": 0.8701,
+      "step": 12832
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 3.813755521346038e-05,
+      "loss": 0.9309,
+      "step": 12833
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 3.812384485697003e-05,
+      "loss": 0.8041,
+      "step": 12834
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 3.811013638491634e-05,
+      "loss": 0.7962,
+      "step": 12835
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 3.809642979771683e-05,
+      "loss": 0.8968,
+      "step": 12836
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 3.808272509578892e-05,
+      "loss": 0.9317,
+      "step": 12837
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 3.806902227954996e-05,
+      "loss": 0.892,
+      "step": 12838
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 3.805532134941734e-05,
+      "loss": 0.8196,
+      "step": 12839
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 3.8041622305808274e-05,
+      "loss": 0.8785,
+      "step": 12840
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 3.802792514913999e-05,
+      "loss": 0.7816,
+      "step": 12841
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 3.801422987982961e-05,
+      "loss": 0.8801,
+      "step": 12842
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 3.800053649829419e-05,
+      "loss": 0.8662,
+      "step": 12843
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 3.798684500495086e-05,
+      "loss": 0.8284,
+      "step": 12844
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 3.797315540021652e-05,
+      "loss": 0.856,
+      "step": 12845
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 3.79594676845081e-05,
+      "loss": 0.895,
+      "step": 12846
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 3.794578185824243e-05,
+      "loss": 0.9282,
+      "step": 12847
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 3.793209792183639e-05,
+      "loss": 0.9033,
+      "step": 12848
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 3.791841587570668e-05,
+      "loss": 0.7731,
+      "step": 12849
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 3.7904735720269946e-05,
+      "loss": 0.8345,
+      "step": 12850
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 3.7891057455942924e-05,
+      "loss": 0.959,
+      "step": 12851
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 3.787738108314205e-05,
+      "loss": 0.8705,
+      "step": 12852
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 3.7863706602283934e-05,
+      "loss": 0.9095,
+      "step": 12853
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 3.7850034013785005e-05,
+      "loss": 0.9218,
+      "step": 12854
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 3.783636331806162e-05,
+      "loss": 0.9218,
+      "step": 12855
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 3.78226945155302e-05,
+      "loss": 0.9616,
+      "step": 12856
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 3.7809027606606986e-05,
+      "loss": 0.8445,
+      "step": 12857
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 3.7795362591708204e-05,
+      "loss": 0.9322,
+      "step": 12858
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 3.778169947125002e-05,
+      "loss": 0.8458,
+      "step": 12859
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 3.7768038245648505e-05,
+      "loss": 0.8681,
+      "step": 12860
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 3.77543789153198e-05,
+      "loss": 0.8718,
+      "step": 12861
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 3.774072148067984e-05,
+      "loss": 0.8758,
+      "step": 12862
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 3.772706594214458e-05,
+      "loss": 0.8536,
+      "step": 12863
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 3.771341230012986e-05,
+      "loss": 0.7825,
+      "step": 12864
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 3.7699760555051554e-05,
+      "loss": 0.7862,
+      "step": 12865
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 3.768611070732542e-05,
+      "loss": 0.8298,
+      "step": 12866
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 3.767246275736714e-05,
+      "loss": 0.8614,
+      "step": 12867
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 3.7658816705592373e-05,
+      "loss": 0.9198,
+      "step": 12868
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 3.7645172552416674e-05,
+      "loss": 0.9023,
+      "step": 12869
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 3.7631530298255645e-05,
+      "loss": 0.8918,
+      "step": 12870
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 3.761788994352472e-05,
+      "loss": 0.8677,
+      "step": 12871
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 3.760425148863932e-05,
+      "loss": 0.8832,
+      "step": 12872
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 3.75906149340148e-05,
+      "loss": 0.9356,
+      "step": 12873
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 3.757698028006643e-05,
+      "loss": 0.8302,
+      "step": 12874
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 3.756334752720954e-05,
+      "loss": 0.8667,
+      "step": 12875
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 3.754971667585924e-05,
+      "loss": 0.7742,
+      "step": 12876
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 3.753608772643065e-05,
+      "loss": 0.971,
+      "step": 12877
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 3.7522460679338946e-05,
+      "loss": 0.9563,
+      "step": 12878
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 3.7508835534998984e-05,
+      "loss": 0.9236,
+      "step": 12879
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 3.7495212293825835e-05,
+      "loss": 0.8441,
+      "step": 12880
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 3.7481590956234315e-05,
+      "loss": 0.7824,
+      "step": 12881
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 3.746797152263935e-05,
+      "loss": 1.1004,
+      "step": 12882
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 3.7454353993455646e-05,
+      "loss": 0.7569,
+      "step": 12883
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 3.7440738369097974e-05,
+      "loss": 0.8249,
+      "step": 12884
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 3.742712464998096e-05,
+      "loss": 0.9293,
+      "step": 12885
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 3.741351283651919e-05,
+      "loss": 0.8304,
+      "step": 12886
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 3.7399902929127285e-05,
+      "loss": 0.9189,
+      "step": 12887
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 3.738629492821969e-05,
+      "loss": 0.8538,
+      "step": 12888
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 3.7372688834210836e-05,
+      "loss": 0.8521,
+      "step": 12889
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 3.73590846475151e-05,
+      "loss": 0.9096,
+      "step": 12890
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 3.734548236854676e-05,
+      "loss": 0.9325,
+      "step": 12891
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 3.733188199772015e-05,
+      "loss": 0.907,
+      "step": 12892
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 3.7318283535449396e-05,
+      "loss": 0.8068,
+      "step": 12893
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 3.730468698214874e-05,
+      "loss": 0.7407,
+      "step": 12894
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 3.729109233823216e-05,
+      "loss": 0.9304,
+      "step": 12895
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 3.727749960411368e-05,
+      "loss": 1.0342,
+      "step": 12896
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 3.7263908780207336e-05,
+      "loss": 0.9002,
+      "step": 12897
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 3.7250319866926976e-05,
+      "loss": 0.9562,
+      "step": 12898
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 3.7236732864686506e-05,
+      "loss": 0.7368,
+      "step": 12899
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 3.7223147773899724e-05,
+      "loss": 0.8756,
+      "step": 12900
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 3.720956459498026e-05,
+      "loss": 0.8911,
+      "step": 12901
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 3.71959833283419e-05,
+      "loss": 0.9483,
+      "step": 12902
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 3.718240397439818e-05,
+      "loss": 0.9057,
+      "step": 12903
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 3.716882653356274e-05,
+      "loss": 0.8843,
+      "step": 12904
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 3.715525100624903e-05,
+      "loss": 0.8999,
+      "step": 12905
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 3.71416773928705e-05,
+      "loss": 0.8712,
+      "step": 12906
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 3.712810569384054e-05,
+      "loss": 0.8533,
+      "step": 12907
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 3.711453590957243e-05,
+      "loss": 0.8862,
+      "step": 12908
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 3.7100968040479523e-05,
+      "loss": 0.7189,
+      "step": 12909
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 3.7087402086974975e-05,
+      "loss": 0.9248,
+      "step": 12910
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 3.707383804947194e-05,
+      "loss": 0.805,
+      "step": 12911
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 3.706027592838352e-05,
+      "loss": 0.911,
+      "step": 12912
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 3.704671572412273e-05,
+      "loss": 0.853,
+      "step": 12913
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 3.703315743710257e-05,
+      "loss": 0.7358,
+      "step": 12914
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 3.7019601067735954e-05,
+      "loss": 0.8572,
+      "step": 12915
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 3.700604661643574e-05,
+      "loss": 0.9235,
+      "step": 12916
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 3.699249408361473e-05,
