diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,46902 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 0.999936,
+  "eval_steps": 500,
+  "global_step": 7812,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.255319148936171e-06,
+      "loss": 7.1797,
+      "step": 1
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.510638297872341e-06,
+      "loss": 7.0888,
+      "step": 2
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.2765957446808511e-05,
+      "loss": 7.1929,
+      "step": 3
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.7021276595744682e-05,
+      "loss": 7.1251,
+      "step": 4
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.1276595744680852e-05,
+      "loss": 7.2457,
+      "step": 5
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.5531914893617022e-05,
+      "loss": 6.9995,
+      "step": 6
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.9787234042553195e-05,
+      "loss": 6.5315,
+      "step": 7
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.4042553191489365e-05,
+      "loss": 6.5957,
+      "step": 8
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.829787234042553e-05,
+      "loss": 6.3764,
+      "step": 9
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.2553191489361704e-05,
+      "loss": 6.0142,
+      "step": 10
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.680851063829787e-05,
+      "loss": 6.0318,
+      "step": 11
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.1063829787234044e-05,
+      "loss": 6.2023,
+      "step": 12
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.531914893617021e-05,
+      "loss": 5.9664,
+      "step": 13
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.957446808510639e-05,
+      "loss": 5.8684,
+      "step": 14
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.382978723404255e-05,
+      "loss": 5.8419,
+      "step": 15
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.808510638297873e-05,
+      "loss": 5.6985,
+      "step": 16
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 7.234042553191488e-05,
+      "loss": 5.6815,
+      "step": 17
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 7.659574468085106e-05,
+      "loss": 5.618,
+      "step": 18
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.085106382978724e-05,
+      "loss": 5.6284,
+      "step": 19
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.510638297872341e-05,
+      "loss": 5.5076,
+      "step": 20
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.936170212765958e-05,
+      "loss": 5.5638,
+      "step": 21
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.361702127659574e-05,
+      "loss": 5.5847,
+      "step": 22
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.787234042553192e-05,
+      "loss": 5.4566,
+      "step": 23
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00010212765957446809,
+      "loss": 5.3719,
+      "step": 24
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00010638297872340425,
+      "loss": 5.6425,
+      "step": 25
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00011063829787234042,
+      "loss": 5.5902,
+      "step": 26
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0001148936170212766,
+      "loss": 5.2239,
+      "step": 27
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00011914893617021278,
+      "loss": 5.4397,
+      "step": 28
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00012340425531914893,
+      "loss": 5.4031,
+      "step": 29
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0001276595744680851,
+      "loss": 5.441,
+      "step": 30
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00013191489361702127,
+      "loss": 5.392,
+      "step": 31
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00013617021276595746,
+      "loss": 5.4408,
+      "step": 32
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00014042553191489363,
+      "loss": 5.2701,
+      "step": 33
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00014468085106382977,
+      "loss": 5.2233,
+      "step": 34
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00014893617021276596,
+      "loss": 5.1704,
+      "step": 35
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00015319148936170213,
+      "loss": 5.3029,
+      "step": 36
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00015744680851063832,
+      "loss": 5.213,
+      "step": 37
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00016170212765957449,
+      "loss": 5.3257,
+      "step": 38
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00016595744680851062,
+      "loss": 5.2267,
+      "step": 39
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00017021276595744682,
+      "loss": 5.2484,
+      "step": 40
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00017446808510638298,
+      "loss": 5.0193,
+      "step": 41
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00017872340425531915,
+      "loss": 5.2959,
+      "step": 42
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00018297872340425532,
+      "loss": 5.3462,
+      "step": 43
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00018723404255319148,
+      "loss": 5.1916,
+      "step": 44
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019148936170212765,
+      "loss": 5.2139,
+      "step": 45
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019574468085106384,
+      "loss": 5.2185,
+      "step": 46
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0002,
+      "loss": 5.3135,
+      "step": 47
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00020425531914893618,
+      "loss": 5.2488,
+      "step": 48
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00020851063829787234,
+      "loss": 5.1809,
+      "step": 49
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0002127659574468085,
+      "loss": 5.1186,
+      "step": 50
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0002170212765957447,
+      "loss": 5.2081,
+      "step": 51
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00022127659574468084,
+      "loss": 5.0977,
+      "step": 52
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.000225531914893617,
+      "loss": 5.2865,
+      "step": 53
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0002297872340425532,
+      "loss": 5.2788,
+      "step": 54
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00023404255319148937,
+      "loss": 5.2017,
+      "step": 55
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00023829787234042556,
+      "loss": 5.1572,
+      "step": 56
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0002425531914893617,
+      "loss": 5.2028,
+      "step": 57
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00024680851063829787,
+      "loss": 5.1875,
+      "step": 58
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00025106382978723403,
+      "loss": 5.0508,
+      "step": 59
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0002553191489361702,
+      "loss": 5.0979,
+      "step": 60
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00025957446808510637,
+      "loss": 5.167,
+      "step": 61
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00026382978723404253,
+      "loss": 5.1245,
+      "step": 62
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00026808510638297875,
+      "loss": 5.1894,
+      "step": 63
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0002723404255319149,
+      "loss": 5.0651,
+      "step": 64
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0002765957446808511,
+      "loss": 5.1648,
+      "step": 65
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00028085106382978725,
+      "loss": 5.0701,
+      "step": 66
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0002851063829787234,
+      "loss": 5.1611,
+      "step": 67
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00028936170212765953,
+      "loss": 5.0858,
+      "step": 68
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00029361702127659575,
+      "loss": 5.0026,
+      "step": 69
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0002978723404255319,
+      "loss": 5.1671,
+      "step": 70
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0003021276595744681,
+      "loss": 5.106,
+      "step": 71
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00030638297872340425,
+      "loss": 4.9204,
+      "step": 72
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0003106382978723404,
+      "loss": 5.115,
+      "step": 73
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00031489361702127664,
+      "loss": 4.9747,
+      "step": 74
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0003191489361702128,
+      "loss": 5.0885,
+      "step": 75
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00032340425531914897,
+      "loss": 5.1272,
+      "step": 76
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0003276595744680851,
+      "loss": 5.0675,
+      "step": 77
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00033191489361702125,
+      "loss": 5.0292,
+      "step": 78
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0003361702127659574,
+      "loss": 5.0696,
+      "step": 79
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00034042553191489364,
+      "loss": 5.1578,
+      "step": 80
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0003446808510638298,
+      "loss": 5.0533,
+      "step": 81
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00034893617021276597,
+      "loss": 5.0916,
+      "step": 82
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00035319148936170213,
+      "loss": 5.0296,
+      "step": 83
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0003574468085106383,
+      "loss": 4.9205,
+      "step": 84
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0003617021276595745,
+      "loss": 5.1009,
+      "step": 85
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00036595744680851063,
+      "loss": 5.0388,
+      "step": 86
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0003702127659574468,
+      "loss": 5.105,
+      "step": 87
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00037446808510638297,
+      "loss": 5.1434,
+      "step": 88
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00037872340425531913,
+      "loss": 5.1707,
+      "step": 89
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0003829787234042553,
+      "loss": 5.1167,
+      "step": 90
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0003872340425531915,
+      "loss": 5.166,
+      "step": 91
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0003914893617021277,
+      "loss": 5.0074,
+      "step": 92
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00039574468085106385,
+      "loss": 4.889,
+      "step": 93
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0004,
+      "loss": 4.9876,
+      "step": 94
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00040425531914893613,
+      "loss": 4.9492,
+      "step": 95
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00040851063829787235,
+      "loss": 4.9397,
+      "step": 96
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0004127659574468085,
+      "loss": 4.9026,
+      "step": 97
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0004170212765957447,
+      "loss": 5.0206,
+      "step": 98
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00042127659574468085,
+      "loss": 4.8089,
+      "step": 99
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.000425531914893617,
+      "loss": 5.0353,
+      "step": 100
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0004297872340425532,
+      "loss": 4.9311,
+      "step": 101
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0004340425531914894,
+      "loss": 4.8648,
+      "step": 102
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00043829787234042557,
+      "loss": 4.9794,
+      "step": 103
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0004425531914893617,
+      "loss": 4.6757,
+      "step": 104
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00044680851063829785,
+      "loss": 4.8003,
+      "step": 105
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.000451063829787234,
+      "loss": 4.9657,
+      "step": 106
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00045531914893617024,
+      "loss": 4.8779,
+      "step": 107
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0004595744680851064,
+      "loss": 4.9369,
+      "step": 108
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00046382978723404257,
+      "loss": 4.8687,
+      "step": 109
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00046808510638297874,
+      "loss": 4.9378,
+      "step": 110
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0004723404255319149,
+      "loss": 4.8214,
+      "step": 111
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0004765957446808511,
+      "loss": 4.7465,
+      "step": 112
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00048085106382978723,
+      "loss": 4.9237,
+      "step": 113
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0004851063829787234,
+      "loss": 4.9892,
+      "step": 114
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0004893617021276596,
+      "loss": 4.8622,
+      "step": 115
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0004936170212765957,
+      "loss": 4.7589,
+      "step": 116
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.000497872340425532,
+      "loss": 4.9604,
+      "step": 117
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0005021276595744681,
+      "loss": 4.8609,
+      "step": 118
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0005063829787234042,
+      "loss": 4.8081,
+      "step": 119
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0005106382978723404,
+      "loss": 4.7668,
+      "step": 120
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0005148936170212766,
+      "loss": 4.8921,
+      "step": 121
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0005191489361702127,
+      "loss": 4.8458,
+      "step": 122
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.000523404255319149,
+      "loss": 4.9136,
+      "step": 123
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0005276595744680851,
+      "loss": 4.9846,
+      "step": 124
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0005319148936170213,
+      "loss": 4.7408,
+      "step": 125
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0005361702127659575,
+      "loss": 4.7616,
+      "step": 126
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0005404255319148936,
+      "loss": 4.9788,
+      "step": 127
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0005446808510638298,
+      "loss": 4.9959,
+      "step": 128
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.000548936170212766,
+      "loss": 4.8863,
+      "step": 129
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0005531914893617022,
+      "loss": 4.8144,
+      "step": 130
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0005574468085106383,
+      "loss": 4.792,
+      "step": 131
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0005617021276595745,
+      "loss": 4.7578,
+      "step": 132
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0005659574468085107,
+      "loss": 4.9363,
+      "step": 133
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0005702127659574468,
+      "loss": 4.8889,
+      "step": 134
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0005744680851063831,
+      "loss": 4.6994,
+      "step": 135
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0005787234042553191,
+      "loss": 4.8257,
+      "step": 136
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0005829787234042553,
+      "loss": 4.7841,
+      "step": 137
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0005872340425531915,
+      "loss": 4.6846,
+      "step": 138
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0005914893617021276,
+      "loss": 4.9022,
+      "step": 139
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0005957446808510638,
+      "loss": 4.7742,
+      "step": 140
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0006,
+      "loss": 4.8432,
+      "step": 141
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0006042553191489362,
+      "loss": 4.9234,
+      "step": 142
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0006085106382978724,
+      "loss": 4.8216,
+      "step": 143
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0006127659574468085,
+      "loss": 4.7849,
+      "step": 144
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0006170212765957447,
+      "loss": 4.6616,
+      "step": 145
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0006212765957446808,
+      "loss": 4.7503,
+      "step": 146
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.000625531914893617,
+      "loss": 4.5514,
+      "step": 147
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0006297872340425533,
+      "loss": 4.7243,
+      "step": 148
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0006340425531914894,
+      "loss": 4.79,
+      "step": 149
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0006382978723404256,
+      "loss": 4.6775,
+      "step": 150
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0006425531914893617,
+      "loss": 4.8091,
+      "step": 151
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0006468085106382979,
+      "loss": 4.667,
+      "step": 152
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0006510638297872342,
+      "loss": 4.6569,
+      "step": 153
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0006553191489361702,
+      "loss": 4.6302,
+      "step": 154
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0006595744680851064,
+      "loss": 4.6782,
+      "step": 155
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0006638297872340425,
+      "loss": 4.7412,
+      "step": 156
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0006680851063829787,
+      "loss": 4.582,
+      "step": 157
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0006723404255319148,
+      "loss": 4.6194,
+      "step": 158
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.000676595744680851,
+      "loss": 4.8352,
+      "step": 159
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0006808510638297873,
+      "loss": 4.6905,
+      "step": 160
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0006851063829787234,
+      "loss": 4.591,
+      "step": 161
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0006893617021276596,
+      "loss": 4.8185,
+      "step": 162
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0006936170212765957,
+      "loss": 4.7164,
+      "step": 163
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0006978723404255319,
+      "loss": 4.6823,
+      "step": 164
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0007021276595744682,
+      "loss": 4.6696,
+      "step": 165
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0007063829787234043,
+      "loss": 4.6627,
+      "step": 166
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0007106382978723405,
+      "loss": 4.799,
+      "step": 167
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0007148936170212766,
+      "loss": 4.5271,
+      "step": 168
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0007191489361702128,
+      "loss": 4.7776,
+      "step": 169
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.000723404255319149,
+      "loss": 4.5363,
+      "step": 170
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0007276595744680852,
+      "loss": 4.7055,
+      "step": 171
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0007319148936170213,
+      "loss": 4.6881,
+      "step": 172
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0007361702127659574,
+      "loss": 4.5895,
+      "step": 173
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0007404255319148936,
+      "loss": 4.7209,
+      "step": 174
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0007446808510638298,
+      "loss": 4.7778,
+      "step": 175
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0007489361702127659,
+      "loss": 4.5703,
+      "step": 176
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0007531914893617022,
+      "loss": 4.7525,
+      "step": 177
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0007574468085106383,
+      "loss": 4.584,
+      "step": 178
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0007617021276595745,
+      "loss": 4.5994,
+      "step": 179
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0007659574468085106,
+      "loss": 4.6461,
+      "step": 180
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0007702127659574468,
+      "loss": 4.6053,
+      "step": 181
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.000774468085106383,
+      "loss": 4.5722,
+      "step": 182
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0007787234042553192,
+      "loss": 4.5702,
+      "step": 183
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0007829787234042554,
+      "loss": 4.5885,
+      "step": 184
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0007872340425531915,
+      "loss": 4.7327,
+      "step": 185
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0007914893617021277,
+      "loss": 4.6439,
+      "step": 186
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0007957446808510639,
+      "loss": 4.6348,
+      "step": 187
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0008,
+      "loss": 4.5735,
+      "step": 188
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0008042553191489363,
+      "loss": 4.6003,
+      "step": 189
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0008085106382978723,
+      "loss": 4.5682,
+      "step": 190
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0008127659574468085,
+      "loss": 4.4693,
+      "step": 191
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0008170212765957447,
+      "loss": 4.5299,
+      "step": 192
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0008212765957446808,
+      "loss": 4.7399,
+      "step": 193
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.000825531914893617,
+      "loss": 4.5125,
+      "step": 194
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0008297872340425531,
+      "loss": 4.7958,
+      "step": 195
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0008340425531914894,
+      "loss": 4.6565,
+      "step": 196
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0008382978723404256,
+      "loss": 4.5195,
+      "step": 197
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0008425531914893617,
+      "loss": 4.5109,
+      "step": 198
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0008468085106382979,
+      "loss": 4.6303,
+      "step": 199
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.000851063829787234,
+      "loss": 4.6224,
+      "step": 200
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0008553191489361703,
+      "loss": 4.5786,
+      "step": 201
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0008595744680851064,
+      "loss": 4.5364,
+      "step": 202
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0008638297872340426,
+      "loss": 4.6714,
+      "step": 203
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0008680851063829788,
+      "loss": 4.58,
+      "step": 204
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0008723404255319149,
+      "loss": 4.6119,
+      "step": 205
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0008765957446808511,
+      "loss": 4.5332,
+      "step": 206
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0008808510638297873,
+      "loss": 4.4933,
+      "step": 207
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0008851063829787234,
+      "loss": 4.6313,
+      "step": 208
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0008893617021276596,
+      "loss": 4.53,
+      "step": 209
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0008936170212765957,
+      "loss": 4.67,
+      "step": 210
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0008978723404255319,
+      "loss": 4.4501,
+      "step": 211
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.000902127659574468,
+      "loss": 4.5493,
+      "step": 212
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0009063829787234043,
+      "loss": 4.5255,
+      "step": 213
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0009106382978723405,
+      "loss": 4.4557,
+      "step": 214
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0009148936170212766,
+      "loss": 4.4246,
+      "step": 215
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0009191489361702128,
+      "loss": 4.3941,
+      "step": 216
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0009234042553191489,
+      "loss": 4.4731,
+      "step": 217
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0009276595744680851,
+      "loss": 4.5522,
+      "step": 218
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0009319148936170214,
+      "loss": 4.5119,
+      "step": 219
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0009361702127659575,
+      "loss": 4.6598,
+      "step": 220
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0009404255319148937,
+      "loss": 4.5204,
+      "step": 221
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0009446808510638298,
+      "loss": 4.5019,
+      "step": 222
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.000948936170212766,
+      "loss": 4.4943,
+      "step": 223
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0009531914893617022,
+      "loss": 4.5206,
+      "step": 224
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0009574468085106384,
+      "loss": 4.5076,
+      "step": 225
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0009617021276595745,
+      "loss": 4.525,
+      "step": 226
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0009659574468085106,
+      "loss": 4.516,
+      "step": 227
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0009702127659574468,
+      "loss": 4.5474,
+      "step": 228
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0009744680851063829,
+      "loss": 4.5543,
+      "step": 229
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0009787234042553192,
+      "loss": 4.3835,
+      "step": 230
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0009829787234042554,
+      "loss": 4.5311,
+      "step": 231
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0009872340425531915,
+      "loss": 4.4841,
+      "step": 232
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0009914893617021276,
+      "loss": 4.5842,
+      "step": 233
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.000995744680851064,
+      "loss": 4.3653,
+      "step": 234
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.001,
+      "loss": 4.5177,
+      "step": 235
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0009999999570221019,
+      "loss": 4.4598,
+      "step": 236
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0009999998280884144,
+      "loss": 4.415,
+      "step": 237
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0009999996131989603,
+      "loss": 4.6801,
+      "step": 238
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0009999993123537758,
+      "loss": 4.5289,
+      "step": 239
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0009999989255529132,
+      "loss": 4.4771,
+      "step": 240
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0009999984527964388,
+      "loss": 4.4301,
+      "step": 241
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.000999997894084434,
+      "loss": 4.3532,
+      "step": 242
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0009999972494169947,
+      "loss": 4.53,
+      "step": 243
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0009999965187942315,
+      "loss": 4.5443,
+      "step": 244
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0009999957022162706,
+      "loss": 4.4448,
+      "step": 245
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0009999947996832518,
+      "loss": 4.5396,
+      "step": 246
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0009999938111953305,
+      "loss": 4.5541,
+      "step": 247
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0009999927367526768,
+      "loss": 4.4026,
+      "step": 248
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0009999915763554753,
+      "loss": 4.325,
+      "step": 249
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0009999903300039253,
+      "loss": 4.5837,
+      "step": 250
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0009999889976982413,
+      "loss": 4.386,
+      "step": 251
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.000999987579438652,
+      "loss": 4.6375,
+      "step": 252
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0009999860752254014,
+      "loss": 4.4909,
+      "step": 253
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0009999844850587486,
+      "loss": 4.4272,
+      "step": 254
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.000999982808938966,
+      "loss": 4.5611,
+      "step": 255
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0009999810468663425,
+      "loss": 4.476,
+      "step": 256
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0009999791988411805,
+      "loss": 4.5587,
+      "step": 257
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0009999772648637984,
+      "loss": 4.3942,
+      "step": 258
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.000999975244934528,
+      "loss": 4.4885,
+      "step": 259
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0009999731390537168,
+      "loss": 4.4049,
+      "step": 260
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0009999709472217266,
+      "loss": 4.3371,
+      "step": 261
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0009999686694389348,
+      "loss": 4.4639,
+      "step": 262
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0009999663057057323,
+      "loss": 4.4727,
+      "step": 263
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0009999638560225258,
+      "loss": 4.5027,
+      "step": 264
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0009999613203897364,
+      "loss": 4.3945,
+      "step": 265
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0009999586988078,
+      "loss": 4.4141,
+      "step": 266
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0009999559912771672,
+      "loss": 4.423,
+      "step": 267
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0009999531977983036,
+      "loss": 4.3975,
+      "step": 268
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0009999503183716894,
+      "loss": 4.3833,
+      "step": 269
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0009999473529978193,
+      "loss": 4.3843,
+      "step": 270
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0009999443016772036,
+      "loss": 4.5668,
+      "step": 271
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0009999411644103665,
+      "loss": 4.453,
+      "step": 272
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0009999379411978473,
+      "loss": 4.5031,
+      "step": 273
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009999346320402002,
+      "loss": 4.3407,
+      "step": 274
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009999312369379943,
+      "loss": 4.4873,
+      "step": 275
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.000999927755891813,
+      "loss": 4.4522,
+      "step": 276
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009999241889022549,
+      "loss": 4.4631,
+      "step": 277
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.000999920535969933,
+      "loss": 4.3684,
+      "step": 278
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009999167970954753,
+      "loss": 4.4466,
+      "step": 279
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009999129722795248,
+      "loss": 4.3188,
+      "step": 280
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.000999909061522739,
+      "loss": 4.493,
+      "step": 281
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009999050648257899,
+      "loss": 4.3478,
+      "step": 282
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009999009821893647,
+      "loss": 4.4095,
+      "step": 283
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009998968136141653,
+      "loss": 4.4046,
+      "step": 284
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009998925591009084,
+      "loss": 4.4471,
+      "step": 285
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009998882186503255,
+      "loss": 4.4423,
+      "step": 286
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009998837922631624,
+      "loss": 4.3571,
+      "step": 287
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009998792799401804,
+      "loss": 4.4663,
+      "step": 288
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.000999874681682155,
+      "loss": 4.4846,
+      "step": 289
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009998699974898767,
+      "loss": 4.4407,
+      "step": 290
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.000999865227364151,
+      "loss": 4.4919,
+      "step": 291
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009998603713057977,
+      "loss": 4.5299,
+      "step": 292
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009998554293156516,
+      "loss": 4.3561,
+      "step": 293
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009998504013945626,
+      "loss": 4.3352,
+      "step": 294
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009998452875433948,