+      "loss": 0.8629,
+      "step": 12917
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 3.69789434696856e-05,
+      "loss": 0.8974,
+      "step": 12918
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 3.696539477506115e-05,
+      "loss": 0.8351,
+      "step": 12919
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 3.69518480001539e-05,
+      "loss": 0.9379,
+      "step": 12920
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 3.693830314537654e-05,
+      "loss": 0.8497,
+      "step": 12921
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 3.692476021114146e-05,
+      "loss": 0.8675,
+      "step": 12922
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 3.691121919786112e-05,
+      "loss": 0.8205,
+      "step": 12923
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 3.689768010594798e-05,
+      "loss": 0.8828,
+      "step": 12924
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 3.6884142935814316e-05,
+      "loss": 0.7884,
+      "step": 12925
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 3.687060768787244e-05,
+      "loss": 0.9249,
+      "step": 12926
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 3.685707436253455e-05,
+      "loss": 0.8975,
+      "step": 12927
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 3.684354296021281e-05,
+      "loss": 0.8475,
+      "step": 12928
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 3.68300134813193e-05,
+      "loss": 0.8113,
+      "step": 12929
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 3.681648592626605e-05,
+      "loss": 0.8648,
+      "step": 12930
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 3.680296029546509e-05,
+      "loss": 0.9411,
+      "step": 12931
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 3.678943658932832e-05,
+      "loss": 0.8727,
+      "step": 12932
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 3.6775914808267596e-05,
+      "loss": 0.7729,
+      "step": 12933
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 3.676239495269473e-05,
+      "loss": 0.8226,
+      "step": 12934
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 3.6748877023021434e-05,
+      "loss": 0.8203,
+      "step": 12935
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 3.673536101965945e-05,
+      "loss": 0.7724,
+      "step": 12936
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 3.67218469430204e-05,
+      "loss": 0.8846,
+      "step": 12937
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 3.670833479351583e-05,
+      "loss": 0.8076,
+      "step": 12938
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 3.669482457155724e-05,
+      "loss": 0.8664,
+      "step": 12939
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 3.668131627755613e-05,
+      "loss": 1.005,
+      "step": 12940
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 3.6667809911923877e-05,
+      "loss": 0.8952,
+      "step": 12941
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 3.665430547507177e-05,
+      "loss": 0.7953,
+      "step": 12942
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 3.6640802967411205e-05,
+      "loss": 0.9689,
+      "step": 12943
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 3.662730238935325e-05,
+      "loss": 0.9297,
+      "step": 12944
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 3.6613803741309175e-05,
+      "loss": 0.9327,
+      "step": 12945
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 3.660030702369003e-05,
+      "loss": 0.8386,
+      "step": 12946
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 3.658681223690684e-05,
+      "loss": 0.792,
+      "step": 12947
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 3.6573319381370665e-05,
+      "loss": 0.8369,
+      "step": 12948
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 3.655982845749237e-05,
+      "loss": 0.8465,
+      "step": 12949
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 3.6546339465682844e-05,
+      "loss": 0.8243,
+      "step": 12950
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 3.653285240635289e-05,
+      "loss": 0.8308,
+      "step": 12951
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 3.651936727991321e-05,
+      "loss": 0.9011,
+      "step": 12952
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 3.6505884086774556e-05,
+      "loss": 0.8895,
+      "step": 12953
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 3.649240282734755e-05,
+      "loss": 0.8395,
+      "step": 12954
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 3.647892350204274e-05,
+      "loss": 0.838,
+      "step": 12955
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 3.6465446111270616e-05,
+      "loss": 0.9085,
+      "step": 12956
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 3.6451970655441705e-05,
+      "loss": 0.8393,
+      "step": 12957
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 3.6438497134966344e-05,
+      "loss": 0.6878,
+      "step": 12958
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 3.6425025550254896e-05,
+      "loss": 0.8937,
+      "step": 12959
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 3.641155590171762e-05,
+      "loss": 0.8692,
+      "step": 12960
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 3.6398088189764715e-05,
+      "loss": 0.7946,
+      "step": 12961
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 3.638462241480639e-05,
+      "loss": 0.857,
+      "step": 12962
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 3.637115857725273e-05,
+      "loss": 0.7999,
+      "step": 12963
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 3.635769667751372e-05,
+      "loss": 0.8891,
+      "step": 12964
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 3.634423671599946e-05,
+      "loss": 0.9369,
+      "step": 12965
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 3.633077869311973e-05,
+      "loss": 0.8356,
+      "step": 12966
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 3.63173226092845e-05,
+      "loss": 0.856,
+      "step": 12967
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 3.630386846490354e-05,
+      "loss": 0.7962,
+      "step": 12968
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 3.629041626038656e-05,
+      "loss": 0.9114,
+      "step": 12969
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 3.627696599614334e-05,
+      "loss": 0.9599,
+      "step": 12970
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 3.626351767258339e-05,
+      "loss": 0.8382,
+      "step": 12971
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 3.6250071290116375e-05,
+      "loss": 0.9703,
+      "step": 12972
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 3.6236626849151725e-05,
+      "loss": 0.846,
+      "step": 12973
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 3.622318435009896e-05,
+      "loss": 0.8591,
+      "step": 12974
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 3.6209743793367444e-05,
+      "loss": 0.9555,
+      "step": 12975
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 3.619630517936651e-05,
+      "loss": 0.764,
+      "step": 12976
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 3.6182868508505416e-05,
+      "loss": 0.9342,
+      "step": 12977
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 3.616943378119337e-05,
+      "loss": 0.9032,
+      "step": 12978
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 3.6156000997839565e-05,
+      "loss": 0.8621,
+      "step": 12979
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 3.614257015885309e-05,
+      "loss": 0.998,
+      "step": 12980
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 3.612914126464294e-05,
+      "loss": 0.8117,
+      "step": 12981
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 3.611571431561813e-05,
+      "loss": 0.8104,
+      "step": 12982
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 3.610228931218753e-05,
+      "loss": 0.8273,
+      "step": 12983
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 3.6088866254760066e-05,
+      "loss": 0.7841,
+      "step": 12984
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 3.6075445143744466e-05,
+      "loss": 0.8706,
+      "step": 12985
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 3.6062025979549574e-05,
+      "loss": 0.9096,
+      "step": 12986
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 3.604860876258397e-05,
+      "loss": 0.9602,
+      "step": 12987
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 3.603519349325628e-05,
+      "loss": 0.9466,
+      "step": 12988
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 3.602178017197512e-05,
+      "loss": 0.8004,
+      "step": 12989
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 3.6008368799148926e-05,
+      "loss": 0.9588,
+      "step": 12990
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 3.599495937518622e-05,
+      "loss": 0.9543,
+      "step": 12991
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 3.598155190049539e-05,
+      "loss": 0.8931,
+      "step": 12992
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 3.5968146375484635e-05,
+      "loss": 0.8609,
+      "step": 12993
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 3.5954742800562346e-05,
+      "loss": 0.8483,
+      "step": 12994
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 3.5941341176136645e-05,
+      "loss": 0.8789,
+      "step": 12995
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 3.592794150261575e-05,
+      "loss": 0.8815,
+      "step": 12996
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 3.5914543780407715e-05,
+      "loss": 0.8299,
+      "step": 12997
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 3.590114800992057e-05,
+      "loss": 0.844,
+      "step": 12998
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 3.5887754191562274e-05,
+      "loss": 0.9515,
+      "step": 12999
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 3.5874362325740715e-05,
+      "loss": 0.7861,
+      "step": 13000
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 3.586097241286379e-05,
+      "loss": 0.8275,
+      "step": 13001
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 3.584758445333928e-05,
+      "loss": 0.7989,
+      "step": 13002
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 3.58341984475749e-05,
+      "loss": 0.8399,
+      "step": 13003
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 3.582081439597832e-05,
+      "loss": 0.7329,
+      "step": 13004
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 3.580743229895712e-05,
+      "loss": 0.8976,
+      "step": 13005
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 3.579405215691891e-05,
+      "loss": 0.8928,
+      "step": 13006
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 3.5780673970271126e-05,
+      "loss": 0.9726,
+      "step": 13007
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 3.576729773942131e-05,
+      "loss": 0.8003,
+      "step": 13008
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 3.5753923464776715e-05,
+      "loss": 0.8696,
+      "step": 13009
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 3.5740551146744664e-05,
+      "loss": 0.9561,
+      "step": 13010
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 3.572718078573246e-05,
+      "loss": 0.7909,
+      "step": 13011
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 3.571381238214726e-05,
+      "loss": 0.8701,
+      "step": 13012
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 3.570044593639624e-05,
+      "loss": 0.8724,
+      "step": 13013
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 3.568708144888646e-05,
+      "loss": 0.8933,
+      "step": 13014
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 3.567371892002492e-05,
+      "loss": 0.8464,
+      "step": 13015
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 3.566035835021859e-05,
+      "loss": 0.8822,
+      "step": 13016
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 3.564699973987432e-05,
+      "loss": 0.8839,
+      "step": 13017
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 3.5633643089399016e-05,
+      "loss": 0.8899,
+      "step": 13018
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 3.562028839919942e-05,
+      "loss": 0.7908,
+      "step": 13019
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 3.560693566968225e-05,
+      "loss": 0.8513,
+      "step": 13020
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 3.5593584901254176e-05,
+      "loss": 0.84,
+      "step": 13021
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 3.558023609432174e-05,
+      "loss": 0.8359,
+      "step": 13022
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 3.556688924929156e-05,
+      "loss": 0.97,
+      "step": 13023
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 3.555354436657007e-05,
+      "loss": 0.9314,
+      "step": 13024
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 3.55402014465637e-05,
+      "loss": 1.0307,
+      "step": 13025
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 3.5526860489678804e-05,
+      "loss": 0.8133,
+      "step": 13026
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 3.551352149632165e-05,
+      "loss": 0.8782,
+      "step": 13027
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 3.550018446689853e-05,
+      "loss": 0.8425,
+      "step": 13028
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 3.5486849401815606e-05,
+      "loss": 0.9246,
+      "step": 13029
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 3.5473516301478984e-05,
+      "loss": 0.7501,
+      "step": 13030
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 3.54601851662947e-05,
+      "loss": 0.7576,
+      "step": 13031
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 3.544685599666882e-05,
+      "loss": 0.8322,
+      "step": 13032
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 3.543352879300724e-05,
+      "loss": 0.8267,
+      "step": 13033
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 3.542020355571581e-05,
+      "loss": 0.7967,
+      "step": 13034
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 3.540688028520046e-05,
+      "loss": 0.8429,
+      "step": 13035
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 3.5393558981866814e-05,
+      "loss": 0.9027,
+      "step": 13036
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 3.538023964612066e-05,
+      "loss": 0.8287,
+      "step": 13037
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 3.536692227836761e-05,
+      "loss": 0.9223,
+      "step": 13038
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 3.5353606879013214e-05,
+      "loss": 0.8888,
+      "step": 13039
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 3.5340293448463056e-05,
+      "loss": 0.8604,
+      "step": 13040
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 3.5326981987122565e-05,
+      "loss": 0.9297,
+      "step": 13041
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 3.531367249539713e-05,
+      "loss": 0.9081,
+      "step": 13042
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 3.5300364973692115e-05,
+      "loss": 0.8528,
+      "step": 13043
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 3.5287059422412747e-05,
+      "loss": 0.8131,
+      "step": 13044
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 3.527375584196433e-05,
+      "loss": 0.7986,
+      "step": 13045
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 3.526045423275196e-05,
+      "loss": 0.9022,
+      "step": 13046
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 3.5247154595180776e-05,
+      "loss": 0.7171,
+      "step": 13047
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 3.5233856929655754e-05,
+      "loss": 0.8292,
+      "step": 13048
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 3.522056123658196e-05,
+      "loss": 0.8327,
+      "step": 13049
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 3.5207267516364275e-05,
+      "loss": 0.7674,
+      "step": 13050
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 3.519397576940755e-05,
+      "loss": 0.8968,
+      "step": 13051
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 3.518068599611659e-05,
+      "loss": 0.8357,
+      "step": 13052
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 3.516739819689611e-05,
+      "loss": 0.7982,
+      "step": 13053
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 3.5154112372150846e-05,
+      "loss": 0.9165,
+      "step": 13054
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 3.5140828522285385e-05,
+      "loss": 0.7883,
+      "step": 13055
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 3.5127546647704266e-05,
+      "loss": 0.7119,
+      "step": 13056
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 3.5114266748812075e-05,
+      "loss": 0.8235,
+      "step": 13057
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 3.5100988826013115e-05,
+      "loss": 0.8645,
+      "step": 13058
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 3.5087712879711864e-05,
+      "loss": 0.8686,
+      "step": 13059
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 3.507443891031259e-05,
+      "loss": 0.791,
+      "step": 13060
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 3.506116691821961e-05,
+      "loss": 0.9236,
+      "step": 13061
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 3.504789690383709e-05,
+      "loss": 0.9445,
+      "step": 13062
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 3.5034628867569154e-05,
+      "loss": 0.8466,
+      "step": 13063
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 3.502136280981989e-05,
+      "loss": 0.8803,
+      "step": 13064
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 3.500809873099329e-05,
+      "loss": 0.879,
+      "step": 13065
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 3.499483663149338e-05,
+      "loss": 0.9183,
+      "step": 13066
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 3.4981576511724e-05,
+      "loss": 0.8886,
+      "step": 13067
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 3.4968318372089e-05,
+      "loss": 0.8413,
+      "step": 13068
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 3.4955062212992163e-05,
+      "loss": 0.8682,
+      "step": 13069
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 3.4941808034837164e-05,
+      "loss": 0.8882,
+      "step": 13070
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 3.492855583802772e-05,
+      "loss": 0.8691,
+      "step": 13071
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 3.491530562296741e-05,
+      "loss": 0.9601,
+      "step": 13072
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 3.4902057390059765e-05,
+      "loss": 0.8455,
+      "step": 13073
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 3.4888811139708246e-05,
+      "loss": 0.9235,
+      "step": 13074
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 3.487556687231623e-05,
+      "loss": 0.8145,
+      "step": 13075
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 3.486232458828716e-05,
+      "loss": 0.8452,
+      "step": 13076
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 3.484908428802426e-05,
+      "loss": 0.8621,
+      "step": 13077
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 3.4835845971930834e-05,
+      "loss": 0.9306,
+      "step": 13078
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 3.4822609640409985e-05,
+      "loss": 0.9387,
+      "step": 13079
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 3.4809375293864824e-05,
+      "loss": 0.8066,
+      "step": 13080
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 3.479614293269845e-05,
+      "loss": 0.8545,
+      "step": 13081
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 3.4782912557313806e-05,
+      "loss": 0.8589,
+      "step": 13082
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 3.476968416811388e-05,
+      "loss": 0.8612,
+      "step": 13083
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 3.475645776550154e-05,
+      "loss": 0.858,
+      "step": 13084
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 3.47432333498795e-05,
+      "loss": 0.7697,
+      "step": 13085
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 3.4730010921650616e-05,
+      "loss": 0.7633,
+      "step": 13086