+      "loss": 4.3755,
+      "step": 295
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009998400877630272,
+      "loss": 4.2908,
+      "step": 296
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.000999834802054354,
+      "loss": 4.3443,
+      "step": 297
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009998294304182837,
+      "loss": 4.4393,
+      "step": 298
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009998239728557397,
+      "loss": 4.3029,
+      "step": 299
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009998184293676603,
+      "loss": 4.4092,
+      "step": 300
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009998127999549988,
+      "loss": 4.5313,
+      "step": 301
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009998070846187223,
+      "loss": 4.4596,
+      "step": 302
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.000999801283359814,
+      "loss": 4.4296,
+      "step": 303
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009997953961792708,
+      "loss": 4.5592,
+      "step": 304
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009997894230781047,
+      "loss": 4.4496,
+      "step": 305
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009997833640573428,
+      "loss": 4.4531,
+      "step": 306
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009997772191180268,
+      "loss": 4.5266,
+      "step": 307
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009997709882612128,
+      "loss": 4.3608,
+      "step": 308
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.000999764671487972,
+      "loss": 4.3768,
+      "step": 309
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009997582687993904,
+      "loss": 4.3571,
+      "step": 310
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.000999751780196569,
+      "loss": 4.3082,
+      "step": 311
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009997452056806227,
+      "loss": 4.3894,
+      "step": 312
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.000999738545252682,
+      "loss": 4.3844,
+      "step": 313
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009997317989138919,
+      "loss": 4.4798,
+      "step": 314
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009997249666654122,
+      "loss": 4.3873,
+      "step": 315
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009997180485084175,
+      "loss": 4.3916,
+      "step": 316
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.000999711044444097,
+      "loss": 4.4433,
+      "step": 317
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009997039544736546,
+      "loss": 4.396,
+      "step": 318
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009996967785983097,
+      "loss": 4.3425,
+      "step": 319
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009996895168192955,
+      "loss": 4.3503,
+      "step": 320
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009996821691378602,
+      "loss": 4.2771,
+      "step": 321
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009996747355552674,
+      "loss": 4.3853,
+      "step": 322
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.000999667216072795,
+      "loss": 4.4948,
+      "step": 323
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009996596106917352,
+      "loss": 4.3164,
+      "step": 324
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.000999651919413396,
+      "loss": 4.2347,
+      "step": 325
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009996441422390992,
+      "loss": 4.3117,
+      "step": 326
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009996362791701821,
+      "loss": 4.3534,
+      "step": 327
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009996283302079964,
+      "loss": 4.371,
+      "step": 328
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009996202953539083,
+      "loss": 4.4537,
+      "step": 329
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009996121746092997,
+      "loss": 4.3757,
+      "step": 330
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.000999603967975566,
+      "loss": 4.2849,
+      "step": 331
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009995956754541182,
+      "loss": 4.3442,
+      "step": 332
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009995872970463822,
+      "loss": 4.4481,
+      "step": 333
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009995788327537983,
+      "loss": 4.3853,
+      "step": 334
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009995702825778211,
+      "loss": 4.2853,
+      "step": 335
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009995616465199209,
+      "loss": 4.2632,
+      "step": 336
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009995529245815822,
+      "loss": 4.2882,
+      "step": 337
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009995441167643046,
+      "loss": 4.2917,
+      "step": 338
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.000999535223069602,
+      "loss": 4.3144,
+      "step": 339
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009995262434990037,
+      "loss": 4.323,
+      "step": 340
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.000999517178054053,
+      "loss": 4.4014,
+      "step": 341
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009995080267363081,
+      "loss": 4.273,
+      "step": 342
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.000999498789547343,
+      "loss": 4.3668,
+      "step": 343
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009994894664887453,
+      "loss": 4.3452,
+      "step": 344
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009994800575621175,
+      "loss": 4.3811,
+      "step": 345
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009994705627690776,
+      "loss": 4.3462,
+      "step": 346
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009994609821112575,
+      "loss": 4.2733,
+      "step": 347
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009994513155903043,
+      "loss": 4.2863,
+      "step": 348
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009994415632078796,
+      "loss": 4.3406,
+      "step": 349
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009994317249656607,
+      "loss": 4.2386,
+      "step": 350
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.000999421800865338,
+      "loss": 4.24,
+      "step": 351
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009994117909086179,
+      "loss": 4.2968,
+      "step": 352
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009994016950972213,
+      "loss": 4.3605,
+      "step": 353
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.000999391513432884,
+      "loss": 4.2786,
+      "step": 354
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009993812459173556,
+      "loss": 4.3255,
+      "step": 355
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.000999370892552402,
+      "loss": 4.2382,
+      "step": 356
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.000999360453339803,
+      "loss": 4.3485,
+      "step": 357
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009993499282813527,
+      "loss": 4.2823,
+      "step": 358
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009993393173788607,
+      "loss": 4.4509,
+      "step": 359
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009993286206341513,
+      "loss": 4.4178,
+      "step": 360
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009993178380490635,
+      "loss": 4.4537,
+      "step": 361
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009993069696254506,
+      "loss": 4.3264,
+      "step": 362
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009992960153651811,
+      "loss": 4.4247,
+      "step": 363
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009992849752701383,
+      "loss": 4.4142,
+      "step": 364
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00099927384934222,
+      "loss": 4.1958,
+      "step": 365
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009992626375833389,
+      "loss": 4.2842,
+      "step": 366
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009992513399954225,
+      "loss": 4.3154,
+      "step": 367
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009992399565804128,
+      "loss": 4.1561,
+      "step": 368
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.000999228487340267,
+      "loss": 4.2472,
+      "step": 369
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009992169322769566,
+      "loss": 4.2992,
+      "step": 370
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009992052913924683,
+      "loss": 4.4289,
+      "step": 371
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.000999193564688803,
+      "loss": 4.2912,
+      "step": 372
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009991817521679768,
+      "loss": 4.4671,
+      "step": 373
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009991698538320203,
+      "loss": 4.3618,
+      "step": 374
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009991578696829794,
+      "loss": 4.1809,
+      "step": 375
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009991457997229134,
+      "loss": 4.431,
+      "step": 376
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009991336439538984,
+      "loss": 4.3021,
+      "step": 377
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.000999121402378023,
+      "loss": 4.2584,
+      "step": 378
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009991090749973926,
+      "loss": 4.3353,
+      "step": 379
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.000999096661814126,
+      "loss": 4.3731,
+      "step": 380
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009990841628303572,
+      "loss": 4.3414,
+      "step": 381
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009990715780482347,
+      "loss": 4.2675,
+      "step": 382
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009990589074699225,
+      "loss": 4.39,
+      "step": 383
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009990461510975984,
+      "loss": 4.328,
+      "step": 384
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009990333089334555,
+      "loss": 4.3561,
+      "step": 385
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009990203809797013,
+      "loss": 4.3047,
+      "step": 386
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009990073672385587,
+      "loss": 4.3361,
+      "step": 387
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009989942677122647,
+      "loss": 4.2677,
+      "step": 388
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009989810824030713,
+      "loss": 4.2771,
+      "step": 389
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.000998967811313245,
+      "loss": 4.3491,
+      "step": 390
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009989544544450673,
+      "loss": 4.4156,
+      "step": 391
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009989410118008348,
+      "loss": 4.0586,
+      "step": 392
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.000998927483382858,
+      "loss": 4.2559,
+      "step": 393
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009989138691934628,
+      "loss": 4.2544,
+      "step": 394
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009989001692349894,
+      "loss": 4.4056,
+      "step": 395
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009988863835097932,
+      "loss": 4.284,
+      "step": 396
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009988725120202442,
+      "loss": 4.215,
+      "step": 397
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.000998858554768727,
+      "loss": 4.2201,
+      "step": 398
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009988445117576409,
+      "loss": 4.2996,
+      "step": 399
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009988303829894002,
+      "loss": 4.123,
+      "step": 400
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009988161684664335,
+      "loss": 4.3475,
+      "step": 401
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009988018681911849,
+      "loss": 4.3206,
+      "step": 402
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009987874821661123,
+      "loss": 4.1729,
+      "step": 403
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009987730103936895,
+      "loss": 4.3035,
+      "step": 404
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009987584528764035,
+      "loss": 4.3161,
+      "step": 405
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009987438096167576,
+      "loss": 4.2774,
+      "step": 406
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009987290806172688,
+      "loss": 4.2526,
+      "step": 407
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009987142658804693,
+      "loss": 4.417,
+      "step": 408
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009986993654089059,
+      "loss": 4.2937,
+      "step": 409
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00099868437920514,
+      "loss": 4.3641,
+      "step": 410
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009986693072717484,
+      "loss": 4.3047,
+      "step": 411
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009986541496113215,
+      "loss": 4.2684,
+      "step": 412
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009986389062264657,
+      "loss": 4.3095,
+      "step": 413
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009986235771198009,
+      "loss": 4.2352,
+      "step": 414
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.000998608162293963,
+      "loss": 4.2743,
+      "step": 415
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009985926617516012,
+      "loss": 4.3155,
+      "step": 416
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009985770754953808,
+      "loss": 4.3693,
+      "step": 417
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009985614035279812,
+      "loss": 4.3679,
+      "step": 418
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009985456458520968,
+      "loss": 4.3151,
+      "step": 419
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.000998529802470436,
+      "loss": 4.2477,
+      "step": 420
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009985138733857224,
+      "loss": 4.1424,
+      "step": 421
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.000998497858600695,
+      "loss": 4.2872,
+      "step": 422
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009984817581181068,
+      "loss": 4.2843,
+      "step": 423
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009984655719407253,
+      "loss": 4.4452,
+      "step": 424
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.000998449300071333,
+      "loss": 4.2484,
+      "step": 425
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009984329425127279,
+      "loss": 4.353,
+      "step": 426
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009984164992677214,
+      "loss": 4.2656,
+      "step": 427
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009983999703391407,
+      "loss": 4.1845,
+      "step": 428
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.000998383355729827,
+      "loss": 4.3028,
+      "step": 429
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009983666554426367,
+      "loss": 4.4146,
+      "step": 430
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.000998349869480441,
+      "loss": 4.3967,
+      "step": 431
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009983329978461252,
+      "loss": 4.4148,
+      "step": 432
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009983160405425899,
+      "loss": 4.2718,
+      "step": 433
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009982989975727502,
+      "loss": 4.2474,
+      "step": 434
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009982818689395362,
+      "loss": 4.369,
+      "step": 435
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009982646546458922,
+      "loss": 4.1269,
+      "step": 436
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009982473546947777,
+      "loss": 4.2366,
+      "step": 437
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009982299690891668,
+      "loss": 4.2365,
+      "step": 438
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009982124978320482,
+      "loss": 4.2319,
+      "step": 439
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009981949409264255,
+      "loss": 4.3534,
+      "step": 440
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009981772983753167,
+      "loss": 4.2998,
+      "step": 441
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.000998159570181755,
+      "loss": 4.2675,
+      "step": 442
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009981417563487884,
+      "loss": 4.2846,
+      "step": 443
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009981238568794787,
+      "loss": 4.3809,
+      "step": 444
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.000998105871776903,
+      "loss": 4.2481,
+      "step": 445
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009980878010441537,
+      "loss": 4.2871,
+      "step": 446
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.000998069644684337,
+      "loss": 4.207,
+      "step": 447
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009980514027005743,
+      "loss": 4.3354,
+      "step": 448
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009980330750960014,
+      "loss": 4.1758,
+      "step": 449
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009980146618737693,
+      "loss": 4.3089,
+      "step": 450
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009979961630370434,
+      "loss": 4.2996,
+      "step": 451
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009979775785890037,
+      "loss": 4.3227,
+      "step": 452
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009979589085328453,
+      "loss": 4.3129,
+      "step": 453
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009979401528717776,
+      "loss": 4.2824,
+      "step": 454
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009979213116090252,
+      "loss": 4.0775,
+      "step": 455
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009979023847478266,
+      "loss": 4.1649,
+      "step": 456
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009978833722914362,
+      "loss": 4.276,
+      "step": 457
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.000997864274243122,
+      "loss": 4.3562,
+      "step": 458
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009978450906061673,
+      "loss": 4.2172,
+      "step": 459
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00099782582138387,
+      "loss": 4.2407,
+      "step": 460
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009978064665795428,
+      "loss": 4.2175,
+      "step": 461
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009977870261965128,
+      "loss": 4.2504,
+      "step": 462
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.000997767500238122,
+      "loss": 4.1729,
+      "step": 463
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009977478887077276,
+      "loss": 4.3534,
+      "step": 464
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009977281916087007,
+      "loss": 4.1908,
+      "step": 465
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009977084089444275,
+      "loss": 4.2188,
+      "step": 466
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.000997688540718309,
+      "loss": 4.287,
+      "step": 467
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009976685869337602,
+      "loss": 4.325,
+      "step": 468
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009976485475942122,
+      "loss": 4.2347,
+      "step": 469
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009976284227031097,
+      "loss": 4.1932,
+      "step": 470
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.000997608212263912,
+      "loss": 4.3117,
+      "step": 471
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.000997587916280094,
+      "loss": 4.2574,
+      "step": 472
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009975675347551447,
+      "loss": 4.3125,
+      "step": 473
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.000997547067692568,
+      "loss": 4.2276,
+      "step": 474
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.000997526515095882,
+      "loss": 4.3199,
+      "step": 475
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009975058769686204,
+      "loss": 4.1536,
+      "step": 476
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.000997485153314331,
+      "loss": 4.2108,
+      "step": 477
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009974643441365764,
+      "loss": 4.3284,
+      "step": 478
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.000997443449438934,
+      "loss": 4.3745,
+      "step": 479
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009974224692249956,
+      "loss": 4.153,
+      "step": 480
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009974014034983684,
+      "loss": 4.1692,
+      "step": 481
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009973802522626732,
+      "loss": 4.3113,
+      "step": 482
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009973590155215467,
+      "loss": 4.286,
+      "step": 483
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009973376932786396,
+      "loss": 4.1465,
+      "step": 484
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009973162855376172,
+      "loss": 4.3578,
+      "step": 485
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009972947923021602,
+      "loss": 4.3713,
+      "step": 486
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.000997273213575963,
+      "loss": 4.3207,
+      "step": 487
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009972515493627356,
+      "loss": 4.2425,
+      "step": 488
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009972297996662023,
+      "loss": 4.1976,
+      "step": 489
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009972079644901019,
+      "loss": 4.3326,
+      "step": 490
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009971860438381885,
+      "loss": 4.3012,
+      "step": 491
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009971640377142302,
+      "loss": 4.1957,
+      "step": 492
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009971419461220102,
+      "loss": 4.2511,
+      "step": 493
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009971197690653263,
+      "loss": 4.1707,
+      "step": 494
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.000997097506547991,
+      "loss": 4.3112,
+      "step": 495
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009970751585738314,
+      "loss": 4.1791,
+      "step": 496
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009970527251466894,
+      "loss": 4.2053,
+      "step": 497
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009970302062704216,
+      "loss": 4.3588,
+      "step": 498
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009970076019488996,
+      "loss": 4.2959,
+      "step": 499
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009969849121860088,
+      "loss": 4.1987,
+      "step": 500
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009969621369856501,
+      "loss": 4.3253,
+      "step": 501
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009969392763517387,
+      "loss": 4.2348,
+      "step": 502
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009969163302882048,
+      "loss": 4.2469,
+      "step": 503
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009968932987989927,
+      "loss": 4.3061,
+      "step": 504
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009968701818880623,
+      "loss": 4.2116,
+      "step": 505
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009968469795593874,
+      "loss": 4.2746,
+      "step": 506
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009968236918169566,
+      "loss": 4.2775,
+      "step": 507
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009968003186647735,
+      "loss": 4.1734,
+      "step": 508
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009967768601068566,
+      "loss": 4.1794,
+      "step": 509
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.000996753316147238,
+      "loss": 4.106,
+      "step": 510
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009967296867899656,
+      "loss": 4.2004,
+      "step": 511
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009967059720391014,
+      "loss": 4.2034,
+      "step": 512
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009966821718987223,
+      "loss": 4.3276,
+      "step": 513
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009966582863729198,
+      "loss": 4.2353,
+      "step": 514
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009966343154658001,
+      "loss": 4.1553,
+      "step": 515
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009966102591814842,
+      "loss": 4.2862,
+      "step": 516
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009965861175241075,
+      "loss": 4.2796,
+      "step": 517
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009965618904978202,
+      "loss": 4.343,
+      "step": 518
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009965375781067873,
+      "loss": 4.2059,
+      "step": 519
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009965131803551886,
+      "loss": 4.4071,
+      "step": 520
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009964886972472179,
+      "loss": 4.2661,
+      "step": 521
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009964641287870842,
+      "loss": 4.209,
+      "step": 522
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009964394749790116,
+      "loss": 4.2014,
+      "step": 523
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009964147358272378,
+      "loss": 4.2543,
+      "step": 524
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.000996389911336016,
+      "loss": 4.1866,
+      "step": 525
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.000996365001509614,
+      "loss": 4.1834,
+      "step": 526
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009963400063523135,
+      "loss": 4.1575,
+      "step": 527
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009963149258684122,
+      "loss": 4.2933,
+      "step": 528
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.000996289760062221,
+      "loss": 4.2531,
+      "step": 529
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.000996264508938067,
+      "loss": 4.2318,
+      "step": 530
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009962391725002904,
+      "loss": 4.2052,
+      "step": 531
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009962137507532475,
+      "loss": 4.2975,
+      "step": 532
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.000996188243701308,
+      "loss": 4.1915,
+      "step": 533
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.000996162651348857,
+      "loss": 4.2488,
+      "step": 534
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009961369737002943,
+      "loss": 4.1815,
+      "step": 535
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009961112107600343,
+      "loss": 4.2991,
+      "step": 536
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009960853625325054,
+      "loss": 4.0556,
+      "step": 537
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009960594290221519,
+      "loss": 4.1558,
+      "step": 538
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009960334102334315,
+      "loss": 4.3469,
+      "step": 539
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009960073061708176,
+      "loss": 4.1308,
+      "step": 540
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009959811168387973,
+      "loss": 4.2683,
+      "step": 541
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009959548422418733,
+      "loss": 4.1038,
+      "step": 542
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009959284823845623,
+      "loss": 4.109,
+      "step": 543
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009959020372713957,
+      "loss": 4.4033,
+      "step": 544
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00099587550690692,
+      "loss": 4.1941,
+      "step": 545
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.000995848891295696,
+      "loss": 4.1802,
+      "step": 546
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009958221904422992,
+      "loss": 4.1735,
+      "step": 547
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00099579540435132,
+      "loss": 4.0861,
+      "step": 548
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009957685330273629,
+      "loss": 4.2076,
+      "step": 549
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009957415764750475,
+      "loss": 4.1514,
+      "step": 550
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.000995714534699008,
+      "loss": 4.2164,
+      "step": 551
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.000995687407703893,
+      "loss": 4.248,
+      "step": 552
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009956601954943663,
+      "loss": 4.4212,
+      "step": 553
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009956328980751058,
+      "loss": 4.228,
+      "step": 554
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.000995605515450804,
+      "loss": 4.2322,
+      "step": 555
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009955780476261689,
+      "loss": 4.2615,
+      "step": 556
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.000995550494605922,
+      "loss": 4.254,
+      "step": 557
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009955228563948003,
+      "loss": 4.2075,
+      "step": 558
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009954951329975547,
+      "loss": 4.2033,
+      "step": 559
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009954673244189519,
+      "loss": 4.1651,
+      "step": 560
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009954394306637717,
+      "loss": 4.1107,
+      "step": 561
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00099541145173681,
+      "loss": 4.1561,
+      "step": 562
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009953833876428762,
+      "loss": 4.201,
+      "step": 563
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009953552383867952,
+      "loss": 4.184,
+      "step": 564
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.000995327003973406,
+      "loss": 4.1394,
+      "step": 565
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009952986844075627,
+      "loss": 4.1967,
+      "step": 566
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009952702796941334,
+      "loss": 4.2485,
+      "step": 567
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009952417898380014,
+      "loss": 4.1671,
+      "step": 568
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009952132148440644,
+      "loss": 4.3208,
+      "step": 569
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009951845547172347,
+      "loss": 4.1909,
+      "step": 570
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009951558094624395,
+      "loss": 4.1181,
+      "step": 571
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009951269790846203,
+      "loss": 4.2444,
+      "step": 572
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009950980635887333,
+      "loss": 4.2139,
+      "step": 573
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009950690629797494,
+      "loss": 4.1536,
+      "step": 574
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009950399772626543,
+      "loss": 4.1716,
+      "step": 575
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009950108064424481,
+      "loss": 4.1666,
+      "step": 576
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009949815505241458,
+      "loss": 4.2409,
+      "step": 577
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009949522095127766,
+      "loss": 4.0499,
+      "step": 578
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009949227834133846,
+      "loss": 4.2323,
+      "step": 579
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009948932722310283,
+      "loss": 4.1923,
+      "step": 580
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009948636759707814,
+      "loss": 4.2339,
+      "step": 581
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009948339946377316,
+      "loss": 4.1881,
+      "step": 582
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009948042282369816,
+      "loss": 4.3077,
+      "step": 583
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009947743767736484,
+      "loss": 4.1423,
+      "step": 584
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009947444402528638,
+      "loss": 4.1615,
+      "step": 585
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009947144186797745,
+      "loss": 4.2461,
+      "step": 586
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009946843120595414,
+      "loss": 4.1967,
+      "step": 587
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00099465412039734,
+      "loss": 4.1054,
+      "step": 588
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.000994623843698361,
+      "loss": 4.1144,
+      "step": 589
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.000994593481967809,
+      "loss": 4.1514,
+      "step": 590
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009945630352109038,
+      "loss": 4.1572,
+      "step": 591
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009945325034328792,
+      "loss": 4.1278,
+      "step": 592
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009945018866389843,
+      "loss": 4.1622,
+      "step": 593
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009944711848344822,
+      "loss": 4.1781,
+      "step": 594
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.000994440398024651,
+      "loss": 4.2073,
+      "step": 595
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009944095262147835,
+      "loss": 4.3525,
+      "step": 596
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009943785694101865,
+      "loss": 4.1708,
+      "step": 597
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009943475276161823,
+      "loss": 4.1981,
+      "step": 598
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009943164008381071,
+      "loss": 4.1878,
+      "step": 599
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.000994285189081312,
+      "loss": 4.2562,
+      "step": 600
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009942538923511626,
+      "loss": 4.2218,
+      "step": 601
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.000994222510653039,
+      "loss": 4.0447,
+      "step": 602
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009941910439923366,
+      "loss": 4.3266,
+      "step": 603
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009941594923744647,
+      "loss": 4.3206,
+      "step": 604
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.000994127855804847,
+      "loss": 4.1174,
+      "step": 605
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009940961342889225,
+      "loss": 4.2775,
+      "step": 606
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009940643278321447,
+      "loss": 4.2774,
+      "step": 607
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009940324364399809,
+      "loss": 4.1721,
+      "step": 608
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009940004601179141,
+      "loss": 4.2187,
+      "step": 609
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009939683988714413,
+      "loss": 4.2521,
+      "step": 610
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009939362527060743,
+      "loss": 4.1898,
+      "step": 611
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009939040216273391,
+      "loss": 4.1981,
+      "step": 612
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009938717056407768,
+      "loss": 4.1847,
+      "step": 613
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009938393047519427,
+      "loss": 4.2188,
+      "step": 614
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009938068189664071,
+      "loss": 4.1812,
+      "step": 615
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009937742482897549,
+      "loss": 4.1381,
+      "step": 616
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009937415927275847,
+      "loss": 4.0827,
+      "step": 617
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.000993708852285511,
+      "loss": 4.1879,
+      "step": 618
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009936760269691621,
+      "loss": 4.3077,
+      "step": 619
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009936431167841806,
+      "loss": 4.1513,
+      "step": 620
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.000993610121736225,
+      "loss": 4.0601,
+      "step": 621
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009935770418309667,
+      "loss": 4.2478,
+      "step": 622
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.000993543877074093,
+      "loss": 4.2059,
+      "step": 623
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009935106274713053,
+      "loss": 4.3732,
+      "step": 624
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009934772930283193,
+      "loss": 4.1827,
+      "step": 625
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009934438737508658,
+      "loss": 4.2402,
+      "step": 626
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00099341036964469,
+      "loss": 4.1272,
+      "step": 627
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009933767807155515,
+      "loss": 4.0786,
+      "step": 628
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009933431069692245,
+      "loss": 4.1189,
+      "step": 629
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009933093484114982,
+      "loss": 4.2219,
+      "step": 630
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.000993275505048176,
+      "loss": 4.2146,
+      "step": 631
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.000993241576885076,
+      "loss": 4.0692,
+      "step": 632
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009932075639280306,
+      "loss": 4.2116,
+      "step": 633
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009931734661828874,
+      "loss": 4.2287,
+      "step": 634
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.000993139283655508,
+      "loss": 4.2473,
+      "step": 635
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009931050163517687,
+      "loss": 4.3564,
+      "step": 636
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009930706642775607,
+      "loss": 4.208,
+      "step": 637
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009930362274387892,
+      "loss": 4.2075,
+      "step": 638
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009930017058413744,
+      "loss": 3.954,
+      "step": 639
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.000992967099491251,
+      "loss": 4.1679,
+      "step": 640
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009929324083943682,
+      "loss": 4.1029,
+      "step": 641
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00099289763255669,
+      "loss": 4.1714,
+      "step": 642
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009928627719841944,
+      "loss": 4.2196,
+      "step": 643
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009928278266828747,
+      "loss": 4.2297,
+      "step": 644
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009927927966587383,
+      "loss": 4.251,
+      "step": 645
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.000992757681917807,
+      "loss": 4.2131,
+      "step": 646
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009927224824661178,
+      "loss": 4.1081,
+      "step": 647
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009926871983097216,
+      "loss": 4.1528,
+      "step": 648
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009926518294546844,
+      "loss": 4.2689,
+      "step": 649
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009926163759070863,
+      "loss": 4.1577,
+      "step": 650
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009925808376730224,
+      "loss": 4.3138,
+      "step": 651
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.000992545214758602,
+      "loss": 4.1247,
+      "step": 652
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.000992509507169949,
+      "loss": 4.0727,
+      "step": 653
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009924737149132022,
+      "loss": 4.1719,
+      "step": 654
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009924378379945146,
+      "loss": 4.1019,
+      "step": 655
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009924018764200537,
+      "loss": 4.2413,
+      "step": 656
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009923658301960018,
+      "loss": 4.2678,
+      "step": 657
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009923296993285557,
+      "loss": 4.2557,
+      "step": 658
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009922934838239269,
+      "loss": 4.1412,
+      "step": 659
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009922571836883409,
+      "loss": 4.1968,
+      "step": 660
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009922207989280383,
+      "loss": 4.0954,
+      "step": 661
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.000992184329549274,
+      "loss": 4.068,
+      "step": 662
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009921477755583177,
+      "loss": 4.1224,
+      "step": 663
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009921111369614532,
+      "loss": 4.1925,
+      "step": 664
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009920744137649792,
+      "loss": 4.1397,
+      "step": 665
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.000992037605975209,
+      "loss": 4.1334,
+      "step": 666
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00099200071359847,
+      "loss": 4.1592,
+      "step": 667
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009919637366411047,
+      "loss": 4.2981,
+      "step": 668
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009919266751094696,
+      "loss": 4.2947,
+      "step": 669
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009918895290099363,
+      "loss": 4.2564,
+      "step": 670
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009918522983488904,
+      "loss": 4.0701,
+      "step": 671
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009918149831327323,
+      "loss": 4.1001,
+      "step": 672
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009917775833678771,
+      "loss": 4.2423,
+      "step": 673
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.000991740099060754,
+      "loss": 4.2868,
+      "step": 674
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009917025302178073,
+      "loss": 4.2469,
+      "step": 675
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009916648768454955,
+      "loss": 4.0955,
+      "step": 676
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.000991627138950291,
+      "loss": 4.1084,
+      "step": 677
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009915893165386824,
+      "loss": 4.2161,
+      "step": 678
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009915514096171711,
+      "loss": 4.2574,
+      "step": 679
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009915134181922739,
+      "loss": 4.1401,
+      "step": 680
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.000991475342270522,
+      "loss": 4.2548,
+      "step": 681
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009914371818584612,
+      "loss": 4.2647,
+      "step": 682
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009913989369626515,
+      "loss": 4.217,
+      "step": 683
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009913606075896675,
+      "loss": 4.2308,
+      "step": 684
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.000991322193746099,
+      "loss": 4.3254,
+      "step": 685
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009912836954385496,
+      "loss": 4.1446,
+      "step": 686
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009912451126736373,
+      "loss": 4.1579,
+      "step": 687
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009912064454579953,
+      "loss": 4.1351,
+      "step": 688
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009911676937982706,
+      "loss": 4.1584,
+      "step": 689
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009911288577011254,
+      "loss": 4.2695,
+      "step": 690
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009910899371732358,
+      "loss": 4.2378,
+      "step": 691
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009910509322212928,
+      "loss": 4.2285,
+      "step": 692
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.000991011842852002,
+      "loss": 4.0786,
+      "step": 693
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009909726690720828,
+      "loss": 4.043,
+      "step": 694
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00099093341088827,
+      "loss": 4.0948,
+      "step": 695
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009908940683073127,
+      "loss": 4.2152,
+      "step": 696
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.000990854641335974,
+      "loss": 4.1761,
+      "step": 697
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.000990815129981032,
+      "loss": 4.1248,
+      "step": 698
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.000990775534249279,
+      "loss": 4.0936,
+      "step": 699
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009907358541475222,
+      "loss": 4.3576,
+      "step": 700
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009906960896825829,
+      "loss": 4.0967,
+      "step": 701
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009906562408612974,
+      "loss": 4.1129,
+      "step": 702
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009906163076905158,
+      "loss": 3.9179,
+      "step": 703
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.000990576290177103,
+      "loss": 4.207,
+      "step": 704
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009905361883279387,
+      "loss": 4.1484,
+      "step": 705
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.000990496002149917,
+      "loss": 4.1597,
+      "step": 706
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009904557316499463,
+      "loss": 4.1566,
+      "step": 707
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009904153768349492,
+      "loss": 4.1692,
+      "step": 708
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009903749377118637,
+      "loss": 4.17,
+      "step": 709
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009903344142876414,
+      "loss": 4.2475,
+      "step": 710
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009902938065692486,
+      "loss": 4.2338,
+      "step": 711
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.000990253114563667,
+      "loss": 4.0843,
+      "step": 712
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.000990212338277891,
+      "loss": 4.0303,
+      "step": 713
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009901714777189312,
+      "loss": 4.1928,
+      "step": 714
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009901305328938118,
+      "loss": 4.1946,
+      "step": 715
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.000990089503809572,
+      "loss": 4.1935,
+      "step": 716
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009900483904732645,
+      "loss": 4.2213,
+      "step": 717
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009900071928919578,
+      "loss": 4.0803,
+      "step": 718
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009899659110727337,
+      "loss": 4.2221,
+      "step": 719
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009899245450226896,
+      "loss": 4.0755,
+      "step": 720
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009898830947489366,
+      "loss": 4.1433,
+      "step": 721
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009898415602586001,
+      "loss": 4.0479,
+      "step": 722
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.000989799941558821,
+      "loss": 4.1367,
+      "step": 723
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.000989758238656753,
+      "loss": 4.1752,
+      "step": 724
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009897164515595667,
+      "loss": 4.1695,
+      "step": 725
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009896745802744449,
+      "loss": 4.1199,
+      "step": 726
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.000989632624808586,
+      "loss": 4.1922,
+      "step": 727
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009895905851692026,
+      "loss": 4.124,
+      "step": 728
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009895484613635216,
+      "loss": 4.1287,
+      "step": 729
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009895062533987848,
+      "loss": 4.166,
+      "step": 730
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009894639612822486,
+      "loss": 4.157,
+      "step": 731
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009894215850211824,
+      "loss": 4.2073,
+      "step": 732
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009893791246228724,
+      "loss": 4.129,
+      "step": 733
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009893365800946175,
+      "loss": 4.2051,
+      "step": 734
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009892939514437313,
+      "loss": 4.2601,
+      "step": 735
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009892512386775428,
+      "loss": 4.2251,
+      "step": 736
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.000989208441803394,
+      "loss": 4.0462,
+      "step": 737
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009891655608286433,
+      "loss": 4.1964,
+      "step": 738
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009891225957606613,
+      "loss": 3.9848,
+      "step": 739
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.000989079546606835,
+      "loss": 4.1685,
+      "step": 740
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009890364133745646,
+      "loss": 4.2485,
+      "step": 741
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009889931960712652,
+      "loss": 4.0872,
+      "step": 742
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009889498947043668,
+      "loss": 4.1116,
+      "step": 743
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.000988906509281313,
+      "loss": 4.2286,
+      "step": 744
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009888630398095623,
+      "loss": 4.121,
+      "step": 745
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009888194862965878,
+      "loss": 4.1468,
+      "step": 746
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009887758487498765,
+      "loss": 4.2079,
+      "step": 747
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009887321271769304,
+      "loss": 4.0431,
+      "step": 748
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.000988688321585266,
+      "loss": 4.1662,
+      "step": 749
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009886444319824137,
+      "loss": 4.0857,
+      "step": 750
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009886004583759187,
+      "loss": 4.1687,
+      "step": 751
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009885564007733405,
+      "loss": 4.1933,
+      "step": 752
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009885122591822533,
+      "loss": 4.2113,
+      "step": 753
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009884680336102452,
+      "loss": 4.2054,
+      "step": 754
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009884237240649193,
+      "loss": 4.1364,
+      "step": 755
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009883793305538932,
+      "loss": 4.1587,
+      "step": 756
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009883348530847983,
+      "loss": 4.1246,
+      "step": 757
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009882902916652809,
+      "loss": 4.3303,
+      "step": 758
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009882456463030016,
+      "loss": 3.9803,
+      "step": 759
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009882009170056353,
+      "loss": 4.1615,
+      "step": 760
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.000988156103780872,
+      "loss": 4.1683,
+      "step": 761
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.000988111206636415,
+      "loss": 4.1089,
+      "step": 762
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.000988066225579983,
+      "loss": 4.3317,
+      "step": 763
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009880211606193088,
+      "loss": 4.0918,
+      "step": 764
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009879760117621392,
+      "loss": 4.0331,
+      "step": 765
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009879307790162362,
+      "loss": 4.0704,
+      "step": 766
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.000987885462389376,
+      "loss": 3.9987,
+      "step": 767
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009878400618893483,
+      "loss": 4.1075,
+      "step": 768
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009877945775239585,
+      "loss": 4.1919,
+      "step": 769
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009877490093010259,
+      "loss": 4.1534,
+      "step": 770
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009877033572283842,
+      "loss": 4.1773,
+      "step": 771
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009876576213138815,
+      "loss": 4.2362,
+      "step": 772
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009876118015653803,
+      "loss": 4.0887,
+      "step": 773
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009875658979907574,
+      "loss": 4.287,
+      "step": 774
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009875199105979042,
+      "loss": 4.1098,
+      "step": 775
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009874738393947267,
+      "loss": 4.1825,
+      "step": 776
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009874276843891449,
+      "loss": 4.1222,
+      "step": 777
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009873814455890935,
+      "loss": 4.2031,
+      "step": 778
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009873351230025211,
+      "loss": 4.1513,
+      "step": 779
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009872887166373916,
+      "loss": 4.2349,
+      "step": 780
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009872422265016826,
+      "loss": 4.1616,
+      "step": 781
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009871956526033862,
+      "loss": 4.0945,
+      "step": 782
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009871489949505091,
+      "loss": 4.1322,
+      "step": 783
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009871022535510722,
+      "loss": 4.1013,
+      "step": 784
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009870554284131111,
+      "loss": 4.0359,
+      "step": 785
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009870085195446752,
+      "loss": 4.1805,
+      "step": 786
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.000986961526953829,
+      "loss": 4.1985,
+      "step": 787
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.000986914450648651,
+      "loss": 4.129,
+      "step": 788
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009868672906372341,
+      "loss": 4.1444,
+      "step": 789
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009868200469276857,
+      "loss": 4.2283,
+      "step": 790
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009867727195281275,
+      "loss": 4.2048,
+      "step": 791
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009867253084466956,
+      "loss": 4.2429,
+      "step": 792
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009866778136915408,
+      "loss": 4.1204,
+      "step": 793
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009866302352708276,
+      "loss": 4.1252,
+      "step": 794
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009865825731927354,
+      "loss": 4.105,
+      "step": 795
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.000986534827465458,
+      "loss": 4.1911,
+      "step": 796
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009864869980972034,
+      "loss": 4.0226,
+      "step": 797
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.000986439085096194,
+      "loss": 3.9766,
+      "step": 798
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009863910884706666,
+      "loss": 4.2318,
+      "step": 799
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009863430082288722,
+      "loss": 4.3046,
+      "step": 800
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009862948443790766,
+      "loss": 4.0226,
+      "step": 801
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009862465969295597,
+      "loss": 4.1702,
+      "step": 802
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009861982658886158,
+      "loss": 4.2248,
+      "step": 803
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009861498512645532,
+      "loss": 4.2123,
+      "step": 804
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009861013530656955,
+      "loss": 4.1502,
+      "step": 805
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009860527713003798,
+      "loss": 4.1014,
+      "step": 806
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009860041059769579,
+      "loss": 4.1962,
+      "step": 807
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.000985955357103796,
+      "loss": 4.21,
+      "step": 808
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009859065246892744,
+      "loss": 4.0482,
+      "step": 809
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.000985857608741788,
+      "loss": 4.1745,
+      "step": 810
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009858086092697463,
+      "loss": 4.0492,
+      "step": 811
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009857595262815727,
+      "loss": 4.1776,
+      "step": 812
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.000985710359785705,
+      "loss": 4.2113,
+      "step": 813
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009856611097905955,
+      "loss": 4.1161,
+      "step": 814
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009856117763047114,
+      "loss": 4.1993,
+      "step": 815
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009855623593365331,
+      "loss": 4.2252,
+      "step": 816
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.000985512858894556,
+      "loss": 4.0882,
+      "step": 817
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009854632749872899,
+      "loss": 4.1377,
+      "step": 818
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009854136076232587,
+      "loss": 4.1129,
+      "step": 819
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009853638568110012,
+      "loss": 4.1814,
+      "step": 820
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009853140225590698,
+      "loss": 4.23,
+      "step": 821
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009852641048760317,
+      "loss": 4.1589,
+      "step": 822
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009852141037704682,
+      "loss": 4.1674,
+      "step": 823
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009851640192509753,
+      "loss": 4.032,
+      "step": 824
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009851138513261631,
+      "loss": 4.1803,
+      "step": 825
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009850636000046558,
+      "loss": 4.1043,
+      "step": 826
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009850132652950922,
+      "loss": 4.0929,
+      "step": 827
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009849628472061258,
+      "loss": 4.1048,
+      "step": 828
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009849123457464236,
+      "loss": 4.1431,
+      "step": 829
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009848617609246677,
+      "loss": 4.1578,
+      "step": 830
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009848110927495538,
+      "loss": 4.1313,
+      "step": 831
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.000984760341229793,
+      "loss": 4.1772,
+      "step": 832
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009847095063741096,
+      "loss": 4.1324,
+      "step": 833
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.000984658588191243,
+      "loss": 4.1438,
+      "step": 834
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009846075866899462,
+      "loss": 4.1436,
+      "step": 835
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009845565018789873,
+      "loss": 4.2032,
+      "step": 836
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009845053337671481,
+      "loss": 4.1832,
+      "step": 837
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009844540823632252,
+      "loss": 4.155,
+      "step": 838
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009844027476760295,
+      "loss": 4.1197,
+      "step": 839
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009843513297143856,
+      "loss": 4.1013,
+      "step": 840
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009842998284871332,
+      "loss": 4.1176,
+      "step": 841
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009842482440031255,
+      "loss": 4.1632,
+      "step": 842
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009841965762712312,
+      "loss": 4.0459,
+      "step": 843
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009841448253003317,
+      "loss": 4.1838,
+      "step": 844
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009840929910993244,
+      "loss": 4.1358,
+      "step": 845
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009840410736771197,
+      "loss": 4.0319,
+      "step": 846
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.000983989073042643,
+      "loss": 4.1447,
+      "step": 847
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009839369892048336,
+      "loss": 4.2852,
+      "step": 848
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009838848221726455,
+      "loss": 4.0786,
+      "step": 849
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.000983832571955047,
+      "loss": 4.2737,
+      "step": 850
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00098378023856102,
+      "loss": 4.0909,
+      "step": 851
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009837278219995617,
+      "loss": 4.2167,
+      "step": 852
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.000983675322279683,
+      "loss": 4.196,
+      "step": 853
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009836227394104092,
+      "loss": 4.0324,
+      "step": 854
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009835700734007796,
+      "loss": 4.1438,
+      "step": 855
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009835173242598485,
+      "loss": 4.2194,
+      "step": 856
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009834644919966841,
+      "loss": 4.0669,
+      "step": 857
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009834115766203686,
+      "loss": 4.1216,
+      "step": 858
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.000983358578139999,
+      "loss": 4.103,
+      "step": 859
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.000983305496564686,
+      "loss": 4.1786,
+      "step": 860
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009832523319035554,
+      "loss": 4.2026,
+      "step": 861
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009831990841657463,
+      "loss": 4.0152,
+      "step": 862
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009831457533604133,
+      "loss": 4.1829,
+      "step": 863
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.000983092339496724,
+      "loss": 4.1268,
+      "step": 864
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.000983038842583861,
+      "loss": 4.2633,
+      "step": 865
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009829852626310211,
+      "loss": 4.0287,
+      "step": 866
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009829315996474155,
+      "loss": 4.0225,
+      "step": 867
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009828778536422691,
+      "loss": 4.1264,
+      "step": 868
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009828240246248217,
+      "loss": 4.2395,
+      "step": 869
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009827701126043271,
+      "loss": 4.2013,
+      "step": 870
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009827161175900534,
+      "loss": 4.0805,
+      "step": 871
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009826620395912832,
+      "loss": 4.1354,
+      "step": 872
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009826078786173125,
+      "loss": 4.0689,
+      "step": 873
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.000982553634677453,
+      "loss": 3.9636,
+      "step": 874
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.000982499307781029,
+      "loss": 4.1864,
+      "step": 875
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009824448979373806,
+      "loss": 3.9714,
+      "step": 876
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009823904051558612,
+      "loss": 4.2349,
+      "step": 877
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009823358294458388,
+      "loss": 4.2455,
+      "step": 878
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009822811708166959,
+      "loss": 4.2024,
+      "step": 879
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009822264292778282,
+      "loss": 4.1579,
+      "step": 880
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009821716048386473,
+      "loss": 3.9942,
+      "step": 881
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009821166975085773,
+      "loss": 4.1935,
+      "step": 882
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009820617072970582,
+      "loss": 4.0318,
+      "step": 883
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009820066342135427,
+      "loss": 4.2514,
+      "step": 884
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009819514782674992,
+      "loss": 4.1634,
+      "step": 885
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009818962394684092,
+      "loss": 4.1175,
+      "step": 886
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.000981840917825769,
+      "loss": 4.1681,
+      "step": 887
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009817855133490891,
+      "loss": 4.2029,
+      "step": 888
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.000981730026047894,