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 3.4716790481217495e-05,
+      "loss": 0.8869,
+      "step": 13087
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 3.4703572028982846e-05,
+      "loss": 0.9494,
+      "step": 13088
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 3.4690355565349186e-05,
+      "loss": 0.7688,
+      "step": 13089
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 3.4677141090719044e-05,
+      "loss": 0.7771,
+      "step": 13090
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 3.466392860549484e-05,
+      "loss": 0.6831,
+      "step": 13091
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 3.4650718110078965e-05,
+      "loss": 0.8768,
+      "step": 13092
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 3.4637509604873776e-05,
+      "loss": 0.823,
+      "step": 13093
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 3.4624303090281516e-05,
+      "loss": 0.6897,
+      "step": 13094
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 3.461109856670439e-05,
+      "loss": 0.8847,
+      "step": 13095
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 3.459789603454454e-05,
+      "loss": 0.8777,
+      "step": 13096
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 3.458469549420402e-05,
+      "loss": 0.9157,
+      "step": 13097
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 3.4571496946084914e-05,
+      "loss": 0.8839,
+      "step": 13098
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 3.4558300390589116e-05,
+      "loss": 0.9267,
+      "step": 13099
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 3.454510582811863e-05,
+      "loss": 0.8371,
+      "step": 13100
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 3.45319132590752e-05,
+      "loss": 1.062,
+      "step": 13101
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 3.4518722683860594e-05,
+      "loss": 0.8827,
+      "step": 13102
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 3.45055341028766e-05,
+      "loss": 0.7858,
+      "step": 13103
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 3.4492347516524815e-05,
+      "loss": 0.8679,
+      "step": 13104
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 3.447916292520691e-05,
+      "loss": 0.8928,
+      "step": 13105
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 3.446598032932437e-05,
+      "loss": 0.8117,
+      "step": 13106
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 3.445279972927867e-05,
+      "loss": 0.7571,
+      "step": 13107
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 3.443962112547125e-05,
+      "loss": 0.8436,
+      "step": 13108
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 3.4426444518303404e-05,
+      "loss": 0.9101,
+      "step": 13109
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 3.44132699081765e-05,
+      "loss": 0.7718,
+      "step": 13110
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 3.440009729549175e-05,
+      "loss": 0.8216,
+      "step": 13111
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 3.438692668065031e-05,
+      "loss": 0.8268,
+      "step": 13112
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 3.4373758064053293e-05,
+      "loss": 0.9526,
+      "step": 13113
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 3.436059144610173e-05,
+      "loss": 0.9262,
+      "step": 13114
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 3.434742682719666e-05,
+      "loss": 0.9358,
+      "step": 13115
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 3.433426420773897e-05,
+      "loss": 0.8008,
+      "step": 13116
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 3.4321103588129555e-05,
+      "loss": 0.8262,
+      "step": 13117
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 3.4307944968769204e-05,
+      "loss": 0.8558,
+      "step": 13118
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 3.429478835005863e-05,
+      "loss": 0.9185,
+      "step": 13119
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 3.428163373239859e-05,
+      "loss": 0.8574,
+      "step": 13120
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 3.426848111618966e-05,
+      "loss": 0.8412,
+      "step": 13121
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 3.4255330501832414e-05,
+      "loss": 0.8249,
+      "step": 13122
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 3.424218188972732e-05,
+      "loss": 0.8805,
+      "step": 13123
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 3.422903528027489e-05,
+      "loss": 0.9666,
+      "step": 13124
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 3.421589067387545e-05,
+      "loss": 0.8537,
+      "step": 13125
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 3.4202748070929314e-05,
+      "loss": 0.795,
+      "step": 13126
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 3.4189607471836825e-05,
+      "loss": 0.8949,
+      "step": 13127
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 3.417646887699804e-05,
+      "loss": 0.7485,
+      "step": 13128
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 3.4163332286813196e-05,
+      "loss": 0.899,
+      "step": 13129
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 3.415019770168235e-05,
+      "loss": 0.7364,
+      "step": 13130
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 3.413706512200546e-05,
+      "loss": 0.8395,
+      "step": 13131
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 3.4123934548182566e-05,
+      "loss": 0.8715,
+      "step": 13132
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 3.4110805980613525e-05,
+      "loss": 0.9241,
+      "step": 13133
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 3.409767941969815e-05,
+      "loss": 0.7795,
+      "step": 13134
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 3.408455486583621e-05,
+      "loss": 0.9595,
+      "step": 13135
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 3.407143231942739e-05,
+      "loss": 0.815,
+      "step": 13136
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 3.405831178087141e-05,
+      "loss": 0.8393,
+      "step": 13137
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 3.404519325056781e-05,
+      "loss": 0.8103,
+      "step": 13138
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 3.403207672891612e-05,
+      "loss": 0.81,
+      "step": 13139
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 3.4018962216315765e-05,
+      "loss": 0.846,
+      "step": 13140
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 3.400584971316621e-05,
+      "loss": 0.7654,
+      "step": 13141
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 3.3992739219866786e-05,
+      "loss": 0.8353,
+      "step": 13142
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 3.397963073681675e-05,
+      "loss": 0.8887,
+      "step": 13143
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 3.396652426441532e-05,
+      "loss": 0.88,
+      "step": 13144
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 3.3953419803061635e-05,
+      "loss": 0.7518,
+      "step": 13145
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 3.394031735315485e-05,
+      "loss": 0.7919,
+      "step": 13146
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 3.3927216915093976e-05,
+      "loss": 0.8657,
+      "step": 13147
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 3.391411848927793e-05,
+      "loss": 0.917,
+      "step": 13148
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 3.390102207610574e-05,
+      "loss": 0.9643,
+      "step": 13149
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 3.388792767597612e-05,
+      "loss": 0.8759,
+      "step": 13150
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 3.3874835289287954e-05,
+      "loss": 0.9261,
+      "step": 13151
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 3.386174491643992e-05,
+      "loss": 0.889,
+      "step": 13152
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 3.384865655783074e-05,
+      "loss": 0.8672,
+      "step": 13153
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 3.383557021385899e-05,
+      "loss": 0.8614,
+      "step": 13154
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 3.382248588492321e-05,
+      "loss": 0.8304,
+      "step": 13155
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 3.380940357142188e-05,
+      "loss": 0.8546,
+      "step": 13156
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 3.379632327375341e-05,
+      "loss": 0.8832,
+      "step": 13157
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 3.3783244992316187e-05,
+      "loss": 0.8235,
+      "step": 13158
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 3.3770168727508524e-05,
+      "loss": 0.8068,
+      "step": 13159
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 3.3757094479728615e-05,
+      "loss": 0.8978,
+      "step": 13160
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 3.3744022249374666e-05,
+      "loss": 0.8188,
+      "step": 13161
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 3.373095203684474e-05,
+      "loss": 0.8641,
+      "step": 13162
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 3.371788384253697e-05,
+      "loss": 0.8403,
+      "step": 13163
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 3.370481766684931e-05,
+      "loss": 0.903,
+      "step": 13164
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 3.369175351017968e-05,
+      "loss": 0.8221,
+      "step": 13165
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 3.367869137292596e-05,
+      "loss": 0.774,
+      "step": 13166
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 3.3665631255485916e-05,
+      "loss": 0.8873,
+      "step": 13167
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 3.365257315825737e-05,
+      "loss": 0.8758,
+      "step": 13168
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 3.363951708163794e-05,
+      "loss": 0.7819,
+      "step": 13169
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 3.3626463026025304e-05,
+      "loss": 0.8941,
+      "step": 13170
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 3.361341099181702e-05,
+      "loss": 0.9066,
+      "step": 13171
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 3.36003609794105e-05,
+      "loss": 0.8998,
+      "step": 13172
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 3.3587312989203276e-05,
+      "loss": 0.8647,
+      "step": 13173
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 3.3574267021592657e-05,
+      "loss": 0.9241,
+      "step": 13174
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 3.356122307697603e-05,
+      "loss": 0.8596,
+      "step": 13175
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 3.35481811557506e-05,
+      "loss": 0.8395,
+      "step": 13176
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 3.353514125831358e-05,
+      "loss": 0.7811,
+      "step": 13177
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 3.352210338506209e-05,
+      "loss": 0.9296,
+      "step": 13178
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 3.350906753639316e-05,
+      "loss": 0.7813,
+      "step": 13179
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 3.3496033712703865e-05,
+      "loss": 0.7827,
+      "step": 13180
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 3.348300191439112e-05,
+      "loss": 0.9461,
+      "step": 13181
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 3.34699721418518e-05,
+      "loss": 0.9047,
+      "step": 13182
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 3.3456944395482734e-05,
+      "loss": 0.7312,
+      "step": 13183
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 3.344391867568065e-05,
+      "loss": 0.8543,
+      "step": 13184
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 3.343089498284231e-05,
+      "loss": 1.0098,
+      "step": 13185
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 3.341787331736431e-05,
+      "loss": 1.038,
+      "step": 13186
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 3.3404853679643246e-05,
+      "loss": 0.9605,
+      "step": 13187
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 3.33918360700756e-05,
+      "loss": 0.8879,
+      "step": 13188
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 3.3378820489057804e-05,
+      "loss": 0.8649,
+      "step": 13189
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 3.336580693698632e-05,
+      "loss": 0.8579,
+      "step": 13190
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 3.335279541425741e-05,
+      "loss": 0.6761,
+      "step": 13191
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 3.333978592126743e-05,
+      "loss": 0.8829,
+      "step": 13192
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 3.332677845841248e-05,
+      "loss": 0.8365,
+      "step": 13193
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 3.3313773026088715e-05,
+      "loss": 0.823,
+      "step": 13194
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 3.330076962469226e-05,
+      "loss": 0.7989,
+      "step": 13195
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 3.32877682546191e-05,
+      "loss": 0.8927,
+      "step": 13196
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 3.327476891626524e-05,
+      "loss": 0.7845,
+      "step": 13197
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 3.3261771610026537e-05,
+      "loss": 0.8819,
+      "step": 13198
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 3.3248776336298826e-05,
+      "loss": 0.8461,
+      "step": 13199
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 3.323578309547788e-05,
+      "loss": 0.9021,
+      "step": 13200
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 3.322279188795937e-05,
+      "loss": 0.8617,
+      "step": 13201
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 3.320980271413903e-05,
+      "loss": 1.0268,
+      "step": 13202
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 3.319681557441238e-05,
+      "loss": 0.7239,
+      "step": 13203
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 3.318383046917496e-05,
+      "loss": 0.8661,
+      "step": 13204
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 3.3170847398822224e-05,
+      "loss": 0.9722,
+      "step": 13205
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 3.3157866363749555e-05,
+      "loss": 0.9334,
+      "step": 13206
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 3.314488736435234e-05,
+      "loss": 0.8048,
+      "step": 13207
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 3.313191040102583e-05,
+      "loss": 0.7907,
+      "step": 13208
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 3.3118935474165223e-05,
+      "loss": 0.8528,
+      "step": 13209
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 3.31059625841657e-05,
+      "loss": 0.9533,
+      "step": 13210
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 3.309299173142227e-05,
+      "loss": 0.7673,
+      "step": 13211
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 3.308002291633008e-05,
+      "loss": 0.8973,
+      "step": 13212
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 3.306705613928399e-05,
+      "loss": 0.8926,
+      "step": 13213
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 3.305409140067901e-05,
+      "loss": 0.9402,
+      "step": 13214
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 3.3041128700909865e-05,
+      "loss": 0.8562,
+      "step": 13215
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 3.302816804037142e-05,
+      "loss": 0.9157,
+      "step": 13216
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 3.3015209419458345e-05,
+      "loss": 0.8306,
+      "step": 13217
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 3.30022528385653e-05,
+      "loss": 0.8782,
+      "step": 13218
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 3.29892982980869e-05,
+      "loss": 0.9769,
+      "step": 13219
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 3.297634579841766e-05,
+      "loss": 0.9423,
+      "step": 13220
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 3.296339533995206e-05,
+      "loss": 0.8306,
+      "step": 13221
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 3.2950446923084486e-05,
+      "loss": 0.8217,
+      "step": 13222
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 3.2937500548209265e-05,
+      "loss": 0.8285,
+      "step": 13223
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 3.2924556215720744e-05,
+      "loss": 0.9415,
+      "step": 13224
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 3.29116139260131e-05,
+      "loss": 0.9092,
+      "step": 13225
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 3.289867367948048e-05,
+      "loss": 0.8317,
+      "step": 13226
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 3.2885735476517e-05,
+      "loss": 0.7745,
+      "step": 13227
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 3.2872799317516655e-05,
+      "loss": 0.8356,
+      "step": 13228
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 3.2859865202873476e-05,
+      "loss": 0.8632,
+      "step": 13229
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 3.2846933132981336e-05,
+      "loss": 0.931,
+      "step": 13230
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 3.283400310823409e-05,
+      "loss": 1.0105,
+      "step": 13231
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 3.282107512902548e-05,
+      "loss": 0.7803,
+      "step": 13232
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 3.28081491957493e-05,
+      "loss": 0.8512,
+      "step": 13233
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 3.279522530879917e-05,
+      "loss": 0.9596,
+      "step": 13234
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 3.27823034685687e-05,
+      "loss": 0.8293,
+      "step": 13235
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 3.276938367545139e-05,
+      "loss": 0.7368,
+      "step": 13236
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 3.275646592984072e-05,
+      "loss": 0.8078,
+      "step": 13237
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 3.274355023213015e-05,
+      "loss": 0.7844,
+      "step": 13238
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 3.273063658271298e-05,
+      "loss": 0.7567,
+      "step": 13239
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 3.2717724981982486e-05,
+      "loss": 0.7521,
+      "step": 13240
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 3.270481543033197e-05,
+      "loss": 0.9977,
+      "step": 13241
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 3.2691907928154475e-05,
+      "loss": 0.8629,
+      "step": 13242
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 3.267900247584319e-05,
+      "loss": 0.9281,
+      "step": 13243
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 3.266609907379108e-05,
+      "loss": 0.7441,
+      "step": 13244
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 3.265319772239118e-05,
+      "loss": 0.8071,
+      "step": 13245
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 3.2640298422036396e-05,
+      "loss": 0.8516,
+      "step": 13246
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 3.262740117311954e-05,
+      "loss": 0.9944,
+      "step": 13247
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 3.2614505976033413e-05,
+      "loss": 0.8218,
+      "step": 13248
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 3.260161283117071e-05,
+      "loss": 1.0057,
+      "step": 13249
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 3.258872173892415e-05,
+      "loss": 0.867,
+      "step": 13250
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 3.25758326996863e-05,
+      "loss": 0.7734,
+      "step": 13251
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 3.2562945713849704e-05,
+      "loss": 0.7021,
+      "step": 13252
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 3.2550060781806825e-05,
+      "loss": 0.8354,
+      "step": 13253
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 3.253717790395003e-05,
+      "loss": 0.8417,
+      "step": 13254
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 3.252429708067175e-05,
+      "loss": 0.8554,
+      "step": 13255
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 3.251141831236424e-05,