+      "loss": 4.1404,
+      "step": 889
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009816744559317228,
+      "loss": 4.0324,
+      "step": 890
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009816188030101285,
+      "loss": 4.0386,
+      "step": 891
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009815630672926788,
+      "loss": 4.0485,
+      "step": 892
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009815072487889549,
+      "loss": 4.2157,
+      "step": 893
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.000981451347508553,
+      "loss": 4.0065,
+      "step": 894
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009813953634610826,
+      "loss": 4.1506,
+      "step": 895
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009813392966561687,
+      "loss": 4.03,
+      "step": 896
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009812831471034495,
+      "loss": 4.1681,
+      "step": 897
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009812269148125777,
+      "loss": 4.0408,
+      "step": 898
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009811705997932205,
+      "loss": 4.0587,
+      "step": 899
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.000981114202055059,
+      "loss": 4.1181,
+      "step": 900
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009810577216077885,
+      "loss": 4.2013,
+      "step": 901
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.000981001158461119,
+      "loss": 4.1615,
+      "step": 902
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009809445126247739,
+      "loss": 4.1013,
+      "step": 903
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009808877841084914,
+      "loss": 4.0467,
+      "step": 904
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.000980830972922024,
+      "loss": 3.9904,
+      "step": 905
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009807740790751382,
+      "loss": 4.0013,
+      "step": 906
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009807171025776145,
+      "loss": 4.1784,
+      "step": 907
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009806600434392478,
+      "loss": 4.1889,
+      "step": 908
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009806029016698474,
+      "loss": 4.1934,
+      "step": 909
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009805456772792368,
+      "loss": 4.0725,
+      "step": 910
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009804883702772531,
+      "loss": 4.1802,
+      "step": 911
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009804309806737482,
+      "loss": 4.07,
+      "step": 912
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009803735084785883,
+      "loss": 4.0809,
+      "step": 913
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.000980315953701653,
+      "loss": 4.1719,
+      "step": 914
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009802583163528373,
+      "loss": 4.089,
+      "step": 915
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009802005964420493,
+      "loss": 4.1511,
+      "step": 916
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009801427939792118,
+      "loss": 4.1394,
+      "step": 917
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009800849089742615,
+      "loss": 4.1518,
+      "step": 918
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009800269414371502,
+      "loss": 4.165,
+      "step": 919
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.000979968891377842,
+      "loss": 4.3027,
+      "step": 920
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009799107588063178,
+      "loss": 4.063,
+      "step": 921
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009798525437325703,
+      "loss": 4.256,
+      "step": 922
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009797942461666075,
+      "loss": 4.1155,
+      "step": 923
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009797358661184517,
+      "loss": 4.1306,
+      "step": 924
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009796774035981388,
+      "loss": 4.0449,
+      "step": 925
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009796188586157193,
+      "loss": 4.1766,
+      "step": 926
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.000979560231181258,
+      "loss": 4.0182,
+      "step": 927
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009795015213048333,
+      "loss": 4.1126,
+      "step": 928
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009794427289965381,
+      "loss": 3.9529,
+      "step": 929
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009793838542664797,
+      "loss": 4.0495,
+      "step": 930
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009793248971247791,
+      "loss": 4.0549,
+      "step": 931
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009792658575815724,
+      "loss": 4.1239,
+      "step": 932
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009792067356470082,
+      "loss": 4.1902,
+      "step": 933
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009791475313312508,
+      "loss": 4.1224,
+      "step": 934
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.000979088244644478,
+      "loss": 4.1289,
+      "step": 935
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.000979028875596882,
+      "loss": 4.1422,
+      "step": 936
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009789694241986687,
+      "loss": 3.9274,
+      "step": 937
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009789098904600588,
+      "loss": 3.9911,
+      "step": 938
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009788502743912867,
+      "loss": 4.1116,
+      "step": 939
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.000978790576002601,
+      "loss": 4.0463,
+      "step": 940
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.000978730795304265,
+      "loss": 4.0868,
+      "step": 941
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009786709323065552,
+      "loss": 4.0992,
+      "step": 942
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009786109870197628,
+      "loss": 4.2598,
+      "step": 943
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009785509594541933,
+      "loss": 4.0948,
+      "step": 944
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.000978490849620166,
+      "loss": 4.1486,
+      "step": 945
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009784306575280144,
+      "loss": 4.2188,
+      "step": 946
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009783703831880866,
+      "loss": 4.0695,
+      "step": 947
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.000978310026610744,
+      "loss": 4.1557,
+      "step": 948
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009782495878063629,
+      "loss": 4.0066,
+      "step": 949
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.000978189066785333,
+      "loss": 4.0985,
+      "step": 950
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009781284635580591,
+      "loss": 4.195,
+      "step": 951
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009780677781349595,
+      "loss": 4.2982,
+      "step": 952
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009780070105264666,
+      "loss": 4.0992,
+      "step": 953
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009779461607430267,
+      "loss": 4.2505,
+      "step": 954
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009778852287951013,
+      "loss": 4.1149,
+      "step": 955
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.000977824214693165,
+      "loss": 4.1279,
+      "step": 956
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009777631184477065,
+      "loss": 4.0033,
+      "step": 957
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009777019400692291,
+      "loss": 4.1167,
+      "step": 958
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009776406795682505,
+      "loss": 4.1324,
+      "step": 959
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009775793369553017,
+      "loss": 4.0275,
+      "step": 960
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009775179122409284,
+      "loss": 4.0508,
+      "step": 961
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00097745640543569,
+      "loss": 3.9927,
+      "step": 962
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009773948165501605,
+      "loss": 3.9659,
+      "step": 963
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009773331455949275,
+      "loss": 4.0269,
+      "step": 964
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.000977271392580593,
+      "loss": 4.2054,
+      "step": 965
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009772095575177732,
+      "loss": 4.0671,
+      "step": 966
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009771476404170983,
+      "loss": 4.04,
+      "step": 967
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009770856412892122,
+      "loss": 4.1317,
+      "step": 968
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009770235601447737,
+      "loss": 4.2806,
+      "step": 969
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009769613969944548,
+      "loss": 4.0912,
+      "step": 970
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009768991518489425,
+      "loss": 4.1326,
+      "step": 971
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009768368247189375,
+      "loss": 4.1433,
+      "step": 972
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009767744156151542,
+      "loss": 4.0329,
+      "step": 973
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009767119245483215,
+      "loss": 4.1192,
+      "step": 974
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009766493515291827,
+      "loss": 4.2426,
+      "step": 975
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009765866965684945,
+      "loss": 4.0268,
+      "step": 976
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009765239596770279,
+      "loss": 4.0795,
+      "step": 977
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009764611408655685,
+      "loss": 4.2039,
+      "step": 978
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009763982401449153,
+      "loss": 4.1461,
+      "step": 979
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009763352575258817,
+      "loss": 4.1269,
+      "step": 980
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009762721930192952,
+      "loss": 4.1214,
+      "step": 981
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009762090466359974,
+      "loss": 3.9959,
+      "step": 982
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009761458183868437,
+      "loss": 4.1424,
+      "step": 983
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009760825082827039,
+      "loss": 4.0756,
+      "step": 984
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009760191163344617,
+      "loss": 4.0161,
+      "step": 985
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.000975955642553015,
+      "loss": 4.0205,
+      "step": 986
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009758920869492754,
+      "loss": 4.1251,
+      "step": 987
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009758284495341692,
+      "loss": 4.1379,
+      "step": 988
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009757647303186361,
+      "loss": 4.1193,
+      "step": 989
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009757009293136303,
+      "loss": 4.0312,
+      "step": 990
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00097563704653012,
+      "loss": 4.1608,
+      "step": 991
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009755730819790876,
+      "loss": 4.203,
+      "step": 992
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009755090356715288,
+      "loss": 4.1225,
+      "step": 993
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009754449076184543,
+      "loss": 4.0162,
+      "step": 994
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009753806978308883,
+      "loss": 4.0869,
+      "step": 995
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009753164063198693,
+      "loss": 4.057,
+      "step": 996
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009752520330964496,
+      "loss": 4.0544,
+      "step": 997
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.000975187578171696,
+      "loss": 3.9998,
+      "step": 998
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009751230415566887,
+      "loss": 4.3571,
+      "step": 999
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009750584232625226,
+      "loss": 4.1817,
+      "step": 1000
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009749937233003061,
+      "loss": 3.9591,
+      "step": 1001
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009749289416811621,
+      "loss": 4.0477,
+      "step": 1002
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.000974864078416227,
+      "loss": 4.232,
+      "step": 1003
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009747991335166519,
+      "loss": 4.094,
+      "step": 1004
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009747341069936013,
+      "loss": 4.1215,
+      "step": 1005
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.000974668998858254,
+      "loss": 4.0932,
+      "step": 1006
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009746038091218031,
+      "loss": 4.09,
+      "step": 1007
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009745385377954555,
+      "loss": 4.1226,
+      "step": 1008
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009744731848904315,
+      "loss": 4.0264,
+      "step": 1009
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009744077504179669,
+      "loss": 3.9806,
+      "step": 1010
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009743422343893098,
+      "loss": 4.0412,
+      "step": 1011
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009742766368157239,
+      "loss": 4.0336,
+      "step": 1012
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009742109577084856,
+      "loss": 4.1596,
+      "step": 1013
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009741451970788862,
+      "loss": 3.9753,
+      "step": 1014
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009740793549382307,
+      "loss": 4.2997,
+      "step": 1015
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009740134312978382,
+      "loss": 4.0126,
+      "step": 1016
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009739474261690415,
+      "loss": 4.0334,
+      "step": 1017
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009738813395631878,
+      "loss": 4.1232,
+      "step": 1018
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009738151714916381,
+      "loss": 3.9615,
+      "step": 1019
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009737489219657675,
+      "loss": 3.9855,
+      "step": 1020
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.000973682590996965,
+      "loss": 4.0515,
+      "step": 1021
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009736161785966337,
+      "loss": 4.1659,
+      "step": 1022
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009735496847761908,
+      "loss": 4.0647,
+      "step": 1023
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009734831095470671,
+      "loss": 4.0405,
+      "step": 1024
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009734164529207079,
+      "loss": 4.0625,
+      "step": 1025
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009733497149085719,
+      "loss": 4.0419,
+      "step": 1026
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009732828955221327,
+      "loss": 4.0858,
+      "step": 1027
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009732159947728767,
+      "loss": 4.0797,
+      "step": 1028
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009731490126723052,
+      "loss": 4.0759,
+      "step": 1029
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009730819492319335,
+      "loss": 4.1115,
+      "step": 1030
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009730148044632901,
+      "loss": 4.0034,
+      "step": 1031
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009729475783779181,
+      "loss": 4.256,
+      "step": 1032
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009728802709873747,
+      "loss": 3.9631,
+      "step": 1033
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009728128823032305,
+      "loss": 4.1181,
+      "step": 1034
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009727454123370705,
+      "loss": 4.0626,
+      "step": 1035
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009726778611004935,
+      "loss": 4.1457,
+      "step": 1036
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009726102286051127,
+      "loss": 4.0448,
+      "step": 1037
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009725425148625542,
+      "loss": 4.0471,
+      "step": 1038
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009724747198844595,
+      "loss": 4.134,
+      "step": 1039
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009724068436824831,
+      "loss": 4.1982,
+      "step": 1040
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009723388862682936,
+      "loss": 4.1057,
+      "step": 1041
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009722708476535736,
+      "loss": 4.1778,
+      "step": 1042
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009722027278500201,
+      "loss": 4.1788,
+      "step": 1043
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009721345268693432,
+      "loss": 4.1621,
+      "step": 1044
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009720662447232677,
+      "loss": 4.1622,
+      "step": 1045
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009719978814235322,
+      "loss": 4.2716,
+      "step": 1046
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.000971929436981889,
+      "loss": 4.0947,
+      "step": 1047
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009718609114101045,
+      "loss": 4.0,
+      "step": 1048
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009717923047199591,
+      "loss": 4.0383,
+      "step": 1049
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009717236169232468,
+      "loss": 4.1521,
+      "step": 1050
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009716548480317763,
+      "loss": 4.0647,
+      "step": 1051
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009715859980573694,
+      "loss": 4.1425,
+      "step": 1052
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009715170670118624,
+      "loss": 3.8636,
+      "step": 1053
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009714480549071052,
+      "loss": 4.0656,
+      "step": 1054
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.000971378961754962,
+      "loss": 4.058,
+      "step": 1055
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009713097875673105,
+      "loss": 4.1017,
+      "step": 1056
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009712405323560427,
+      "loss": 4.105,
+      "step": 1057
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009711711961330643,
+      "loss": 4.1439,
+      "step": 1058
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009711017789102948,
+      "loss": 4.0925,
+      "step": 1059
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009710322806996682,
+      "loss": 4.1315,
+      "step": 1060
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009709627015131319,
+      "loss": 4.17,
+      "step": 1061
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009708930413626473,
+      "loss": 4.1502,
+      "step": 1062
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009708233002601898,
+      "loss": 4.0575,
+      "step": 1063
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009707534782177487,
+      "loss": 4.1062,
+      "step": 1064
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009706835752473273,
+      "loss": 4.2034,
+      "step": 1065
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009706135913609426,
+      "loss": 4.0238,
+      "step": 1066
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009705435265706258,
+      "loss": 4.0687,
+      "step": 1067
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009704733808884218,
+      "loss": 4.2064,
+      "step": 1068
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009704031543263893,
+      "loss": 3.9334,
+      "step": 1069
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009703328468966015,
+      "loss": 4.0958,
+      "step": 1070
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009702624586111445,
+      "loss": 3.964,
+      "step": 1071
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009701919894821192,
+      "loss": 4.1434,
+      "step": 1072
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009701214395216399,
+      "loss": 4.091,
+      "step": 1073
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009700508087418351,
+      "loss": 4.0919,
+      "step": 1074
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009699800971548469,
+      "loss": 4.0635,
+      "step": 1075
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009699093047728316,
+      "loss": 4.0618,
+      "step": 1076
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009698384316079591,
+      "loss": 4.104,
+      "step": 1077
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009697674776724135,
+      "loss": 3.9966,
+      "step": 1078
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009696964429783923,
+      "loss": 4.1471,
+      "step": 1079
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009696253275381074,
+      "loss": 4.1298,
+      "step": 1080
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009695541313637843,
+      "loss": 4.0441,
+      "step": 1081
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009694828544676626,
+      "loss": 4.1918,
+      "step": 1082
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009694114968619954,
+      "loss": 4.0242,
+      "step": 1083
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009693400585590501,
+      "loss": 4.0071,
+      "step": 1084
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009692685395711076,
+      "loss": 4.0279,
+      "step": 1085
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.000969196939910463,
+      "loss": 4.1848,
+      "step": 1086
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.000969125259589425,
+      "loss": 4.0549,
+      "step": 1087
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009690534986203164,
+      "loss": 4.0651,
+      "step": 1088
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009689816570154734,
+      "loss": 3.9695,
+      "step": 1089
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009689097347872469,
+      "loss": 3.9525,
+      "step": 1090
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009688377319480009,
+      "loss": 3.9434,
+      "step": 1091
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009687656485101132,
+      "loss": 4.024,
+      "step": 1092
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009686934844859765,
+      "loss": 4.0272,
+      "step": 1093
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009686212398879962,
+      "loss": 4.2582,
+      "step": 1094
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.000968548914728592,
+      "loss": 4.0902,
+      "step": 1095
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009684765090201974,
+      "loss": 4.0441,
+      "step": 1096
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009684040227752599,
+      "loss": 4.1233,
+      "step": 1097
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009683314560062408,
+      "loss": 3.8987,
+      "step": 1098
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009682588087256148,
+      "loss": 4.1683,
+      "step": 1099
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009681860809458713,
+      "loss": 4.0463,
+      "step": 1100
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009681132726795128,
+      "loss": 4.0684,
+      "step": 1101
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009680403839390557,
+      "loss": 4.0144,
+      "step": 1102
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009679674147370308,
+      "loss": 4.1296,
+      "step": 1103
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009678943650859819,
+      "loss": 4.0719,
+      "step": 1104
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009678212349984676,
+      "loss": 4.008,
+      "step": 1105
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009677480244870596,
+      "loss": 4.017,
+      "step": 1106
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009676747335643435,
+      "loss": 4.0732,
+      "step": 1107
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009676013622429187,
+      "loss": 4.0904,
+      "step": 1108
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009675279105353992,
+      "loss": 4.1163,
+      "step": 1109
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009674543784544117,
+      "loss": 4.1633,
+      "step": 1110
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009673807660125973,
+      "loss": 4.0908,
+      "step": 1111
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009673070732226108,
+      "loss": 3.9621,
+      "step": 1112
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.000967233300097121,
+      "loss": 4.0295,
+      "step": 1113
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009671594466488104,
+      "loss": 4.065,
+      "step": 1114
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.000967085512890375,
+      "loss": 4.1317,
+      "step": 1115
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.000967011498834525,
+      "loss": 3.9693,
+      "step": 1116
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009669374044939845,
+      "loss": 4.031,
+      "step": 1117
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009668632298814908,
+      "loss": 3.9767,
+      "step": 1118
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009667889750097956,
+      "loss": 3.9814,
+      "step": 1119
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.000966714639891664,
+      "loss": 4.2,
+      "step": 1120
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009666402245398754,
+      "loss": 4.0724,
+      "step": 1121
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009665657289672222,
+      "loss": 4.0009,
+      "step": 1122
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009664911531865114,
+      "loss": 4.1494,
+      "step": 1123
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009664164972105633,
+      "loss": 4.0471,
+      "step": 1124
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009663417610522123,
+      "loss": 4.0942,
+      "step": 1125
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009662669447243061,
+      "loss": 4.0696,
+      "step": 1126
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009661920482397068,
+      "loss": 3.9861,
+      "step": 1127
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009661170716112897,
+      "loss": 4.1162,
+      "step": 1128
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009660420148519444,
+      "loss": 4.0994,
+      "step": 1129
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.000965966877974574,
+      "loss": 4.1327,
+      "step": 1130
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009658916609920951,
+      "loss": 4.1561,
+      "step": 1131
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009658163639174389,
+      "loss": 4.0713,
+      "step": 1132
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009657409867635492,
+      "loss": 3.9842,
+      "step": 1133
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009656655295433848,
+      "loss": 4.1157,
+      "step": 1134
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009655899922699174,
+      "loss": 4.165,
+      "step": 1135
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009655143749561325,
+      "loss": 4.0365,
+      "step": 1136
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009654386776150302,
+      "loss": 4.0221,
+      "step": 1137
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009653629002596231,
+      "loss": 3.9876,
+      "step": 1138
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009652870429029386,
+      "loss": 4.042,
+      "step": 1139
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009652111055580174,
+      "loss": 4.1545,
+      "step": 1140
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009651350882379139,
+      "loss": 3.9491,
+      "step": 1141
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009650589909556964,
+      "loss": 4.2504,
+      "step": 1142
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.000964982813724447,
+      "loss": 4.0123,
+      "step": 1143
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009649065565572614,
+      "loss": 4.217,
+      "step": 1144
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009648302194672489,
+      "loss": 4.1235,
+      "step": 1145
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009647538024675331,
+      "loss": 4.1791,
+      "step": 1146
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009646773055712506,
+      "loss": 4.0154,
+      "step": 1147
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009646007287915523,
+      "loss": 4.0947,
+      "step": 1148
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009645240721416028,
+      "loss": 4.2082,
+      "step": 1149
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00096444733563458,
+      "loss": 4.2883,
+      "step": 1150
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009643705192836758,
+      "loss": 4.082,
+      "step": 1151
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009642936231020958,
+      "loss": 4.2579,
+      "step": 1152
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009642166471030596,
+      "loss": 4.0718,
+      "step": 1153
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009641395912998,
+      "loss": 4.2002,
+      "step": 1154
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009640624557055639,
+      "loss": 4.1518,
+      "step": 1155
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009639852403336118,
+      "loss": 4.1189,
+      "step": 1156
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009639079451972179,
+      "loss": 4.1194,
+      "step": 1157
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009638305703096701,
+      "loss": 4.1645,
+      "step": 1158
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009637531156842702,
+      "loss": 4.0545,
+      "step": 1159
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009636755813343334,
+      "loss": 4.2036,
+      "step": 1160
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009635979672731887,
+      "loss": 4.1876,
+      "step": 1161
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.000963520273514179,
+      "loss": 4.1012,
+      "step": 1162
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009634425000706607,
+      "loss": 4.1601,
+      "step": 1163
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009633646469560039,
+      "loss": 4.0087,
+      "step": 1164
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009632867141835926,
+      "loss": 4.0845,
+      "step": 1165
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.000963208701766824,
+      "loss": 4.0637,
+      "step": 1166
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009631306097191099,
+      "loss": 3.9741,
+      "step": 1167
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009630524380538748,
+      "loss": 3.983,
+      "step": 1168
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009629741867845573,
+      "loss": 4.1033,
+      "step": 1169
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00096289585592461,
+      "loss": 3.9737,
+      "step": 1170
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009628174454874987,
+      "loss": 4.1085,
+      "step": 1171
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009627389554867031,
+      "loss": 4.1167,
+      "step": 1172
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009626603859357165,
+      "loss": 4.3003,
+      "step": 1173
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009625817368480458,
+      "loss": 4.0287,
+      "step": 1174
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.000962503008237212,
+      "loss": 4.0754,
+      "step": 1175
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009624242001167493,
+      "loss": 4.1397,
+      "step": 1176
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009623453125002056,
+      "loss": 3.9751,
+      "step": 1177
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009622663454011427,
+      "loss": 4.1285,
+      "step": 1178
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009621872988331362,
+      "loss": 4.1195,
+      "step": 1179
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009621081728097747,
+      "loss": 3.9238,
+      "step": 1180
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.000962028967344661,
+      "loss": 3.9871,
+      "step": 1181
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009619496824514117,
+      "loss": 3.9829,
+      "step": 1182
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009618703181436566,
+      "loss": 4.2251,
+      "step": 1183
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009617908744350393,
+      "loss": 4.0413,
+      "step": 1184