+      "loss": 0.8183,
+      "step": 13256
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 3.249854159941971e-05,
+      "loss": 0.9476,
+      "step": 13257
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 3.248566694223031e-05,
+      "loss": 0.7619,
+      "step": 13258
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 3.2472794341188126e-05,
+      "loss": 0.9283,
+      "step": 13259
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 3.2459923796685255e-05,
+      "loss": 0.8832,
+      "step": 13260
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 3.244705530911361e-05,
+      "loss": 0.9109,
+      "step": 13261
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 3.243418887886513e-05,
+      "loss": 0.9387,
+      "step": 13262
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 3.2421324506331705e-05,
+      "loss": 0.7479,
+      "step": 13263
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 3.2408462191905e-05,
+      "loss": 0.9243,
+      "step": 13264
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 3.239560193597683e-05,
+      "loss": 0.8637,
+      "step": 13265
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 3.2382743738938806e-05,
+      "loss": 0.904,
+      "step": 13266
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 3.236988760118257e-05,
+      "loss": 0.9384,
+      "step": 13267
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 3.2357033523099636e-05,
+      "loss": 0.7319,
+      "step": 13268
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 3.234418150508147e-05,
+      "loss": 0.8049,
+      "step": 13269
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 3.2331331547519486e-05,
+      "loss": 0.8255,
+      "step": 13270
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 3.2318483650804984e-05,
+      "loss": 0.8493,
+      "step": 13271
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.230563781532932e-05,
+      "loss": 0.9431,
+      "step": 13272
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.229279404148369e-05,
+      "loss": 0.8132,
+      "step": 13273
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.2279952329659224e-05,
+      "loss": 0.8165,
+      "step": 13274
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.2267112680247045e-05,
+      "loss": 0.8734,
+      "step": 13275
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.225427509363813e-05,
+      "loss": 0.8281,
+      "step": 13276
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.224143957022353e-05,
+      "loss": 0.8361,
+      "step": 13277
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.22286061103941e-05,
+      "loss": 0.8879,
+      "step": 13278
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.2215774714540704e-05,
+      "loss": 0.8659,
+      "step": 13279
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.2202945383054095e-05,
+      "loss": 0.9303,
+      "step": 13280
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.2190118116324984e-05,
+      "loss": 0.9924,
+      "step": 13281
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.2177292914744076e-05,
+      "loss": 0.8041,
+      "step": 13282
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.21644697787019e-05,
+      "loss": 0.8931,
+      "step": 13283
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.2151648708589075e-05,
+      "loss": 0.9067,
+      "step": 13284
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.213882970479598e-05,
+      "loss": 0.7456,
+      "step": 13285
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.212601276771301e-05,
+      "loss": 0.9533,
+      "step": 13286
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.2113197897730574e-05,
+      "loss": 0.7262,
+      "step": 13287
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.210038509523888e-05,
+      "loss": 0.811,
+      "step": 13288
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.208757436062822e-05,
+      "loss": 0.8803,
+      "step": 13289
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.2074765694288686e-05,
+      "loss": 0.8651,
+      "step": 13290
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.2061959096610396e-05,
+      "loss": 0.9197,
+      "step": 13291
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.204915456798334e-05,
+      "loss": 0.8253,
+      "step": 13292
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.203635210879747e-05,
+      "loss": 0.9016,
+      "step": 13293
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.202355171944276e-05,
+      "loss": 0.911,
+      "step": 13294
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.201075340030898e-05,
+      "loss": 0.9056,
+      "step": 13295
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.199795715178593e-05,
+      "loss": 0.9733,
+      "step": 13296
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.19851629742633e-05,
+      "loss": 0.8148,
+      "step": 13297
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.197237086813073e-05,
+      "loss": 0.8266,
+      "step": 13298
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.195958083377785e-05,
+      "loss": 0.8598,
+      "step": 13299
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.194679287159415e-05,
+      "loss": 0.7864,
+      "step": 13300
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.193400698196907e-05,
+      "loss": 0.77,
+      "step": 13301
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.1921223165292034e-05,
+      "loss": 0.8493,
+      "step": 13302
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.190844142195233e-05,
+      "loss": 0.7809,
+      "step": 13303
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.189566175233928e-05,
+      "loss": 0.7414,
+      "step": 13304
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.188288415684203e-05,
+      "loss": 0.8762,
+      "step": 13305
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.1870108635849826e-05,
+      "loss": 0.7429,
+      "step": 13306
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.18573351897516e-05,
+      "loss": 0.8673,
+      "step": 13307
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.184456381893649e-05,
+      "loss": 0.7716,
+      "step": 13308
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.183179452379339e-05,
+      "loss": 0.8665,
+      "step": 13309
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.181902730471118e-05,
+      "loss": 0.9073,
+      "step": 13310
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.1806262162078714e-05,
+      "loss": 0.7721,
+      "step": 13311
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.179349909628476e-05,
+      "loss": 0.7966,
+      "step": 13312
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.1780738107718e-05,
+      "loss": 0.7952,
+      "step": 13313
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.176797919676706e-05,
+      "loss": 0.7832,
+      "step": 13314
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.17552223638205e-05,
+      "loss": 1.0072,
+      "step": 13315
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.174246760926689e-05,
+      "loss": 0.8656,
+      "step": 13316
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.172971493349463e-05,
+      "loss": 0.8227,
+      "step": 13317
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.1716964336892104e-05,
+      "loss": 0.7856,
+      "step": 13318
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.1704215819847636e-05,
+      "loss": 0.9223,
+      "step": 13319
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.1691469382749464e-05,
+      "loss": 0.7581,
+      "step": 13320
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.167872502598582e-05,
+      "loss": 0.7313,
+      "step": 13321
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.166598274994482e-05,
+      "loss": 0.7816,
+      "step": 13322
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.1653242555014516e-05,
+      "loss": 0.7585,
+      "step": 13323
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.164050444158289e-05,
+      "loss": 0.8991,
+      "step": 13324
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.162776841003793e-05,
+      "loss": 0.7358,
+      "step": 13325
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.161503446076749e-05,
+      "loss": 0.8702,
+      "step": 13326
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.160230259415935e-05,
+      "loss": 0.8191,
+      "step": 13327
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.1589572810601366e-05,
+      "loss": 0.8626,
+      "step": 13328
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.157684511048107e-05,
+      "loss": 0.9091,
+      "step": 13329
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.15641194941862e-05,
+      "loss": 0.8435,
+      "step": 13330
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.155139596210426e-05,
+      "loss": 0.7913,
+      "step": 13331
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.153867451462273e-05,
+      "loss": 0.7465,
+      "step": 13332
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.152595515212915e-05,
+      "loss": 0.8585,
+      "step": 13333
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.1513237875010726e-05,
+      "loss": 0.8379,
+      "step": 13334
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.150052268365489e-05,
+      "loss": 0.9503,
+      "step": 13335
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.14878095784488e-05,
+      "loss": 0.8224,
+      "step": 13336
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.14750985597797e-05,
+      "loss": 0.911,
+      "step": 13337
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.146238962803468e-05,
+      "loss": 0.7261,
+      "step": 13338
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.14496827836008e-05,
+      "loss": 0.8915,
+      "step": 13339
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.1436978026865e-05,
+      "loss": 0.7746,
+      "step": 13340
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.1424275358214226e-05,
+      "loss": 0.8456,
+      "step": 13341
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.141157477803538e-05,
+      "loss": 0.8985,
+      "step": 13342
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.139887628671521e-05,
+      "loss": 0.7948,
+      "step": 13343
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.138617988464047e-05,
+      "loss": 0.8051,
+      "step": 13344
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.137348557219781e-05,
+      "loss": 0.8497,
+      "step": 13345
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.136079334977383e-05,
+      "loss": 0.8712,
+      "step": 13346
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.134810321775511e-05,
+      "loss": 0.9078,
+      "step": 13347
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.13354151765281e-05,
+      "loss": 0.8444,
+      "step": 13348
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.132272922647923e-05,
+      "loss": 0.9096,
+      "step": 13349
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.131004536799483e-05,
+      "loss": 0.9199,
+      "step": 13350
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.129736360146117e-05,
+      "loss": 0.8477,
+      "step": 13351
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.128468392726453e-05,
+      "loss": 0.7469,
+      "step": 13352
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.1272006345790993e-05,
+      "loss": 0.8752,
+      "step": 13353
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.1259330857426736e-05,
+      "loss": 0.8507,
+      "step": 13354
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.124665746255779e-05,
+      "loss": 0.8251,
+      "step": 13355
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.1233986161570025e-05,
+      "loss": 0.8563,
+      "step": 13356
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.122131695484944e-05,
+      "loss": 0.9358,
+      "step": 13357
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.12086498427818e-05,
+      "loss": 0.9521,
+      "step": 13358
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.1195984825752966e-05,
+      "loss": 0.8817,
+      "step": 13359
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.1183321904148597e-05,
+      "loss": 0.8803,
+      "step": 13360
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.1170661078354366e-05,
+      "loss": 0.8291,
+      "step": 13361
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 3.115800234875584e-05,
+      "loss": 0.992,
+      "step": 13362
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 3.1145345715738515e-05,
+      "loss": 0.8493,
+      "step": 13363
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 3.11326911796879e-05,
+      "loss": 0.9273,
+      "step": 13364
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 3.112003874098938e-05,
+      "loss": 0.8431,
+      "step": 13365
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 3.110738840002827e-05,
+      "loss": 0.8845,
+      "step": 13366
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 3.109474015718984e-05,
+      "loss": 0.7825,
+      "step": 13367
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 3.108209401285925e-05,
+      "loss": 0.7719,
+      "step": 13368
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 3.106944996742172e-05,
+      "loss": 0.8197,
+      "step": 13369
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 3.105680802126224e-05,
+      "loss": 0.8083,
+      "step": 13370
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 3.104416817476593e-05,
+      "loss": 0.8864,
+      "step": 13371
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 3.103153042831764e-05,
+      "loss": 0.9075,
+      "step": 13372
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 3.101889478230224e-05,
+      "loss": 0.8721,
+      "step": 13373
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 3.100626123710463e-05,
+      "loss": 0.8844,
+      "step": 13374
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 3.099362979310949e-05,
+      "loss": 1.0032,
+      "step": 13375
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 3.098100045070157e-05,
+      "loss": 1.0016,
+      "step": 13376
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 3.09683732102655e-05,
+      "loss": 0.9313,
+      "step": 13377
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 3.095574807218575e-05,
+      "loss": 0.8711,
+      "step": 13378
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 3.094312503684692e-05,
+      "loss": 0.9666,
+      "step": 13379
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 3.093050410463337e-05,
+      "loss": 0.7768,
+      "step": 13380
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 3.091788527592954e-05,
+      "loss": 0.9163,
+      "step": 13381
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 3.090526855111969e-05,
+      "loss": 0.7861,
+      "step": 13382
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 3.089265393058808e-05,
+      "loss": 0.8456,
+      "step": 13383
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 3.088004141471887e-05,
+      "loss": 0.7045,
+      "step": 13384
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 3.086743100389616e-05,
+      "loss": 0.783,
+      "step": 13385
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 3.085482269850406e-05,
+      "loss": 0.8057,
+      "step": 13386
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 3.084221649892652e-05,
+      "loss": 0.8178,
+      "step": 13387
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 3.0829612405547456e-05,
+      "loss": 0.9438,
+      "step": 13388
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 3.0817010418750735e-05,
+      "loss": 0.8373,
+      "step": 13389
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 3.080441053892011e-05,
+      "loss": 0.9108,
+      "step": 13390
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 3.079181276643939e-05,
+      "loss": 0.9455,
+      "step": 13391
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 3.077921710169218e-05,
+      "loss": 0.9134,
+      "step": 13392
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 3.0766623545062104e-05,
+      "loss": 0.834,
+      "step": 13393
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 3.0754032096932694e-05,
+      "loss": 0.8343,
+      "step": 13394
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 3.0741442757687375e-05,
+      "loss": 0.9366,
+      "step": 13395
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 3.0728855527709644e-05,
+      "loss": 0.906,
+      "step": 13396
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 3.071627040738277e-05,
+      "loss": 0.8381,
+      "step": 13397
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 3.070368739709011e-05,
+      "loss": 0.7808,
+      "step": 13398
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 3.0691106497214775e-05,
+      "loss": 0.9157,
+      "step": 13399
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 3.067852770814e-05,
+      "loss": 0.9169,
+      "step": 13400
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 3.0665951030248854e-05,
+      "loss": 0.7581,
+      "step": 13401
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 3.065337646392431e-05,
+      "loss": 0.9328,
+      "step": 13402
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 3.0640804009549406e-05,
+      "loss": 0.9185,
+      "step": 13403
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 3.062823366750699e-05,
+      "loss": 0.834,
+      "step": 13404
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 3.06156654381799e-05,
+      "loss": 0.7731,
+      "step": 13405
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 3.060309932195089e-05,
+      "loss": 0.9883,
+      "step": 13406
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 3.059053531920264e-05,
+      "loss": 0.7883,
+      "step": 13407
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 3.0577973430317854e-05,
+      "loss": 0.9381,
+      "step": 13408
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 3.0565413655679056e-05,
+      "loss": 0.8846,
+      "step": 13409
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 3.055285599566876e-05,
+      "loss": 0.8508,
+      "step": 13410
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 3.0540300450669426e-05,
+      "loss": 0.8355,
+      "step": 13411
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 3.052774702106338e-05,
+      "loss": 0.9784,
+      "step": 13412
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 3.0515195707233e-05,
+      "loss": 0.7655,
+      "step": 13413
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 3.0502646509560517e-05,
+      "loss": 0.9121,
+      "step": 13414
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 3.0490099428428108e-05,
+      "loss": 0.9221,
+      "step": 13415
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 3.0477554464217872e-05,
+      "loss": 0.8871,
+      "step": 13416
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 3.0465011617311924e-05,
+      "loss": 1.0464,
+      "step": 13417
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 3.045247088809221e-05,
+      "loss": 0.8126,
+      "step": 13418
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 3.043993227694064e-05,
+      "loss": 0.8409,
+      "step": 13419
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 3.0427395784239175e-05,
+      "loss": 0.9994,
+      "step": 13420
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 3.0414861410369478e-05,
+      "loss": 0.9381,
+      "step": 13421
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 3.0402329155713395e-05,
+      "loss": 0.8903,
+      "step": 13422
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 3.0389799020652532e-05,
+      "loss": 0.8598,
+      "step": 13423
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 3.0377271005568485e-05,
+      "loss": 0.9178,
+      "step": 13424
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 3.0364745110842864e-05,
+      "loss": 0.7808,
+      "step": 13425
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 3.03522213368571e-05,
+      "loss": 0.7718,
+      "step": 13426
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 3.0339699683992607e-05,