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.000961711351339217,
+      "loss": 4.1047,
+      "step": 1185
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009616317488698609,
+      "loss": 4.0341,
+      "step": 1186
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009615520670406554,
+      "loss": 4.0053,
+      "step": 1187
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.000961472305865299,
+      "loss": 4.0393,
+      "step": 1188
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009613924653575032,
+      "loss": 4.0055,
+      "step": 1189
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009613125455309939,
+      "loss": 3.9678,
+      "step": 1190
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009612325463995099,
+      "loss": 4.0906,
+      "step": 1191
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.000961152467976804,
+      "loss": 4.0547,
+      "step": 1192
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009610723102766428,
+      "loss": 3.9525,
+      "step": 1193
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009609920733128064,
+      "loss": 3.958,
+      "step": 1194
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.000960911757099088,
+      "loss": 4.1006,
+      "step": 1195
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009608313616492955,
+      "loss": 4.0431,
+      "step": 1196
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009607508869772495,
+      "loss": 3.9155,
+      "step": 1197
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009606703330967843,
+      "loss": 4.0901,
+      "step": 1198
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009605897000217485,
+      "loss": 3.9418,
+      "step": 1199
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009605089877660036,
+      "loss": 4.1932,
+      "step": 1200
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.000960428196343425,
+      "loss": 4.1826,
+      "step": 1201
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009603473257679018,
+      "loss": 4.0193,
+      "step": 1202
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009602663760533363,
+      "loss": 4.0131,
+      "step": 1203
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.000960185347213645,
+      "loss": 4.1226,
+      "step": 1204
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009601042392627576,
+      "loss": 3.9713,
+      "step": 1205
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009600230522146176,
+      "loss": 4.0104,
+      "step": 1206
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009599417860831817,
+      "loss": 4.0973,
+      "step": 1207
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009598604408824207,
+      "loss": 4.1125,
+      "step": 1208
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009597790166263189,
+      "loss": 4.0144,
+      "step": 1209
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009596975133288738,
+      "loss": 4.2263,
+      "step": 1210
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009596159310040969,
+      "loss": 4.0152,
+      "step": 1211
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009595342696660132,
+      "loss": 4.1508,
+      "step": 1212
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.000959452529328661,
+      "loss": 4.185,
+      "step": 1213
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009593707100060926,
+      "loss": 4.0531,
+      "step": 1214
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.000959288811712374,
+      "loss": 4.0043,
+      "step": 1215
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009592068344615837,
+      "loss": 4.0415,
+      "step": 1216
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009591247782678152,
+      "loss": 4.012,
+      "step": 1217
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009590426431451746,
+      "loss": 4.1145,
+      "step": 1218
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.000958960429107782,
+      "loss": 4.0815,
+      "step": 1219
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009588781361697709,
+      "loss": 3.9523,
+      "step": 1220
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009587957643452885,
+      "loss": 4.1376,
+      "step": 1221
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009587133136484953,
+      "loss": 4.1164,
+      "step": 1222
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009586307840935657,
+      "loss": 4.0166,
+      "step": 1223
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009585481756946874,
+      "loss": 4.1362,
+      "step": 1224
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009584654884660618,
+      "loss": 3.9452,
+      "step": 1225
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009583827224219035,
+      "loss": 3.9666,
+      "step": 1226
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009582998775764413,
+      "loss": 4.2034,
+      "step": 1227
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009582169539439171,
+      "loss": 3.9628,
+      "step": 1228
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009581339515385865,
+      "loss": 4.1143,
+      "step": 1229
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009580508703747184,
+      "loss": 4.1015,
+      "step": 1230
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009579677104665956,
+      "loss": 4.1109,
+      "step": 1231
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.000957884471828514,
+      "loss": 3.9662,
+      "step": 1232
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009578011544747835,
+      "loss": 4.0822,
+      "step": 1233
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009577177584197274,
+      "loss": 4.0519,
+      "step": 1234
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009576342836776821,
+      "loss": 4.15,
+      "step": 1235
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009575507302629982,
+      "loss": 4.1819,
+      "step": 1236
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009574670981900393,
+      "loss": 3.9392,
+      "step": 1237
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009573833874731829,
+      "loss": 3.9558,
+      "step": 1238
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009572995981268198,
+      "loss": 4.1319,
+      "step": 1239
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009572157301653542,
+      "loss": 4.126,
+      "step": 1240
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009571317836032041,
+      "loss": 3.9894,
+      "step": 1241
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009570477584548007,
+      "loss": 3.8905,
+      "step": 1242
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009569636547345894,
+      "loss": 3.9959,
+      "step": 1243
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009568794724570281,
+      "loss": 4.0937,
+      "step": 1244
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009567952116365889,
+      "loss": 4.0076,
+      "step": 1245
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009567108722877572,
+      "loss": 4.0984,
+      "step": 1246
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009566264544250318,
+      "loss": 4.0906,
+      "step": 1247
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009565419580629253,
+      "loss": 4.0505,
+      "step": 1248
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009564573832159637,
+      "loss": 3.9967,
+      "step": 1249
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009563727298986859,
+      "loss": 4.1327,
+      "step": 1250
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009562879981256454,
+      "loss": 4.0877,
+      "step": 1251
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009562031879114081,
+      "loss": 4.1298,
+      "step": 1252
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009561182992705541,
+      "loss": 3.9737,
+      "step": 1253
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009560333322176767,
+      "loss": 3.7801,
+      "step": 1254
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009559482867673825,
+      "loss": 4.063,
+      "step": 1255
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009558631629342922,
+      "loss": 4.0237,
+      "step": 1256
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009557779607330393,
+      "loss": 4.0983,
+      "step": 1257
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009556926801782713,
+      "loss": 4.1285,
+      "step": 1258
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009556073212846485,
+      "loss": 3.8953,
+      "step": 1259
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009555218840668454,
+      "loss": 4.0212,
+      "step": 1260
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009554363685395495,
+      "loss": 4.0194,
+      "step": 1261
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009553507747174622,
+      "loss": 4.0606,
+      "step": 1262
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009552651026152976,
+      "loss": 4.0986,
+      "step": 1263
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009551793522477841,
+      "loss": 3.9767,
+      "step": 1264
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009550935236296629,
+      "loss": 4.2397,
+      "step": 1265
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009550076167756892,
+      "loss": 3.9893,
+      "step": 1266
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009549216317006313,
+      "loss": 4.0515,
+      "step": 1267
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009548355684192711,
+      "loss": 4.1027,
+      "step": 1268
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009547494269464036,
+      "loss": 4.2617,
+      "step": 1269
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009546632072968378,
+      "loss": 3.9796,
+      "step": 1270
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009545769094853958,
+      "loss": 4.0893,
+      "step": 1271
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.000954490533526913,
+      "loss": 4.0477,
+      "step": 1272
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009544040794362388,
+      "loss": 4.0954,
+      "step": 1273
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009543175472282352,
+      "loss": 4.1777,
+      "step": 1274
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009542309369177784,
+      "loss": 4.0593,
+      "step": 1275
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009541442485197577,
+      "loss": 4.0349,
+      "step": 1276
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009540574820490757,
+      "loss": 3.9666,
+      "step": 1277
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009539706375206487,
+      "loss": 4.0738,
+      "step": 1278
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009538837149494064,
+      "loss": 3.9808,
+      "step": 1279
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009537967143502915,
+      "loss": 4.0622,
+      "step": 1280
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009537096357382604,
+      "loss": 4.138,
+      "step": 1281
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009536224791282833,
+      "loss": 4.139,
+      "step": 1282
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.000953535244535343,
+      "loss": 4.1675,
+      "step": 1283
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009534479319744365,
+      "loss": 4.1437,
+      "step": 1284
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009533605414605735,
+      "loss": 4.0769,
+      "step": 1285
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009532730730087778,
+      "loss": 4.0425,
+      "step": 1286
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009531855266340859,
+      "loss": 3.9394,
+      "step": 1287
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009530979023515483,
+      "loss": 4.0109,
+      "step": 1288
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009530102001762284,
+      "loss": 4.1523,
+      "step": 1289
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009529224201232034,
+      "loss": 4.1154,
+      "step": 1290
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009528345622075635,
+      "loss": 4.0405,
+      "step": 1291
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.000952746626444413,
+      "loss": 3.8818,
+      "step": 1292
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009526586128488685,
+      "loss": 4.0882,
+      "step": 1293
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009525705214360609,
+      "loss": 4.1219,
+      "step": 1294
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009524823522211339,
+      "loss": 4.0244,
+      "step": 1295
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.000952394105219245,
+      "loss": 4.1541,
+      "step": 1296
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009523057804455648,
+      "loss": 4.1028,
+      "step": 1297
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009522173779152773,
+      "loss": 3.9829,
+      "step": 1298
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00095212889764358,
+      "loss": 4.1809,
+      "step": 1299
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009520403396456838,
+      "loss": 3.9643,
+      "step": 1300
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009519517039368126,
+      "loss": 4.1378,
+      "step": 1301
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.000951862990532204,
+      "loss": 4.0475,
+      "step": 1302
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.000951774199447109,
+      "loss": 4.109,
+      "step": 1303
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009516853306967915,
+      "loss": 4.1004,
+      "step": 1304
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009515963842965294,
+      "loss": 3.9608,
+      "step": 1305
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009515073602616135,
+      "loss": 3.9302,
+      "step": 1306
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.000951418258607348,
+      "loss": 4.0643,
+      "step": 1307
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009513290793490505,
+      "loss": 4.0533,
+      "step": 1308
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009512398225020522,
+      "loss": 3.9447,
+      "step": 1309
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.000951150488081697,
+      "loss": 3.9854,
+      "step": 1310
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009510610761033427,
+      "loss": 3.9939,
+      "step": 1311
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009509715865823604,
+      "loss": 4.0158,
+      "step": 1312
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009508820195341342,
+      "loss": 4.0477,
+      "step": 1313
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009507923749740618,
+      "loss": 3.9354,
+      "step": 1314
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.000950702652917554,
+      "loss": 3.9445,
+      "step": 1315
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009506128533800354,
+      "loss": 4.0178,
+      "step": 1316
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009505229763769431,
+      "loss": 4.0567,
+      "step": 1317
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009504330219237283,
+      "loss": 4.1455,
+      "step": 1318
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009503429900358553,
+      "loss": 4.1352,
+      "step": 1319
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009502528807288014,
+      "loss": 4.0662,
+      "step": 1320
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009501626940180575,
+      "loss": 4.1505,
+      "step": 1321
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009500724299191278,
+      "loss": 3.952,
+      "step": 1322
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009499820884475296,
+      "loss": 3.8814,
+      "step": 1323
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.000949891669618794,
+      "loss": 4.0232,
+      "step": 1324
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009498011734484646,
+      "loss": 4.1034,
+      "step": 1325
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009497105999520989,
+      "loss": 4.061,
+      "step": 1326
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009496199491452679,
+      "loss": 4.0495,
+      "step": 1327
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009495292210435548,
+      "loss": 4.0728,
+      "step": 1328
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009494384156625575,
+      "loss": 4.1667,
+      "step": 1329
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.000949347533017886,
+      "loss": 4.0916,
+      "step": 1330
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009492565731251644,
+      "loss": 4.1544,
+      "step": 1331
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009491655360000297,
+      "loss": 4.0987,
+      "step": 1332
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009490744216581322,
+      "loss": 4.0323,
+      "step": 1333
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009489832301151353,
+      "loss": 3.993,
+      "step": 1334
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009488919613867162,
+      "loss": 3.9442,
+      "step": 1335
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.000948800615488565,
+      "loss": 4.0641,
+      "step": 1336
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009487091924363851,
+      "loss": 4.0724,
+      "step": 1337
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009486176922458929,
+      "loss": 4.0664,
+      "step": 1338
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009485261149328188,
+      "loss": 3.9276,
+      "step": 1339
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009484344605129056,
+      "loss": 4.0093,
+      "step": 1340
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009483427290019101,
+      "loss": 4.1796,
+      "step": 1341
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009482509204156018,
+      "loss": 4.0086,
+      "step": 1342
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009481590347697637,
+      "loss": 4.0832,
+      "step": 1343
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009480670720801921,
+      "loss": 3.9351,
+      "step": 1344
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009479750323626962,
+      "loss": 4.0834,
+      "step": 1345
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.000947882915633099,
+      "loss": 4.0503,
+      "step": 1346
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009477907219072361,
+      "loss": 4.2202,
+      "step": 1347
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009476984512009571,
+      "loss": 4.047,
+      "step": 1348
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009476061035301241,
+      "loss": 3.9834,
+      "step": 1349
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009475136789106128,
+      "loss": 4.1642,
+      "step": 1350
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009474211773583121,
+      "loss": 3.9971,
+      "step": 1351
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.000947328598889124,
+      "loss": 4.2284,
+      "step": 1352
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.000947235943518964,
+      "loss": 4.0257,
+      "step": 1353
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009471432112637604,
+      "loss": 3.9323,
+      "step": 1354
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.000947050402139455,
+      "loss": 4.1062,
+      "step": 1355
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009469575161620029,
+      "loss": 4.1078,
+      "step": 1356
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009468645533473721,
+      "loss": 4.038,
+      "step": 1357
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009467715137115442,
+      "loss": 4.0763,
+      "step": 1358
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009466783972705136,
+      "loss": 3.9585,
+      "step": 1359
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009465852040402883,
+      "loss": 4.1559,
+      "step": 1360
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.000946491934036889,
+      "loss": 4.1095,
+      "step": 1361
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00094639858727635,
+      "loss": 4.037,
+      "step": 1362
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009463051637747191,
+      "loss": 4.0777,
+      "step": 1363
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009462116635480562,
+      "loss": 3.9852,
+      "step": 1364
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009461180866124356,
+      "loss": 3.9694,
+      "step": 1365
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.000946024432983944,
+      "loss": 4.1436,
+      "step": 1366
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009459307026786816,
+      "loss": 3.9859,
+      "step": 1367
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009458368957127617,
+      "loss": 4.2226,
+      "step": 1368
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.000945743012102311,
+      "loss": 4.0265,
+      "step": 1369
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009456490518634688,
+      "loss": 3.9484,
+      "step": 1370
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009455550150123883,
+      "loss": 3.9299,
+      "step": 1371
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009454609015652353,
+      "loss": 4.0716,
+      "step": 1372
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009453667115381892,
+      "loss": 4.0615,
+      "step": 1373
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009452724449474422,
+      "loss": 4.0629,
+      "step": 1374
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009451781018091999,
+      "loss": 4.0158,
+      "step": 1375
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.000945083682139681,
+      "loss": 4.0003,
+      "step": 1376
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009449891859551172,
+      "loss": 4.1186,
+      "step": 1377
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009448946132717536,
+      "loss": 3.9652,
+      "step": 1378
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009447999641058485,
+      "loss": 3.9484,
+      "step": 1379
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009447052384736728,
+      "loss": 3.9876,
+      "step": 1380
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009446104363915112,
+      "loss": 4.0474,
+      "step": 1381
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009445155578756611,
+      "loss": 4.0614,
+      "step": 1382
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009444206029424334,
+      "loss": 4.0793,
+      "step": 1383
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009443255716081521,
+      "loss": 3.9895,
+      "step": 1384
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009442304638891537,
+      "loss": 4.0415,
+      "step": 1385
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009441352798017886,
+      "loss": 3.9818,
+      "step": 1386
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009440400193624201,
+      "loss": 4.1298,
+      "step": 1387
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009439446825874247,
+      "loss": 4.0385,
+      "step": 1388
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009438492694931915,
+      "loss": 4.1082,
+      "step": 1389
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009437537800961234,
+      "loss": 4.0767,
+      "step": 1390
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009436582144126361,
+      "loss": 4.1133,
+      "step": 1391
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009435625724591583,
+      "loss": 3.9418,
+      "step": 1392
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009434668542521323,
+      "loss": 4.1004,
+      "step": 1393
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009433710598080127,
+      "loss": 4.0713,
+      "step": 1394
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.000943275189143268,
+      "loss": 3.9862,
+      "step": 1395
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009431792422743794,
+      "loss": 3.961,
+      "step": 1396
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009430832192178413,
+      "loss": 4.0355,
+      "step": 1397
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009429871199901613,
+      "loss": 4.166,
+      "step": 1398
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009428909446078598,
+      "loss": 4.0836,
+      "step": 1399
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009427946930874703,
+      "loss": 4.0184,
+      "step": 1400
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009426983654455398,
+      "loss": 4.003,
+      "step": 1401
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009426019616986281,
+      "loss": 4.1242,
+      "step": 1402
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009425054818633081,
+      "loss": 4.165,
+      "step": 1403
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009424089259561657,
+      "loss": 4.0571,
+      "step": 1404
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009423122939938003,
+      "loss": 3.983,
+      "step": 1405
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009422155859928236,
+      "loss": 4.0846,
+      "step": 1406
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009421188019698612,
+      "loss": 4.0876,
+      "step": 1407
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009420219419415514,
+      "loss": 4.1904,
+      "step": 1408
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009419250059245452,
+      "loss": 3.9151,
+      "step": 1409
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009418279939355074,
+      "loss": 3.9734,
+      "step": 1410
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009417309059911152,
+      "loss": 4.2369,
+      "step": 1411
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009416337421080593,
+      "loss": 4.0578,
+      "step": 1412
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009415365023030433,
+      "loss": 4.0931,
+      "step": 1413
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009414391865927838,
+      "loss": 3.8957,
+      "step": 1414
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009413417949940107,
+      "loss": 3.973,
+      "step": 1415
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009412443275234663,
+      "loss": 4.2092,
+      "step": 1416
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009411467841979068,
+      "loss": 3.921,
+      "step": 1417
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009410491650341009,
+      "loss": 4.081,
+      "step": 1418
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009409514700488304,
+      "loss": 4.1174,
+      "step": 1419
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009408536992588902,
+      "loss": 4.1459,
+      "step": 1420
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009407558526810883,
+      "loss": 4.1095,
+      "step": 1421
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009406579303322458,
+      "loss": 4.0729,
+      "step": 1422
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009405599322291962,
+      "loss": 4.1076,
+      "step": 1423
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009404618583887872,
+      "loss": 4.2486,
+      "step": 1424
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009403637088278783,
+      "loss": 4.0817,
+      "step": 1425
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009402654835633427,
+      "loss": 3.9873,
+      "step": 1426
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009401671826120665,
+      "loss": 4.0699,
+      "step": 1427
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009400688059909487,
+      "loss": 4.1474,
+      "step": 1428
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009399703537169016,
+      "loss": 4.0326,
+      "step": 1429
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009398718258068502,
+      "loss": 3.9599,
+      "step": 1430
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009397732222777322,
+      "loss": 4.0413,
+      "step": 1431
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009396745431464993,
+      "loss": 4.0178,
+      "step": 1432
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.000939575788430115,
+      "loss": 3.9228,
+      "step": 1433
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009394769581455569,
+      "loss": 3.9885,
+      "step": 1434
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009393780523098146,
+      "loss": 4.0655,
+      "step": 1435
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009392790709398917,
+      "loss": 3.9338,
+      "step": 1436
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009391800140528037,
+      "loss": 4.0398,
+      "step": 1437
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009390808816655801,
+      "loss": 3.8816,
+      "step": 1438
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009389816737952623,
+      "loss": 3.9897,
+      "step": 1439
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009388823904589061,
+      "loss": 3.9598,
+      "step": 1440
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009387830316735788,
+      "loss": 3.9095,
+      "step": 1441
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009386835974563615,
+      "loss": 4.0158,
+      "step": 1442
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009385840878243483,
+      "loss": 4.0119,
+      "step": 1443
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009384845027946458,
+      "loss": 3.9886,
+      "step": 1444
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.000938384842384374,
+      "loss": 3.822,
+      "step": 1445
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009382851066106655,
+      "loss": 4.1733,
+      "step": 1446
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009381852954906661,
+      "loss": 3.9688,
+      "step": 1447
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009380854090415347,
+      "loss": 3.99,
+      "step": 1448
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009379854472804425,
+      "loss": 3.9522,
+      "step": 1449
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009378854102245746,
+      "loss": 4.1056,
+      "step": 1450
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009377852978911281,
+      "loss": 4.1132,
+      "step": 1451
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009376851102973139,
+      "loss": 3.9606,
+      "step": 1452
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009375848474603549,
+      "loss": 4.1541,
+      "step": 1453
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009374845093974878,
+      "loss": 4.1712,
+      "step": 1454
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009373840961259619,
+      "loss": 3.9351,
+      "step": 1455
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009372836076630392,
+      "loss": 4.0318,
+      "step": 1456
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009371830440259947,
+      "loss": 4.0634,
+      "step": 1457
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009370824052321168,
+      "loss": 4.0253,
+      "step": 1458
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009369816912987065,
+      "loss": 4.0097,
+      "step": 1459
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009368809022430773,
+      "loss": 4.0104,
+      "step": 1460
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009367800380825564,
+      "loss": 4.0016,
+      "step": 1461
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009366790988344834,
+      "loss": 4.0761,
+      "step": 1462
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009365780845162108,
+      "loss": 4.0691,
+      "step": 1463
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009364769951451044,
+      "loss": 3.9252,
+      "step": 1464
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009363758307385422,
+      "loss": 4.05,
+      "step": 1465
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009362745913139159,
+      "loss": 4.2087,
+      "step": 1466
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009361732768886298,
+      "loss": 4.079,
+      "step": 1467
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009360718874801006,
+      "loss": 3.9175,
+      "step": 1468
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009359704231057586,
+      "loss": 4.0432,
+      "step": 1469
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009358688837830468,
+      "loss": 4.06,
+      "step": 1470
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009357672695294208,
+      "loss": 4.1234,
+      "step": 1471
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009356655803623492,
+      "loss": 4.0294,
+      "step": 1472
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009355638162993139,
+      "loss": 4.0095,
+      "step": 1473
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009354619773578088,
+      "loss": 4.1249,
+      "step": 1474
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009353600635553416,
+      "loss": 4.1672,
+      "step": 1475
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009352580749094324,
+      "loss": 3.9645,
+      "step": 1476
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009351560114376142,
+      "loss": 4.0184,
+      "step": 1477
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009350538731574329,
+      "loss": 4.0819,
+      "step": 1478
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009349516600864471,
+      "loss": 4.2173,
+      "step": 1479
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009348493722422287,
+      "loss": 4.1243,
+      "step": 1480
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009347470096423618,
+      "loss": 4.0679,
+      "step": 1481
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009346445723044441,
+      "loss": 4.0627,
+      "step": 1482
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009345420602460855,
+      "loss": 3.9668,
+      "step": 1483
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009344394734849091,
+      "loss": 4.0217,
+      "step": 1484
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009343368120385507,
+      "loss": 4.1139,
+      "step": 1485
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009342340759246591,
+      "loss": 4.0709,
+      "step": 1486
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009341312651608958,
+      "loss": 3.9767,
+      "step": 1487
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009340283797649352,
+      "loss": 4.0382,
+      "step": 1488
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009339254197544642,
+      "loss": 4.0795,
+      "step": 1489
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009338223851471832,
+      "loss": 4.0064,
+      "step": 1490
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009337192759608049,
+      "loss": 3.9133,
+      "step": 1491
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.000933616092213055,
+      "loss": 4.0419,
+      "step": 1492
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009335128339216719,
+      "loss": 4.1654,
+      "step": 1493
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009334095011044069,
+      "loss": 3.9055,
+      "step": 1494
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009333060937790242,
+      "loss": 4.037,
+      "step": 1495
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009332026119633006,
+      "loss": 4.0472,
+      "step": 1496
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.000933099055675026,
+      "loss": 3.9239,
+      "step": 1497
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009329954249320027,
+      "loss": 3.8994,
+      "step": 1498
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009328917197520461,
+      "loss": 3.9131,
+      "step": 1499
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009327879401529846,
+      "loss": 3.988,
+      "step": 1500
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009326840861526586,
+      "loss": 3.9509,
+      "step": 1501
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009325801577689223,
+      "loss": 3.9941,
+      "step": 1502
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009324761550196419,
+      "loss": 4.0164,
+      "step": 1503
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009323720779226967,
+      "loss": 4.0956,
+      "step": 1504
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009322679264959789,
+      "loss": 3.9282,
+      "step": 1505
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.000932163700757393,
+      "loss": 3.9812,
+      "step": 1506
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009320594007248573,
+      "loss": 4.0785,
+      "step": 1507
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009319550264163015,
+      "loss": 3.9974,
+      "step": 1508
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009318505778496691,
+      "loss": 3.9276,
+      "step": 1509
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009317460550429159,
+      "loss": 3.975,
+      "step": 1510
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009316414580140105,
+      "loss": 4.0509,
+      "step": 1511
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009315367867809345,
+      "loss": 4.1418,
+      "step": 1512
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009314320413616821,
+      "loss": 3.9865,
+      "step": 1513
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009313272217742603,
+      "loss": 4.0417,
+      "step": 1514
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009312223280366885,
+      "loss": 3.9021,
+      "step": 1515
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009311173601669996,
+      "loss": 4.0037,
+      "step": 1516
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009310123181832382,
+      "loss": 3.9101,
+      "step": 1517
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009309072021034628,
+      "loss": 4.0322,
+      "step": 1518
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009308020119457439,
+      "loss": 4.0834,
+      "step": 1519
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009306967477281647,
+      "loss": 3.8791,
+      "step": 1520
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009305914094688216,
+      "loss": 4.0628,
+      "step": 1521
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009304859971858233,
+      "loss": 4.1073,
+      "step": 1522
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009303805108972916,
+      "loss": 3.9446,
+      "step": 1523
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009302749506213606,
+      "loss": 4.0929,
+      "step": 1524
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009301693163761773,
+      "loss": 3.9905,
+      "step": 1525
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009300636081799017,
+      "loss": 4.0654,
+      "step": 1526
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009299578260507061,
+      "loss": 4.0581,
+      "step": 1527
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009298519700067757,
+      "loss": 3.9567,
+      "step": 1528
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009297460400663084,
+      "loss": 3.989,
+      "step": 1529
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009296400362475148,
+      "loss": 3.8759,
+      "step": 1530
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.000929533958568618,
+      "loss": 4.0737,
+      "step": 1531
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009294278070478544,
+      "loss": 4.0914,
+      "step": 1532
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009293215817034722,
+      "loss": 4.1015,
+      "step": 1533
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.000929215282553733,
+      "loss": 4.2259,
+      "step": 1534
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009291089096169109,
+      "loss": 4.1295,
+      "step": 1535
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009290024629112926,
+      "loss": 4.1381,
+      "step": 1536
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009288959424551774,
+      "loss": 4.0051,
+      "step": 1537
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009287893482668774,
+      "loss": 3.9718,
+      "step": 1538
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009286826803647177,
+      "loss": 4.1899,
+      "step": 1539
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009285759387670356,
+      "loss": 4.0248,
+      "step": 1540
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009284691234921809,
+      "loss": 4.0956,
+      "step": 1541
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009283622345585168,
+      "loss": 3.988,
+      "step": 1542
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009282552719844185,
+      "loss": 4.2232,
+      "step": 1543
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009281482357882743,
+      "loss": 4.0811,
+      "step": 1544
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009280411259884846,
+      "loss": 3.971,
+      "step": 1545
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009279339426034633,
+      "loss": 4.0171,
+      "step": 1546
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009278266856516362,
+      "loss": 4.1326,
+      "step": 1547
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.000927719355151442,
+      "loss": 4.0178,
+      "step": 1548
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009276119511213321,
+      "loss": 4.0147,
+      "step": 1549
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009275044735797705,
+      "loss": 4.0663,
+      "step": 1550
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009273969225452339,
+      "loss": 4.043,
+      "step": 1551
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009272892980362114,
+      "loss": 4.1409,
+      "step": 1552
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.000927181600071205,
+      "loss": 4.1622,
+      "step": 1553
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009270738286687293,
+      "loss": 4.0845,
+      "step": 1554
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009269659838473113,
+      "loss": 4.0885,
+      "step": 1555
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009268580656254911,
+      "loss": 4.0747,
+      "step": 1556
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009267500740218205,
+      "loss": 4.0866,
+      "step": 1557
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.000926642009054865,
+      "loss": 4.2401,
+      "step": 1558
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009265338707432023,
+      "loss": 3.9376,
+      "step": 1559
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009264256591054222,
+      "loss": 4.0663,
+      "step": 1560
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009263173741601279,
+      "loss": 3.8404,
+      "step": 1561
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009262090159259344,
+      "loss": 3.99,
+      "step": 1562
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009261005844214703,
+      "loss": 4.0528,
+      "step": 1563
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009259920796653758,
+      "loss": 4.0858,
+      "step": 1564
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009258835016763043,
+      "loss": 3.9549,
+      "step": 1565
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009257748504729218,
+      "loss": 3.995,
+      "step": 1566
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009256661260739063,
+      "loss": 4.0467,
+      "step": 1567
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.000925557328497949,
+      "loss": 4.0278,
+      "step": 1568
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009254484577637534,
+      "loss": 3.9825,
+      "step": 1569
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009253395138900358,
+      "loss": 4.0255,
+      "step": 1570
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009252304968955247,
+      "loss": 3.908,
+      "step": 1571
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009251214067989616,
+      "loss": 4.108,
+      "step": 1572
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009250122436191002,
+      "loss": 4.0043,
+      "step": 1573
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009249030073747069,
+      "loss": 4.1026,
+      "step": 1574
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009247936980845608,
+      "loss": 4.0719,
+      "step": 1575
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009246843157674534,
+      "loss": 3.9402,
+      "step": 1576
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009245748604421886,
+      "loss": 4.0967,
+      "step": 1577
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009244653321275833,
+      "loss": 4.0659,
+      "step": 1578
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009243557308424666,
+      "loss": 4.1137,
+      "step": 1579
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009242460566056803,
+      "loss": 4.0753,
+      "step": 1580
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009241363094360784,
+      "loss": 4.0641,
+      "step": 1581
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009240264893525281,
+      "loss": 4.0911,
+      "step": 1582
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009239165963739084,
+      "loss": 4.0972,
+      "step": 1583
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009238066305191113,
+      "loss": 4.0898,
+      "step": 1584
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009236965918070412,
+      "loss": 4.0863,
+      "step": 1585
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009235864802566152,
+      "loss": 3.969,
+      "step": 1586
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009234762958867626,
+      "loss": 3.9496,
+      "step": 1587
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009233660387164254,
+      "loss": 3.9625,
+      "step": 1588
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009232557087645579,
+      "loss": 3.8946,
+      "step": 1589
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009231453060501274,
+      "loss": 4.0515,
+      "step": 1590
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009230348305921131,
+      "loss": 4.0894,
+      "step": 1591
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009229242824095073,
+      "loss": 4.1553,
+      "step": 1592
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009228136615213144,
+      "loss": 4.0002,
+      "step": 1593
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009227029679465514,
+      "loss": 4.102,
+      "step": 1594
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009225922017042479,
+      "loss": 4.0553,
+      "step": 1595
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009224813628134457,
+      "loss": 3.9675,
+      "step": 1596
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009223704512931994,
+      "loss": 4.0641,
+      "step": 1597
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009222594671625759,
+      "loss": 4.0435,
+      "step": 1598
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.000922148410440655,
+      "loss": 4.1507,
+      "step": 1599
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009220372811465282,
+      "loss": 3.9936,
+      "step": 1600
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009219260792993001,
+      "loss": 3.95,
+      "step": 1601
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009218148049180877,
+      "loss": 3.9487,
+      "step": 1602
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009217034580220201,
+      "loss": 4.0467,
+      "step": 1603
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009215920386302394,
+      "loss": 4.0376,
+      "step": 1604
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009214805467618997,
+      "loss": 4.0359,
+      "step": 1605
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009213689824361678,
+      "loss": 4.1372,
+      "step": 1606
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009212573456722226,
+      "loss": 4.1503,
+      "step": 1607
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009211456364892564,
+      "loss": 3.9035,
+      "step": 1608
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009210338549064728,
+      "loss": 3.9566,
+      "step": 1609
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009209220009430885,
+      "loss": 4.1117,
+      "step": 1610
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009208100746183326,
+      "loss": 3.9513,
+      "step": 1611
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009206980759514464,
+      "loss": 4.085,
+      "step": 1612
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009205860049616837,
+      "loss": 3.977,
+      "step": 1613
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.000920473861668311,
+      "loss": 4.0151,
+      "step": 1614
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009203616460906068,
+      "loss": 3.9969,
+      "step": 1615
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009202493582478625,
+      "loss": 3.9712,
+      "step": 1616
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009201369981593814,
+      "loss": 4.0732,
+      "step": 1617
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009200245658444799,
+      "loss": 4.0281,
+      "step": 1618
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009199120613224861,
+      "loss": 3.8713,
+      "step": 1619
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009197994846127409,
+      "loss": 4.0481,
+      "step": 1620
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009196868357345976,
+      "loss": 4.3254,
+      "step": 1621
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009195741147074217,
+      "loss": 4.1358,
+      "step": 1622
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009194613215505915,
+      "loss": 3.9668,
+      "step": 1623
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009193484562834973,
+      "loss": 4.0297,
+      "step": 1624
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009192355189255418,
+      "loss": 4.0982,
+      "step": 1625
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009191225094961407,
+      "loss": 3.9766,
+      "step": 1626
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.000919009428014721,
+      "loss": 4.0638,
+      "step": 1627
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009188962745007233,
+      "loss": 4.0957,
+      "step": 1628
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009187830489735995,
+      "loss": 3.9835,
+      "step": 1629
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009186697514528148,
+      "loss": 4.1554,
+      "step": 1630
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009185563819578462,
+      "loss": 4.0412,
+      "step": 1631
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009184429405081832,
+      "loss": 4.0445,
+      "step": 1632
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009183294271233277,
+      "loss": 4.0184,
+      "step": 1633
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.000918215841822794,
+      "loss": 4.0679,
+      "step": 1634
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009181021846261087,
+      "loss": 3.8751,
+      "step": 1635
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009179884555528109,
+      "loss": 3.9911,
+      "step": 1636
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009178746546224517,
+      "loss": 3.988,
+      "step": 1637
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.000917760781854595,
+      "loss": 3.9677,
+      "step": 1638
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009176468372688168,
+      "loss": 3.9923,
+      "step": 1639
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009175328208847056,
+      "loss": 4.0264,
+      "step": 1640
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.000917418732721862,
+      "loss": 4.0725,
+      "step": 1641
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.000917304572799899,
+      "loss": 4.1447,
+      "step": 1642
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009171903411384422,
+      "loss": 4.0427,
+      "step": 1643
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009170760377571292,
+      "loss": 3.9104,
+      "step": 1644
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009169616626756102,
+      "loss": 4.0653,
+      "step": 1645
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009168472159135476,
+      "loss": 4.1383,
+      "step": 1646
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.000916732697490616,
+      "loss": 3.9895,
+      "step": 1647
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009166181074265025,
+      "loss": 3.9749,
+      "step": 1648
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009165034457409066,
+      "loss": 3.9613,
+      "step": 1649
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009163887124535397,
+      "loss": 4.1338,
+      "step": 1650
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009162739075841262,
+      "loss": 3.9229,
+      "step": 1651
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.000916159031152402,
+      "loss": 4.0705,
+      "step": 1652
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009160440831781157,
+      "loss": 4.0278,
+      "step": 1653
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009159290636810283,
+      "loss": 4.0873,
+      "step": 1654
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009158139726809132,
+      "loss": 3.9724,
+      "step": 1655
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009156988101975554,
+      "loss": 4.1247,
+      "step": 1656
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009155835762507531,
+      "loss": 4.1037,
+      "step": 1657
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009154682708603162,
+      "loss": 4.0386,
+      "step": 1658
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009153528940460669,
+      "loss": 4.0689,
+      "step": 1659
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009152374458278401,
+      "loss": 3.9684,
+      "step": 1660
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009151219262254825,
+      "loss": 4.063,
+      "step": 1661
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009150063352588531,
+      "loss": 4.0379,
+      "step": 1662
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009148906729478236,
+      "loss": 3.8877,
+      "step": 1663
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009147749393122775,
+      "loss": 4.0823,
+      "step": 1664
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.000914659134372111,
+      "loss": 4.0936,
+      "step": 1665
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009145432581472322,
+      "loss": 4.0757,
+      "step": 1666
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009144273106575613,
+      "loss": 3.9025,
+      "step": 1667
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009143112919230314,
+      "loss": 4.0401,
+      "step": 1668
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009141952019635873,
+      "loss": 4.0829,
+      "step": 1669
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009140790407991861,
+      "loss": 4.058,
+      "step": 1670
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009139628084497975,
+      "loss": 4.0958,
+      "step": 1671
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009138465049354031,
+      "loss": 4.0158,
+      "step": 1672
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009137301302759968,
+      "loss": 3.9379,
+      "step": 1673
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009136136844915846,
+      "loss": 4.1072,
+      "step": 1674
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.000913497167602185,
+      "loss": 4.1297,
+      "step": 1675
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009133805796278287,
+      "loss": 4.013,
+      "step": 1676
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009132639205885585,
+      "loss": 3.9555,
+      "step": 1677
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009131471905044293,
+      "loss": 3.9626,
+      "step": 1678
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009130303893955084,
+      "loss": 4.0435,
+      "step": 1679
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009129135172818753,
+      "loss": 4.1,
+      "step": 1680
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009127965741836218,
+      "loss": 4.0712,
+      "step": 1681
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009126795601208515,
+      "loss": 4.1198,
+      "step": 1682
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009125624751136807,
+      "loss": 4.0397,
+      "step": 1683
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009124453191822376,
+      "loss": 4.0811,
+      "step": 1684
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009123280923466626,
+      "loss": 3.9932,
+      "step": 1685
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009122107946271086,
+      "loss": 4.0356,
+      "step": 1686
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00091209342604374,
+      "loss": 4.081,
+      "step": 1687
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.000911975986616734,
+      "loss": 4.0161,
+      "step": 1688
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009118584763662803,
+      "loss": 3.9632,
+      "step": 1689
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009117408953125795,
+      "loss": 3.9133,
+      "step": 1690
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009116232434758456,
+      "loss": 4.1096,
+      "step": 1691
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009115055208763041,
+      "loss": 4.0702,
+      "step": 1692
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009113877275341931,
+      "loss": 4.0242,
+      "step": 1693
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009112698634697624,
+      "loss": 3.8557,
+      "step": 1694
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009111519287032743,
+      "loss": 4.0802,
+      "step": 1695
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009110339232550034,
+      "loss": 4.0898,
+      "step": 1696
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009109158471452357,
+      "loss": 4.0041,
+      "step": 1697
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009107977003942702,
+      "loss": 4.0175,
+      "step": 1698
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009106794830224179,
+      "loss": 3.9741,
+      "step": 1699
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009105611950500012,
+      "loss": 4.123,
+      "step": 1700
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009104428364973553,
+      "loss": 3.8432,
+      "step": 1701
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009103244073848278,
+      "loss": 4.2085,
+      "step": 1702
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009102059077327778,
+      "loss": 4.1379,
+      "step": 1703
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009100873375615765,
+      "loss": 3.8301,
+      "step": 1704
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009099686968916082,
+      "loss": 3.8622,
+      "step": 1705
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009098499857432676,
+      "loss": 3.9494,
+      "step": 1706
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009097312041369634,
+      "loss": 3.9883,
+      "step": 1707
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.000909612352093115,
+      "loss": 3.8874,
+      "step": 1708
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009094934296321549,
+      "loss": 4.0754,
+      "step": 1709
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009093744367745267,
+      "loss": 4.0432,
+      "step": 1710
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009092553735406871,
+      "loss": 4.064,
+      "step": 1711
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009091362399511043,
+      "loss": 4.0107,
+      "step": 1712
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009090170360262586,
+      "loss": 3.9698,
+      "step": 1713
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009088977617866428,
+      "loss": 4.0537,
+      "step": 1714
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009087784172527614,
+      "loss": 3.9979,
+      "step": 1715
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009086590024451311,
+      "loss": 3.9976,
+      "step": 1716
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009085395173842807,
+      "loss": 4.1522,
+      "step": 1717
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.000908419962090751,
+      "loss": 4.0999,
+      "step": 1718
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.000908300336585095,
+      "loss": 4.0835,
+      "step": 1719
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009081806408878778,
+      "loss": 4.0538,
+      "step": 1720
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009080608750196764,
+      "loss": 4.0892,
+      "step": 1721
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009079410390010798,
+      "loss": 3.9933,
+      "step": 1722
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009078211328526895,
+      "loss": 3.8734,
+      "step": 1723
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009077011565951185,
+      "loss": 3.8355,
+      "step": 1724
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009075811102489923,
+      "loss": 4.0458,
+      "step": 1725
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009074609938349481,
+      "loss": 4.1393,
+      "step": 1726
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009073408073736354,
+      "loss": 4.0426,
+      "step": 1727
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009072205508857154,
+      "loss": 4.0719,
+      "step": 1728
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009071002243918621,
+      "loss": 4.007,
+      "step": 1729
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009069798279127605,
+      "loss": 4.1063,
+      "step": 1730
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009068593614691085,
+      "loss": 4.1062,
+      "step": 1731
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009067388250816156,
+      "loss": 3.9776,
+      "step": 1732
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009066182187710031,
+      "loss": 4.1081,
+      "step": 1733
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009064975425580051,
+      "loss": 3.9364,
+      "step": 1734
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009063767964633669,
+      "loss": 4.066,
+      "step": 1735
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009062559805078463,
+      "loss": 3.945,
+      "step": 1736
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009061350947122129,
+      "loss": 3.9421,
+      "step": 1737
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009060141390972485,
+      "loss": 3.9366,
+      "step": 1738
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009058931136837465,
+      "loss": 4.1165,
+      "step": 1739
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009057720184925129,
+      "loss": 4.0752,
+      "step": 1740
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009056508535443651,
+      "loss": 3.9412,
+      "step": 1741
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009055296188601329,
+      "loss": 4.1332,
+      "step": 1742
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009054083144606579,
+      "loss": 3.9327,
+      "step": 1743
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009052869403667938,
+      "loss": 4.116,
+      "step": 1744
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.000905165496599406,
+      "loss": 4.0616,
+      "step": 1745
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009050439831793726,
+      "loss": 4.0199,
+      "step": 1746
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009049224001275825,
+      "loss": 4.0527,
+      "step": 1747
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009048007474649377,
+      "loss": 4.1765,
+      "step": 1748
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009046790252123514,
+      "loss": 3.9794,
+      "step": 1749
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009045572333907495,
+      "loss": 4.1732,
+      "step": 1750
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.000904435372021069,
+      "loss": 4.0193,
+      "step": 1751
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009043134411242593,
+      "loss": 3.9962,
+      "step": 1752
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009041914407212821,
+      "loss": 4.0864,
+      "step": 1753
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009040693708331103,
+      "loss": 3.9046,
+      "step": 1754
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009039472314807293,
+      "loss": 3.9351,
+      "step": 1755
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009038250226851362,
+      "loss": 4.1354,
+      "step": 1756
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009037027444673402,
+      "loss": 3.9234,
+      "step": 1757
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0009035803968483625,
+      "loss": 4.0919,
+      "step": 1758
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0009034579798492356,
+      "loss": 3.9923,
+      "step": 1759
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0009033354934910048,
+      "loss": 4.0864,
+      "step": 1760
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0009032129377947267,
+      "loss": 3.9309,
+      "step": 1761
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0009030903127814703,
+      "loss": 4.0072,
+      "step": 1762
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0009029676184723161,
+      "loss": 3.9408,
+      "step": 1763
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0009028448548883566,
+      "loss": 3.997,
+      "step": 1764
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0009027220220506965,
+      "loss": 3.936,
+      "step": 1765
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0009025991199804517,
+      "loss": 3.8582,
+      "step": 1766
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0009024761486987512,
+      "loss": 3.943,
+      "step": 1767
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0009023531082267347,
+      "loss": 3.9757,
+      "step": 1768
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0009022299985855545,
+      "loss": 4.0978,
+      "step": 1769
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0009021068197963744,
+      "loss": 3.9092,
+      "step": 1770
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0009019835718803704,
+      "loss": 3.9168,
+      "step": 1771
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0009018602548587302,
+      "loss": 4.0227,
+      "step": 1772
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0009017368687526535,
+      "loss": 3.9971,
+      "step": 1773
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0009016134135833516,