+      "loss": 0.8693,
+      "step": 13427
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 3.0327180152630697e-05,
+      "loss": 0.8045,
+      "step": 13428
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 3.031466274315271e-05,
+      "loss": 0.769,
+      "step": 13429
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 3.0302147455939856e-05,
+      "loss": 0.8802,
+      "step": 13430
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 3.0289634291373257e-05,
+      "loss": 0.9704,
+      "step": 13431
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 3.0277123249834026e-05,
+      "loss": 0.8434,
+      "step": 13432
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 3.0264614331703144e-05,
+      "loss": 0.859,
+      "step": 13433
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 3.025210753736164e-05,
+      "loss": 0.784,
+      "step": 13434
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 3.0239602867190363e-05,
+      "loss": 0.9123,
+      "step": 13435
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 3.0227100321570147e-05,
+      "loss": 0.8309,
+      "step": 13436
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 3.0214599900881757e-05,
+      "loss": 0.8712,
+      "step": 13437
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 3.0202101605505872e-05,
+      "loss": 0.8336,
+      "step": 13438
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 3.0189605435823188e-05,
+      "loss": 0.9599,
+      "step": 13439
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 3.0177111392214218e-05,
+      "loss": 0.7582,
+      "step": 13440
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 3.0164619475059507e-05,
+      "loss": 0.8837,
+      "step": 13441
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 3.015212968473947e-05,
+      "loss": 0.8628,
+      "step": 13442
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 3.013964202163445e-05,
+      "loss": 0.8398,
+      "step": 13443
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 3.012715648612484e-05,
+      "loss": 0.9445,
+      "step": 13444
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 3.0114673078590815e-05,
+      "loss": 0.8716,
+      "step": 13445
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 3.0102191799412616e-05,
+      "loss": 0.9813,
+      "step": 13446
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 3.008971264897036e-05,
+      "loss": 0.9085,
+      "step": 13447
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 3.0077235627644007e-05,
+      "loss": 0.8742,
+      "step": 13448
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 3.0064760735813646e-05,
+      "loss": 0.9325,
+      "step": 13449
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 3.0052287973859127e-05,
+      "loss": 0.8814,
+      "step": 13450
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 3.003981734216037e-05,
+      "loss": 0.8396,
+      "step": 13451
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 3.0027348841097148e-05,
+      "loss": 0.7656,
+      "step": 13452
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 3.0014882471049168e-05,
+      "loss": 0.7511,
+      "step": 13453
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 3.0002418232396112e-05,
+      "loss": 0.8508,
+      "step": 13454
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 2.998995612551754e-05,
+      "loss": 0.8854,
+      "step": 13455
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 2.997749615079305e-05,
+      "loss": 0.87,
+      "step": 13456
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 2.9965038308602068e-05,
+      "loss": 0.8219,
+      "step": 13457
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 2.995258259932402e-05,
+      "loss": 0.7898,
+      "step": 13458
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 2.9940129023338215e-05,
+      "loss": 0.7443,
+      "step": 13459
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 2.9927677581023926e-05,
+      "loss": 0.8686,
+      "step": 13460
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 2.99152282727604e-05,
+      "loss": 0.8474,
+      "step": 13461
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 2.9902781098926725e-05,
+      "loss": 0.9176,
+      "step": 13462
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 2.989033605990208e-05,
+      "loss": 0.803,
+      "step": 13463
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 2.987789315606537e-05,
+      "loss": 0.8369,
+      "step": 13464
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 2.9865452387795555e-05,
+      "loss": 0.8932,
+      "step": 13465
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 2.985301375547158e-05,
+      "loss": 0.8027,
+      "step": 13466
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 2.9840577259472203e-05,
+      "loss": 0.8341,
+      "step": 13467
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 2.9828142900176236e-05,
+      "loss": 0.7153,
+      "step": 13468
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 2.981571067796236e-05,
+      "loss": 0.8338,
+      "step": 13469
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 2.9803280593209115e-05,
+      "loss": 0.9109,
+      "step": 13470
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 2.9790852646295143e-05,
+      "loss": 0.7094,
+      "step": 13471
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 2.9778426837598885e-05,
+      "loss": 0.9301,
+      "step": 13472
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 2.9766003167498825e-05,
+      "loss": 0.7323,
+      "step": 13473
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 2.9753581636373296e-05,
+      "loss": 0.9125,
+      "step": 13474
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 2.9741162244600586e-05,
+      "loss": 0.8939,
+      "step": 13475
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 2.9728744992558943e-05,
+      "loss": 0.8464,
+      "step": 13476
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 2.9716329880626492e-05,
+      "loss": 0.7753,
+      "step": 13477
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 2.9703916909181407e-05,
+      "loss": 0.8159,
+      "step": 13478
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 2.9691506078601684e-05,
+      "loss": 0.8892,
+      "step": 13479
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 2.9679097389265297e-05,
+      "loss": 0.808,
+      "step": 13480
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 2.9666690841550158e-05,
+      "loss": 0.76,
+      "step": 13481
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 2.9654286435834067e-05,
+      "loss": 0.947,
+      "step": 13482
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 2.9641884172494873e-05,
+      "loss": 0.7713,
+      "step": 13483
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 2.962948405191025e-05,
+      "loss": 0.7905,
+      "step": 13484
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 2.9617086074457835e-05,
+      "loss": 0.9152,
+      "step": 13485
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 2.960469024051522e-05,
+      "loss": 0.8878,
+      "step": 13486
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 2.9592296550459885e-05,
+      "loss": 0.816,
+      "step": 13487
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 2.957990500466934e-05,
+      "loss": 0.7387,
+      "step": 13488
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 2.956751560352091e-05,
+      "loss": 0.8917,
+      "step": 13489
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 2.9555128347392014e-05,
+      "loss": 0.7432,
+      "step": 13490
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 2.9542743236659764e-05,
+      "loss": 0.7868,
+      "step": 13491
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 2.9530360271701452e-05,
+      "loss": 0.9481,
+      "step": 13492
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 2.9517979452894172e-05,
+      "loss": 0.7788,
+      "step": 13493
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 2.950560078061494e-05,
+      "loss": 0.9577,
+      "step": 13494
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 2.9493224255240825e-05,
+      "loss": 0.8488,
+      "step": 13495
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 2.9480849877148707e-05,
+      "loss": 0.8152,
+      "step": 13496
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 2.9468477646715454e-05,
+      "loss": 0.7731,
+      "step": 13497
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 2.945610756431787e-05,
+      "loss": 0.8138,
+      "step": 13498
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 2.9443739630332647e-05,
+      "loss": 0.9241,
+      "step": 13499
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 2.9431373845136512e-05,
+      "loss": 0.8236,
+      "step": 13500
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 2.9419010209106036e-05,
+      "loss": 0.8227,
+      "step": 13501
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 2.9406648722617748e-05,
+      "loss": 0.864,
+      "step": 13502
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 2.9394289386048124e-05,
+      "loss": 0.817,
+      "step": 13503
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 2.9381932199773522e-05,
+      "loss": 0.8517,
+      "step": 13504
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 2.936957716417036e-05,
+      "loss": 0.922,
+      "step": 13505
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 2.935722427961487e-05,
+      "loss": 0.9774,
+      "step": 13506
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 2.9344873546483266e-05,
+      "loss": 0.8687,
+      "step": 13507
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 2.9332524965151654e-05,
+      "loss": 0.8088,
+      "step": 13508
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 2.9320178535996158e-05,
+      "loss": 0.8623,
+      "step": 13509
+    }
+  ],
+  "logging_steps": 1,
+  "max_steps": 18012,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 4,
+  "save_steps": 4503,
+  "total_flos": 2.814655510156935e+18,
+  "train_batch_size": 2,
+  "trial_name": null,
+  "trial_params": null
+}