+      "loss": 3.9796,
+      "step": 1774
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.000901489889372048,
+      "loss": 4.043,
+      "step": 1775
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.000901366296139978,
+      "loss": 4.0372,
+      "step": 1776
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0009012426339083887,
+      "loss": 4.0565,
+      "step": 1777
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0009011189026985389,
+      "loss": 4.059,
+      "step": 1778
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0009009951025316997,
+      "loss": 4.0281,
+      "step": 1779
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0009008712334291535,
+      "loss": 4.1548,
+      "step": 1780
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0009007472954121952,
+      "loss": 4.1282,
+      "step": 1781
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0009006232885021308,
+      "loss": 4.0329,
+      "step": 1782
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0009004992127202787,
+      "loss": 4.0173,
+      "step": 1783
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0009003750680879689,
+      "loss": 4.1043,
+      "step": 1784
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0009002508546265434,
+      "loss": 4.1219,
+      "step": 1785
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0009001265723573559,
+      "loss": 3.9463,
+      "step": 1786
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0009000022213017719,
+      "loss": 4.0514,
+      "step": 1787
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0008998778014811689,
+      "loss": 4.0551,
+      "step": 1788
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.000899753312916936,
+      "loss": 4.0419,
+      "step": 1789
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0008996287556304743,
+      "loss": 4.0107,
+      "step": 1790
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0008995041296431965,
+      "loss": 4.0074,
+      "step": 1791
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0008993794349765274,
+      "loss": 3.9194,
+      "step": 1792
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0008992546716519034,
+      "loss": 3.9951,
+      "step": 1793
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.000899129839690773,
+      "loss": 3.9947,
+      "step": 1794
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0008990049391145959,
+      "loss": 4.1131,
+      "step": 1795
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0008988799699448441,
+      "loss": 3.9851,
+      "step": 1796
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0008987549322030013,
+      "loss": 3.9671,
+      "step": 1797
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0008986298259105629,
+      "loss": 4.1153,
+      "step": 1798
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.000898504651089036,
+      "loss": 3.9825,
+      "step": 1799
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0008983794077599398,
+      "loss": 4.1276,
+      "step": 1800
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0008982540959448049,
+      "loss": 3.9189,
+      "step": 1801
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.000898128715665174,
+      "loss": 4.0964,
+      "step": 1802
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0008980032669426015,
+      "loss": 4.0864,
+      "step": 1803
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0008978777497986532,
+      "loss": 4.0195,
+      "step": 1804
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0008977521642549072,
+      "loss": 4.0028,
+      "step": 1805
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0008976265103329531,
+      "loss": 4.1364,
+      "step": 1806
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0008975007880543921,
+      "loss": 4.0184,
+      "step": 1807
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0008973749974408374,
+      "loss": 4.1023,
+      "step": 1808
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0008972491385139138,
+      "loss": 4.0719,
+      "step": 1809
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0008971232112952581,
+      "loss": 4.0025,
+      "step": 1810
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0008969972158065185,
+      "loss": 4.1312,
+      "step": 1811
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0008968711520693551,
+      "loss": 3.8754,
+      "step": 1812
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0008967450201054397,
+      "loss": 3.887,
+      "step": 1813
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.000896618819936456,
+      "loss": 4.1313,
+      "step": 1814
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0008964925515840991,
+      "loss": 4.0131,
+      "step": 1815
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.000896366215070076,
+      "loss": 3.8814,
+      "step": 1816
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0008962398104161054,
+      "loss": 3.9334,
+      "step": 1817
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0008961133376439178,
+      "loss": 4.0352,
+      "step": 1818
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0008959867967752556,
+      "loss": 4.0937,
+      "step": 1819
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.000895860187831872,
+      "loss": 4.0416,
+      "step": 1820
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0008957335108355331,
+      "loss": 4.0362,
+      "step": 1821
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0008956067658080158,
+      "loss": 3.929,
+      "step": 1822
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0008954799527711094,
+      "loss": 3.9722,
+      "step": 1823
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0008953530717466142,
+      "loss": 3.8911,
+      "step": 1824
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0008952261227563426,
+      "loss": 4.0659,
+      "step": 1825
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0008950991058221188,
+      "loss": 4.1396,
+      "step": 1826
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0008949720209657781,
+      "loss": 3.9564,
+      "step": 1827
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0008948448682091683,
+      "loss": 3.963,
+      "step": 1828
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0008947176475741481,
+      "loss": 4.0197,
+      "step": 1829
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0008945903590825883,
+      "loss": 3.9886,
+      "step": 1830
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0008944630027563714,
+      "loss": 4.1277,
+      "step": 1831
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0008943355786173912,
+      "loss": 4.0047,
+      "step": 1832
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0008942080866875537,
+      "loss": 3.9065,
+      "step": 1833
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.000894080526988776,
+      "loss": 4.1224,
+      "step": 1834
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0008939528995429869,
+      "loss": 4.0576,
+      "step": 1835
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008938252043721274,
+      "loss": 3.9977,
+      "step": 1836
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008936974414981498,
+      "loss": 3.9847,
+      "step": 1837
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008935696109430177,
+      "loss": 4.0426,
+      "step": 1838
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008934417127287069,
+      "loss": 3.9758,
+      "step": 1839
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008933137468772046,
+      "loss": 3.9676,
+      "step": 1840
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008931857134105093,
+      "loss": 4.0571,
+      "step": 1841
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008930576123506317,
+      "loss": 3.9311,
+      "step": 1842
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.000892929443719594,
+      "loss": 4.019,
+      "step": 1843
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008928012075394296,
+      "loss": 3.8879,
+      "step": 1844
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.000892672903832184,
+      "loss": 3.961,
+      "step": 1845
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008925445326199138,
+      "loss": 4.0004,
+      "step": 1846
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008924160939246877,
+      "loss": 3.9522,
+      "step": 1847
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008922875877685858,
+      "loss": 4.0449,
+      "step": 1848
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008921590141737,
+      "loss": 3.914,
+      "step": 1849
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008920303731621331,
+      "loss": 3.9906,
+      "step": 1850
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008919016647560002,
+      "loss": 4.0996,
+      "step": 1851
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008917728889774279,
+      "loss": 3.9853,
+      "step": 1852
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008916440458485542,
+      "loss": 4.0532,
+      "step": 1853
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008915151353915288,
+      "loss": 4.0802,
+      "step": 1854
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008913861576285127,
+      "loss": 3.9307,
+      "step": 1855
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008912571125816786,
+      "loss": 3.98,
+      "step": 1856
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008911280002732112,
+      "loss": 4.0515,
+      "step": 1857
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008909988207253061,
+      "loss": 4.0544,
+      "step": 1858
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008908695739601709,
+      "loss": 4.0382,
+      "step": 1859
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008907402600000248,
+      "loss": 4.0625,
+      "step": 1860
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008906108788670979,
+      "loss": 3.9914,
+      "step": 1861
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008904814305836326,
+      "loss": 4.0326,
+      "step": 1862
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008903519151718826,
+      "loss": 3.895,
+      "step": 1863
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.000890222332654113,
+      "loss": 3.9777,
+      "step": 1864
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008900926830526005,
+      "loss": 4.1168,
+      "step": 1865
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008899629663896336,
+      "loss": 4.1108,
+      "step": 1866
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008898331826875117,
+      "loss": 4.056,
+      "step": 1867
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008897033319685466,
+      "loss": 4.0359,
+      "step": 1868
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008895734142550609,
+      "loss": 3.8991,
+      "step": 1869
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008894434295693888,
+      "loss": 4.0296,
+      "step": 1870
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008893133779338764,
+      "loss": 4.1027,
+      "step": 1871
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008891832593708811,
+      "loss": 4.0103,
+      "step": 1872
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008890530739027717,
+      "loss": 4.0127,
+      "step": 1873
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008889228215519285,
+      "loss": 4.1297,
+      "step": 1874
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008887925023407437,
+      "loss": 3.8903,
+      "step": 1875
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008886621162916203,
+      "loss": 3.9995,
+      "step": 1876
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008885316634269733,
+      "loss": 3.9477,
+      "step": 1877
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008884011437692294,
+      "loss": 4.0646,
+      "step": 1878
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008882705573408259,
+      "loss": 4.066,
+      "step": 1879
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008881399041642124,
+      "loss": 3.9106,
+      "step": 1880
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008880091842618497,
+      "loss": 4.0341,
+      "step": 1881
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008878783976562102,
+      "loss": 4.1079,
+      "step": 1882
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008877475443697773,
+      "loss": 3.8885,
+      "step": 1883
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008876166244250463,
+      "loss": 3.9097,
+      "step": 1884
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008874856378445241,
+      "loss": 3.9667,
+      "step": 1885
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008873545846507285,
+      "loss": 4.1562,
+      "step": 1886
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008872234648661892,
+      "loss": 3.9871,
+      "step": 1887
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008870922785134472,
+      "loss": 4.1858,
+      "step": 1888
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008869610256150551,
+      "loss": 3.873,
+      "step": 1889
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008868297061935766,
+      "loss": 3.8971,
+      "step": 1890
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.000886698320271587,
+      "loss": 4.0632,
+      "step": 1891
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008865668678716734,
+      "loss": 3.8502,
+      "step": 1892
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008864353490164335,
+      "loss": 4.0519,
+      "step": 1893
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008863037637284772,
+      "loss": 3.917,
+      "step": 1894
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008861721120304256,
+      "loss": 3.9513,
+      "step": 1895
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008860403939449109,
+      "loss": 4.0673,
+      "step": 1896
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008859086094945772,
+      "loss": 4.0175,
+      "step": 1897
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008857767587020798,
+      "loss": 4.0908,
+      "step": 1898
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.000885644841590085,
+      "loss": 4.0222,
+      "step": 1899
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008855128581812714,
+      "loss": 4.0354,
+      "step": 1900
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008853808084983281,
+      "loss": 3.9892,
+      "step": 1901
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008852486925639561,
+      "loss": 4.0638,
+      "step": 1902
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008851165104008679,
+      "loss": 3.9295,
+      "step": 1903
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008849842620317865,
+      "loss": 3.9447,
+      "step": 1904
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008848519474794476,
+      "loss": 4.0501,
+      "step": 1905
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008847195667665973,
+      "loss": 3.9604,
+      "step": 1906
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008845871199159934,
+      "loss": 3.9922,
+      "step": 1907
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008844546069504051,
+      "loss": 4.016,
+      "step": 1908
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008843220278926128,
+      "loss": 3.7922,
+      "step": 1909
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008841893827654086,
+      "loss": 3.9906,
+      "step": 1910
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008840566715915955,
+      "loss": 3.8654,
+      "step": 1911
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008839238943939882,
+      "loss": 3.836,
+      "step": 1912
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008837910511954127,
+      "loss": 4.0698,
+      "step": 1913
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008836581420187062,
+      "loss": 4.1016,
+      "step": 1914
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008835251668867172,
+      "loss": 4.0843,
+      "step": 1915
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008833921258223059,
+      "loss": 3.8661,
+      "step": 1916
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008832590188483436,
+      "loss": 3.8413,
+      "step": 1917
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008831258459877127,
+      "loss": 4.0445,
+      "step": 1918
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008829926072633074,
+      "loss": 4.0326,
+      "step": 1919
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008828593026980327,
+      "loss": 4.1069,
+      "step": 1920
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008827259323148055,
+      "loss": 3.8036,
+      "step": 1921
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008825924961365537,
+      "loss": 4.0144,
+      "step": 1922
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008824589941862163,
+      "loss": 3.9622,
+      "step": 1923
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008823254264867439,
+      "loss": 3.9252,
+      "step": 1924
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008821917930610985,
+      "loss": 4.0511,
+      "step": 1925
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008820580939322532,
+      "loss": 4.0936,
+      "step": 1926
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008819243291231922,
+      "loss": 3.917,
+      "step": 1927
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008817904986569114,
+      "loss": 4.0538,
+      "step": 1928
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008816566025564179,
+      "loss": 3.9891,
+      "step": 1929
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008815226408447296,
+      "loss": 4.0086,
+      "step": 1930
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008813886135448767,
+      "loss": 3.9551,
+      "step": 1931
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008812545206798994,
+      "loss": 3.9812,
+      "step": 1932
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008811203622728503,
+      "loss": 3.9249,
+      "step": 1933
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008809861383467925,
+      "loss": 3.9517,
+      "step": 1934
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008808518489248009,
+      "loss": 3.9851,
+      "step": 1935
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008807174940299612,
+      "loss": 3.942,
+      "step": 1936
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008805830736853705,
+      "loss": 3.8985,
+      "step": 1937
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008804485879141374,
+      "loss": 3.8848,
+      "step": 1938
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008803140367393815,
+      "loss": 4.0136,
+      "step": 1939
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008801794201842337,
+      "loss": 4.0134,
+      "step": 1940
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008800447382718362,
+      "loss": 3.9094,
+      "step": 1941
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008799099910253423,
+      "loss": 3.9684,
+      "step": 1942
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008797751784679166,
+      "loss": 4.0451,
+      "step": 1943
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.000879640300622735,
+      "loss": 4.0839,
+      "step": 1944
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008795053575129846,
+      "loss": 4.2072,
+      "step": 1945
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008793703491618637,
+      "loss": 3.9294,
+      "step": 1946
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008792352755925817,
+      "loss": 4.0481,
+      "step": 1947
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008791001368283594,
+      "loss": 3.9456,
+      "step": 1948
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008789649328924285,
+      "loss": 4.0213,
+      "step": 1949
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008788296638080324,
+      "loss": 3.8836,
+      "step": 1950
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008786943295984253,
+      "loss": 3.9735,
+      "step": 1951
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008785589302868727,
+      "loss": 3.9829,
+      "step": 1952
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008784234658966515,
+      "loss": 3.8428,
+      "step": 1953
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008782879364510494,
+      "loss": 3.8928,
+      "step": 1954
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008781523419733654,
+      "loss": 3.9886,
+      "step": 1955
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00087801668248691,
+      "loss": 4.0852,
+      "step": 1956
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008778809580150045,
+      "loss": 4.003,
+      "step": 1957
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008777451685809816,
+      "loss": 4.0524,
+      "step": 1958
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008776093142081849,
+      "loss": 3.9644,
+      "step": 1959
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008774733949199696,
+      "loss": 3.9737,
+      "step": 1960
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008773374107397016,
+      "loss": 3.9469,
+      "step": 1961
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008772013616907584,
+      "loss": 3.9787,
+      "step": 1962
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.000877065247796528,
+      "loss": 3.9479,
+      "step": 1963
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008769290690804103,
+      "loss": 4.061,
+      "step": 1964
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008767928255658157,
+      "loss": 4.0537,
+      "step": 1965
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008766565172761664,
+      "loss": 4.0786,
+      "step": 1966
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008765201442348954,
+      "loss": 3.8787,
+      "step": 1967
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008763837064654463,
+      "loss": 4.0655,
+      "step": 1968
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008762472039912747,
+      "loss": 3.999,
+      "step": 1969
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008761106368358469,
+      "loss": 4.0673,
+      "step": 1970
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008759740050226406,
+      "loss": 4.027,
+      "step": 1971
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008758373085751439,
+      "loss": 3.8436,
+      "step": 1972
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008757005475168569,
+      "loss": 4.0174,
+      "step": 1973
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008755637218712902,
+      "loss": 4.0093,
+      "step": 1974
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008754268316619659,
+      "loss": 3.9522,
+      "step": 1975
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008752898769124169,
+      "loss": 4.0959,
+      "step": 1976
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008751528576461873,
+      "loss": 4.071,
+      "step": 1977
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008750157738868323,
+      "loss": 3.9647,
+      "step": 1978
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008748786256579182,
+      "loss": 4.0046,
+      "step": 1979
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008747414129830224,
+      "loss": 3.9754,
+      "step": 1980
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008746041358857334,
+      "loss": 4.0474,
+      "step": 1981
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008744667943896506,
+      "loss": 4.1713,
+      "step": 1982
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008743293885183848,
+      "loss": 3.9023,
+      "step": 1983
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008741919182955573,
+      "loss": 3.9157,
+      "step": 1984
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008740543837448012,
+      "loss": 4.0707,
+      "step": 1985
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00087391678488976,
+      "loss": 4.0085,
+      "step": 1986
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008737791217540887,
+      "loss": 3.9351,
+      "step": 1987
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008736413943614532,
+      "loss": 3.9484,
+      "step": 1988
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008735036027355303,
+      "loss": 3.9685,
+      "step": 1989
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008733657469000081,
+      "loss": 4.0127,
+      "step": 1990
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008732278268785856,
+      "loss": 3.9655,
+      "step": 1991
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008730898426949727,
+      "loss": 4.0335,
+      "step": 1992
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008729517943728909,
+      "loss": 3.9079,
+      "step": 1993
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008728136819360717,
+      "loss": 3.939,
+      "step": 1994
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008726755054082588,
+      "loss": 4.049,
+      "step": 1995
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008725372648132059,
+      "loss": 4.1466,
+      "step": 1996
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008723989601746784,
+      "loss": 3.9379,
+      "step": 1997
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008722605915164525,
+      "loss": 3.9889,
+      "step": 1998
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008721221588623154,
+      "loss": 4.0101,
+      "step": 1999
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.000871983662236065,
+      "loss": 4.1249,
+      "step": 2000
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008718451016615108,
+      "loss": 4.0013,
+      "step": 2001
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008717064771624727,
+      "loss": 4.0869,
+      "step": 2002
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.000871567788762782,
+      "loss": 4.0035,
+      "step": 2003
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.000871429036486281,
+      "loss": 4.0128,
+      "step": 2004
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008712902203568225,
+      "loss": 3.9001,
+      "step": 2005
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008711513403982707,
+      "loss": 4.0041,
+      "step": 2006
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008710123966345007,
+      "loss": 3.9554,
+      "step": 2007
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008708733890893987,
+      "loss": 3.916,
+      "step": 2008
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008707343177868617,
+      "loss": 4.0464,
+      "step": 2009
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008705951827507974,
+      "loss": 4.0896,
+      "step": 2010
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008704559840051248,
+      "loss": 3.9994,
+      "step": 2011
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008703167215737739,
+      "loss": 3.887,
+      "step": 2012
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008701773954806855,
+      "loss": 4.0497,
+      "step": 2013
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008700380057498114,
+      "loss": 3.9417,
+      "step": 2014
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008698985524051143,
+      "loss": 4.0084,
+      "step": 2015
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008697590354705678,
+      "loss": 4.0284,
+      "step": 2016
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008696194549701564,
+      "loss": 4.0583,
+      "step": 2017
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008694798109278757,
+      "loss": 4.1197,
+      "step": 2018
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008693401033677321,
+      "loss": 4.0052,
+      "step": 2019
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.000869200332313743,
+      "loss": 4.0022,
+      "step": 2020
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008690604977899368,
+      "loss": 3.9727,
+      "step": 2021
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008689205998203522,
+      "loss": 4.0644,
+      "step": 2022
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008687806384290399,
+      "loss": 3.8338,
+      "step": 2023
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008686406136400604,
+      "loss": 3.8593,
+      "step": 2024
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008685005254774858,
+      "loss": 4.0175,
+      "step": 2025
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008683603739653989,
+      "loss": 4.0185,
+      "step": 2026
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008682201591278933,
+      "loss": 3.9355,
+      "step": 2027
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008680798809890736,
+      "loss": 4.062,
+      "step": 2028
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008679395395730552,
+      "loss": 3.8984,
+      "step": 2029
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008677991349039643,
+      "loss": 4.0778,
+      "step": 2030
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008676586670059383,
+      "loss": 4.0268,
+      "step": 2031
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008675181359031252,
+      "loss": 4.0407,
+      "step": 2032
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008673775416196839,
+      "loss": 4.0478,
+      "step": 2033
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008672368841797841,
+      "loss": 4.0774,
+      "step": 2034
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008670961636076067,
+      "loss": 3.9885,
+      "step": 2035
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008669553799273428,
+      "loss": 4.0734,
+      "step": 2036
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008668145331631953,
+      "loss": 4.0039,
+      "step": 2037
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008666736233393768,
+      "loss": 3.9379,
+      "step": 2038
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008665326504801117,
+      "loss": 3.9848,
+      "step": 2039
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008663916146096348,
+      "loss": 4.1149,
+      "step": 2040
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008662505157521918,
+      "loss": 4.078,
+      "step": 2041
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008661093539320391,
+      "loss": 3.9347,
+      "step": 2042
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008659681291734441,
+      "loss": 4.0805,
+      "step": 2043
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008658268415006851,
+      "loss": 3.8654,
+      "step": 2044
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008656854909380511,
+      "loss": 4.0075,
+      "step": 2045
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008655440775098417,
+      "loss": 4.0585,
+      "step": 2046
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008654026012403677,
+      "loss": 4.052,
+      "step": 2047
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008652610621539505,
+      "loss": 4.1531,
+      "step": 2048
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008651194602749221,
+      "loss": 4.0586,
+      "step": 2049
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008649777956276257,
+      "loss": 3.903,
+      "step": 2050
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008648360682364151,
+      "loss": 3.7751,
+      "step": 2051
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008646942781256548,
+      "loss": 3.7696,
+      "step": 2052
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008645524253197201,
+      "loss": 4.0357,
+      "step": 2053
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008644105098429974,
+      "loss": 4.0302,
+      "step": 2054
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008642685317198833,
+      "loss": 4.093,
+      "step": 2055
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008641264909747858,
+      "loss": 3.8294,
+      "step": 2056
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008639843876321231,
+      "loss": 4.0748,
+      "step": 2057
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008638422217163244,
+      "loss": 3.8836,
+      "step": 2058
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00086369999325183,
+      "loss": 4.0758,
+      "step": 2059
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008635577022630901,
+      "loss": 4.0095,
+      "step": 2060
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008634153487745666,
+      "loss": 3.9365,
+      "step": 2061
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008632729328107317,
+      "loss": 3.8698,
+      "step": 2062
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.000863130454396068,
+      "loss": 4.0562,
+      "step": 2063
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008629879135550694,
+      "loss": 4.0592,
+      "step": 2064
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008628453103122404,
+      "loss": 4.0561,
+      "step": 2065
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008627026446920963,
+      "loss": 4.0194,
+      "step": 2066
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008625599167191626,
+      "loss": 4.0865,
+      "step": 2067
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008624171264179761,
+      "loss": 3.9926,
+      "step": 2068
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.000862274273813084,
+      "loss": 3.9454,
+      "step": 2069
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008621313589290445,
+      "loss": 4.1239,
+      "step": 2070
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008619883817904262,
+      "loss": 3.9849,
+      "step": 2071
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008618453424218084,
+      "loss": 4.022,
+      "step": 2072
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008617022408477815,
+      "loss": 4.0725,
+      "step": 2073
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.000861559077092946,
+      "loss": 4.045,
+      "step": 2074
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008614158511819138,
+      "loss": 3.9529,
+      "step": 2075
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008612725631393068,
+      "loss": 4.0186,
+      "step": 2076
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008611292129897581,
+      "loss": 3.9337,
+      "step": 2077
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.000860985800757911,
+      "loss": 3.9083,
+      "step": 2078
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008608423264684198,
+      "loss": 3.8446,
+      "step": 2079
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008606987901459496,
+      "loss": 3.7968,
+      "step": 2080
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008605551918151755,
+      "loss": 3.9302,
+      "step": 2081
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008604115315007843,
+      "loss": 3.976,
+      "step": 2082
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008602678092274725,
+      "loss": 3.9225,
+      "step": 2083
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008601240250199476,
+      "loss": 3.9452,
+      "step": 2084
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.000859980178902928,
+      "loss": 4.0661,
+      "step": 2085
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008598362709011424,
+      "loss": 4.0203,
+      "step": 2086
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008596923010393303,
+      "loss": 4.0909,
+      "step": 2087
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008595482693422417,
+      "loss": 4.0547,
+      "step": 2088
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008594041758346371,
+      "loss": 3.8155,
+      "step": 2089
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008592600205412883,
+      "loss": 3.9899,
+      "step": 2090
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008591158034869773,
+      "loss": 4.0467,
+      "step": 2091
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008589715246964963,
+      "loss": 4.0339,
+      "step": 2092
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008588271841946484,
+      "loss": 4.033,
+      "step": 2093
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.000858682782006248,
+      "loss": 4.0747,
+      "step": 2094
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.000858538318156119,
+      "loss": 4.0118,
+      "step": 2095
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008583937926690965,
+      "loss": 3.9214,
+      "step": 2096
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008582492055700264,
+      "loss": 4.0169,
+      "step": 2097
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008581045568837645,
+      "loss": 3.9881,
+      "step": 2098
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008579598466351779,
+      "loss": 4.0919,
+      "step": 2099
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008578150748491438,
+      "loss": 3.9355,
+      "step": 2100
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008576702415505501,
+      "loss": 4.0811,
+      "step": 2101
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008575253467642954,
+      "loss": 3.9904,
+      "step": 2102
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.000857380390515289,
+      "loss": 4.0671,
+      "step": 2103
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008572353728284501,
+      "loss": 4.0302,
+      "step": 2104
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008570902937287092,
+      "loss": 4.0266,
+      "step": 2105
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008569451532410074,
+      "loss": 3.9904,
+      "step": 2106
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008567999513902953,
+      "loss": 3.9908,
+      "step": 2107
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008566546882015354,
+      "loss": 3.8378,
+      "step": 2108
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008565093636996999,
+      "loss": 3.8891,
+      "step": 2109
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008563639779097716,
+      "loss": 4.0055,
+      "step": 2110
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008562185308567443,
+      "loss": 4.0414,
+      "step": 2111
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.000856073022565622,
+      "loss": 3.9182,
+      "step": 2112
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008559274530614189,
+      "loss": 4.0603,
+      "step": 2113
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008557818223691607,
+      "loss": 3.9959,
+      "step": 2114
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008556361305138825,
+      "loss": 4.0427,
+      "step": 2115
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008554903775206304,
+      "loss": 3.8908,
+      "step": 2116
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008553445634144614,
+      "loss": 3.9512,
+      "step": 2117
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008551986882204424,
+      "loss": 3.9854,
+      "step": 2118
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008550527519636511,
+      "loss": 3.9554,
+      "step": 2119
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008549067546691756,
+      "loss": 3.9804,
+      "step": 2120
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008547606963621145,
+      "loss": 4.0171,
+      "step": 2121
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.000854614577067577,
+      "loss": 4.1356,
+      "step": 2122
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008544683968106826,
+      "loss": 3.9364,
+      "step": 2123
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008543221556165614,
+      "loss": 3.986,
+      "step": 2124
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008541758535103541,
+      "loss": 4.0013,
+      "step": 2125
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008540294905172117,
+      "loss": 3.9861,
+      "step": 2126
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008538830666622954,
+      "loss": 3.8645,
+      "step": 2127
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008537365819707776,
+      "loss": 3.9682,
+      "step": 2128
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008535900364678402,
+      "loss": 4.0255,
+      "step": 2129
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008534434301786766,
+      "loss": 3.9501,
+      "step": 2130
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008532967631284898,
+      "loss": 3.8264,
+      "step": 2131
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008531500353424937,
+      "loss": 3.9087,
+      "step": 2132
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008530032468459126,
+      "loss": 3.8665,
+      "step": 2133
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008528563976639807,
+      "loss": 4.0286,
+      "step": 2134
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008527094878219434,
+      "loss": 4.0085,
+      "step": 2135
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008525625173450564,
+      "loss": 4.111,
+      "step": 2136
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008524154862585853,
+      "loss": 4.0615,
+      "step": 2137
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008522683945878066,
+      "loss": 3.8995,
+      "step": 2138
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.000852121242358007,
+      "loss": 3.9155,
+      "step": 2139
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008519740295944838,
+      "loss": 4.0014,
+      "step": 2140
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008518267563225444,
+      "loss": 4.102,
+      "step": 2141
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008516794225675069,
+      "loss": 4.0373,
+      "step": 2142
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008515320283546995,
+      "loss": 4.0671,
+      "step": 2143
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008513845737094612,
+      "loss": 4.0092,
+      "step": 2144
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008512370586571409,
+      "loss": 3.9613,
+      "step": 2145
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008510894832230987,
+      "loss": 4.0313,
+      "step": 2146
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.000850941847432704,
+      "loss": 3.9743,
+      "step": 2147
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.000850794151311337,
+      "loss": 3.9472,
+      "step": 2148
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.000850646394884389,
+      "loss": 3.9379,
+      "step": 2149
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008504985781772605,
+      "loss": 3.9501,
+      "step": 2150
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008503507012153632,
+      "loss": 3.9907,
+      "step": 2151
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008502027640241187,
+      "loss": 4.0074,
+      "step": 2152
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.000850054766628959,
+      "loss": 4.0116,
+      "step": 2153
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008499067090553269,
+      "loss": 4.0517,
+      "step": 2154
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008497585913286752,
+      "loss": 4.0912,
+      "step": 2155
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008496104134744667,
+      "loss": 4.1425,
+      "step": 2156
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.000849462175518175,
+      "loss": 3.9101,
+      "step": 2157
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008493138774852841,
+      "loss": 3.9007,
+      "step": 2158
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008491655194012881,
+      "loss": 3.8792,
+      "step": 2159
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008490171012916914,
+      "loss": 4.0785,
+      "step": 2160
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008488686231820088,
+      "loss": 4.078,
+      "step": 2161
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008487200850977657,
+      "loss": 3.9911,
+      "step": 2162
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008485714870644969,
+      "loss": 3.839,
+      "step": 2163
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008484228291077488,
+      "loss": 3.9198,
+      "step": 2164
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008482741112530769,
+      "loss": 3.9257,
+      "step": 2165
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008481253335260478,
+      "loss": 4.0039,
+      "step": 2166
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.000847976495952238,
+      "loss": 3.9526,
+      "step": 2167
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008478275985572346,
+      "loss": 3.9356,
+      "step": 2168
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008476786413666345,
+      "loss": 3.9858,
+      "step": 2169
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008475296244060454,
+      "loss": 3.848,
+      "step": 2170
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008473805477010849,
+      "loss": 3.9141,
+      "step": 2171
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008472314112773811,
+      "loss": 4.0753,
+      "step": 2172
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008470822151605723,
+      "loss": 3.9517,
+      "step": 2173
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008469329593763069,
+      "loss": 3.9035,
+      "step": 2174
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008467836439502439,
+      "loss": 4.045,
+      "step": 2175
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.000846634268908052,
+      "loss": 4.013,
+      "step": 2176
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.000846484834275411,
+      "loss": 3.8404,
+      "step": 2177
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008463353400780101,
+      "loss": 3.9606,
+      "step": 2178
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008461857863415492,
+      "loss": 3.9702,
+      "step": 2179
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008460361730917383,
+      "loss": 3.9759,
+      "step": 2180
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008458865003542975,
+      "loss": 4.0482,
+      "step": 2181
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008457367681549575,
+      "loss": 4.0508,
+      "step": 2182
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008455869765194591,
+      "loss": 3.9565,
+      "step": 2183
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.000845437125473553,
+      "loss": 3.8158,
+      "step": 2184
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008452872150430002,
+      "loss": 3.9182,
+      "step": 2185
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008451372452535724,
+      "loss": 4.0031,
+      "step": 2186
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.000844987216131051,
+      "loss": 3.98,
+      "step": 2187
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008448371277012276,
+      "loss": 3.9257,
+      "step": 2188
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008446869799899042,
+      "loss": 3.8743,
+      "step": 2189
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.000844536773022893,
+      "loss": 4.0317,
+      "step": 2190
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008443865068260164,
+      "loss": 3.9397,
+      "step": 2191
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008442361814251069,
+      "loss": 3.9678,
+      "step": 2192
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008440857968460069,
+      "loss": 3.8654,
+      "step": 2193
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008439353531145695,
+      "loss": 4.0164,
+      "step": 2194
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008437848502566575,
+      "loss": 4.0684,
+      "step": 2195
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008436342882981445,
+      "loss": 4.0153,
+      "step": 2196
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008434836672649135,
+      "loss": 4.0828,
+      "step": 2197
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008433329871828582,
+      "loss": 4.0556,
+      "step": 2198
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008431822480778819,
+      "loss": 3.856,
+      "step": 2199
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008430314499758986,
+      "loss": 3.8762,
+      "step": 2200
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008428805929028326,
+      "loss": 3.9134,
+      "step": 2201
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008427296768846175,
+      "loss": 3.9365,
+      "step": 2202
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008425787019471978,
+      "loss": 3.9068,
+      "step": 2203
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008424276681165275,
+      "loss": 4.0071,
+      "step": 2204
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008422765754185715,
+      "loss": 4.0021,
+      "step": 2205
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.000842125423879304,
+      "loss": 3.8782,
+      "step": 2206
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00084197421352471,
+      "loss": 4.1007,
+      "step": 2207
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008418229443807841,
+      "loss": 3.8716,
+      "step": 2208
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008416716164735314,
+      "loss": 4.0222,
+      "step": 2209
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008415202298289668,
+      "loss": 4.0003,
+      "step": 2210
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008413687844731155,
+      "loss": 4.0573,
+      "step": 2211
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008412172804320126,
+      "loss": 3.8935,
+      "step": 2212
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008410657177317034,
+      "loss": 4.0506,
+      "step": 2213
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008409140963982435,
+      "loss": 3.9753,
+      "step": 2214
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008407624164576981,
+      "loss": 3.9304,
+      "step": 2215
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008406106779361429,
+      "loss": 3.9668,
+      "step": 2216
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008404588808596634,
+      "loss": 4.0144,
+      "step": 2217
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008403070252543555,
+      "loss": 4.0222,
+      "step": 2218
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008401551111463245,
+      "loss": 3.9729,
+      "step": 2219
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008400031385616867,
+      "loss": 4.0494,
+      "step": 2220
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008398511075265677,
+      "loss": 3.9839,
+      "step": 2221
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008396990180671033,
+      "loss": 3.9339,
+      "step": 2222
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008395468702094398,
+      "loss": 3.9731,
+      "step": 2223
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008393946639797328,
+      "loss": 3.9742,
+      "step": 2224
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008392423994041485,
+      "loss": 4.0768,
+      "step": 2225
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.000839090076508863,
+      "loss": 3.9575,
+      "step": 2226
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.000838937695320062,
+      "loss": 3.9292,
+      "step": 2227
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008387852558639421,
+      "loss": 3.9531,
+      "step": 2228
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008386327581667091,
+      "loss": 4.0211,
+      "step": 2229
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008384802022545793,
+      "loss": 3.8498,
+      "step": 2230
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008383275881537786,
+      "loss": 4.0345,
+      "step": 2231
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008381749158905432,
+      "loss": 3.9212,
+      "step": 2232
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008380221854911194,
+      "loss": 4.0574,
+      "step": 2233
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008378693969817632,
+      "loss": 3.8419,
+      "step": 2234
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008377165503887409,
+      "loss": 3.8741,
+      "step": 2235
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008375636457383282,
+      "loss": 4.0165,
+      "step": 2236
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008374106830568117,
+      "loss": 3.9781,
+      "step": 2237
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.000837257662370487,
+      "loss": 3.9773,
+      "step": 2238
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008371045837056604,
+      "loss": 3.9392,
+      "step": 2239
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008369514470886476,
+      "loss": 3.9473,
+      "step": 2240
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.000836798252545775,
+      "loss": 3.9083,
+      "step": 2241
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008366450001033783,
+      "loss": 3.9148,
+      "step": 2242
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008364916897878032,
+      "loss": 3.9922,
+      "step": 2243
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008363383216254057,
+      "loss": 4.0567,
+      "step": 2244
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008361848956425515,
+      "loss": 3.8695,
+      "step": 2245
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008360314118656165,
+      "loss": 3.8348,
+      "step": 2246
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008358778703209861,
+      "loss": 3.978,
+      "step": 2247
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008357242710350561,
+      "loss": 4.0885,
+      "step": 2248
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008355706140342318,
+      "loss": 3.9405,
+      "step": 2249
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008354168993449285,
+      "loss": 3.9269,
+      "step": 2250
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008352631269935718,
+      "loss": 4.0932,
+      "step": 2251
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008351092970065969,
+      "loss": 3.9981,
+      "step": 2252
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.000834955409410449,
+      "loss": 3.9699,
+      "step": 2253
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008348014642315831,
+      "loss": 3.97,
+      "step": 2254
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008346474614964641,
+      "loss": 4.0999,
+      "step": 2255
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008344934012315668,
+      "loss": 3.9817,
+      "step": 2256
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008343392834633761,
+      "loss": 3.8298,
+      "step": 2257
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008341851082183867,
+      "loss": 4.0189,
+      "step": 2258
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008340308755231027,
+      "loss": 4.1205,
+      "step": 2259
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008338765854040392,
+      "loss": 3.9538,
+      "step": 2260
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008337222378877196,
+      "loss": 4.0488,
+      "step": 2261
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008335678330006789,
+      "loss": 3.9334,
+      "step": 2262
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008334133707694604,
+      "loss": 4.0683,
+      "step": 2263
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008332588512206181,
+      "loss": 4.2033,
+      "step": 2264
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.000833104274380716,
+      "loss": 4.0993,
+      "step": 2265
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008329496402763273,
+      "loss": 3.9782,
+      "step": 2266
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008327949489340358,
+      "loss": 3.8877,
+      "step": 2267
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008326402003804343,
+      "loss": 3.9681,
+      "step": 2268
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.000832485394642126,
+      "loss": 3.9094,
+      "step": 2269
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.000832330531745724,
+      "loss": 3.9017,
+      "step": 2270
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.000832175611717851,
+      "loss": 3.9322,
+      "step": 2271
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008320206345851393,
+      "loss": 3.8042,
+      "step": 2272
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008318656003742313,
+      "loss": 4.0893,
+      "step": 2273
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008317105091117794,
+      "loss": 3.9372,
+      "step": 2274
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008315553608244453,
+      "loss": 3.9208,
+      "step": 2275
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008314001555389012,
+      "loss": 3.8366,
+      "step": 2276
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008312448932818283,
+      "loss": 3.8813,
+      "step": 2277
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008310895740799181,
+      "loss": 4.1622,
+      "step": 2278
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008309341979598719,
+      "loss": 3.9332,
+      "step": 2279
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008307787649484004,
+      "loss": 4.0746,
+      "step": 2280
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008306232750722247,
+      "loss": 4.035,
+      "step": 2281
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008304677283580749,
+      "loss": 4.0216,
+      "step": 2282
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008303121248326915,
+      "loss": 3.9148,
+      "step": 2283
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008301564645228248,
+      "loss": 3.9211,
+      "step": 2284
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008300007474552341,
+      "loss": 4.0201,
+      "step": 2285
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008298449736566893,
+      "loss": 3.9867,
+      "step": 2286
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008296891431539695,
+      "loss": 4.1515,
+      "step": 2287
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008295332559738641,
+      "loss": 4.0318,
+      "step": 2288
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008293773121431716,
+      "loss": 4.0711,
+      "step": 2289
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008292213116887008,
+      "loss": 3.9554,
+      "step": 2290
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008290652546372696,
+      "loss": 4.0104,
+      "step": 2291
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008289091410157066,
+      "loss": 4.0811,
+      "step": 2292
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.000828752970850849,
+      "loss": 3.8357,
+      "step": 2293
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008285967441695446,
+      "loss": 3.9138,
+      "step": 2294
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008284404609986504,
+      "loss": 3.9814,
+      "step": 2295
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008282841213650332,
+      "loss": 4.0115,
+      "step": 2296
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008281277252955699,
+      "loss": 4.0559,
+      "step": 2297
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008279712728171466,
+      "loss": 3.9859,
+      "step": 2298
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008278147639566595,
+      "loss": 4.0385,
+      "step": 2299
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008276581987410139,
+      "loss": 3.9971,
+      "step": 2300
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008275015771971255,
+      "loss": 3.7941,
+      "step": 2301
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008273448993519193,
+      "loss": 4.0668,
+      "step": 2302
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008271881652323298,
+      "loss": 4.0252,
+      "step": 2303
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008270313748653018,
+      "loss": 3.932,
+      "step": 2304
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008268745282777891,
+      "loss": 4.0074,
+      "step": 2305
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008267176254967556,
+      "loss": 4.0758,
+      "step": 2306
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008265606665491745,
+      "loss": 3.8423,
+      "step": 2307
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008264036514620293,
+      "loss": 3.7917,
+      "step": 2308
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008262465802623121,
+      "loss": 4.0117,
+      "step": 2309
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008260894529770257,
+      "loss": 3.9147,
+      "step": 2310
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008259322696331819,
+      "loss": 3.8716,
+      "step": 2311
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008257750302578026,
+      "loss": 4.0386,
+      "step": 2312
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008256177348779188,
+      "loss": 3.9582,
+      "step": 2313
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008254603835205714,
+      "loss": 4.0209,
+      "step": 2314
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.000825302976212811,
+      "loss": 3.9907,
+      "step": 2315
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008251455129816979,
+      "loss": 4.007,
+      "step": 2316
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008249879938543017,
+      "loss": 3.7585,
+      "step": 2317
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008248304188577017,
+      "loss": 3.948,
+      "step": 2318
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008246727880189869,
+      "loss": 4.0544,
+      "step": 2319
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008245151013652561,
+      "loss": 4.1454,
+      "step": 2320
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008243573589236171,
+      "loss": 3.9092,
+      "step": 2321
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008241995607211879,
+      "loss": 3.9198,
+      "step": 2322
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008240417067850957,
+      "loss": 4.0626,
+      "step": 2323
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008238837971424775,
+      "loss": 3.9448,
+      "step": 2324
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008237258318204799,
+      "loss": 4.0579,
+      "step": 2325
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008235678108462587,
+      "loss": 3.8667,
+      "step": 2326
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008234097342469798,
+      "loss": 3.8567,
+      "step": 2327
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008232516020498183,
+      "loss": 3.8073,
+      "step": 2328
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008230934142819588,
+      "loss": 3.9332,
+      "step": 2329
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.000822935170970596,
+      "loss": 3.869,
+      "step": 2330
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008227768721429334,
+      "loss": 3.8794,
+      "step": 2331
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008226185178261846,
+      "loss": 3.9648,
+      "step": 2332
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008224601080475723,
+      "loss": 3.9758,
+      "step": 2333
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008223016428343293,
+      "loss": 4.0414,
+      "step": 2334
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008221431222136975,
+      "loss": 4.0277,
+      "step": 2335
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008219845462129283,
+      "loss": 3.9684,
+      "step": 2336
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008218259148592828,
+      "loss": 3.8675,
+      "step": 2337
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008216672281800316,
+      "loss": 3.9772,
+      "step": 2338
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008215084862024549,
+      "loss": 3.8298,
+      "step": 2339
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008213496889538421,
+      "loss": 4.096,
+      "step": 2340
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008211908364614924,
+      "loss": 4.0465,
+      "step": 2341
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008210319287527143,
+      "loss": 3.9058,
+      "step": 2342
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.000820872965854826,
+      "loss": 4.0123,
+      "step": 2343
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008207139477951549,
+      "loss": 3.8967,
+      "step": 2344
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008205548746010382,
+      "loss": 3.8489,
+      "step": 2345
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008203957462998224,
+      "loss": 4.0199,
+      "step": 2346
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008202365629188633,
+      "loss": 3.8912,
+      "step": 2347
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008200773244855267,
+      "loss": 3.9396,
+      "step": 2348
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008199180310271873,
+      "loss": 4.1072,
+      "step": 2349
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008197586825712295,
+      "loss": 4.0439,
+      "step": 2350
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008195992791450474,
+      "loss": 3.8634,
+      "step": 2351
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008194398207760439,
+      "loss": 3.9083,
+      "step": 2352
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008192803074916321,
+      "loss": 4.0236,
+      "step": 2353
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008191207393192339,
+      "loss": 4.0332,
+      "step": 2354
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008189611162862812,
+      "loss": 4.0142,
+      "step": 2355
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008188014384202147,
+      "loss": 3.9377,
+      "step": 2356
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008186417057484852,
+      "loss": 3.9374,
+      "step": 2357
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008184819182985524,
+      "loss": 3.9241,
+      "step": 2358
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008183220760978858,
+      "loss": 3.9495,
+      "step": 2359
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.000818162179173964,
+      "loss": 4.1245,
+      "step": 2360
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008180022275542749,
+      "loss": 4.0877,
+      "step": 2361
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008178422212663166,
+      "loss": 3.9406,
+      "step": 2362
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008176821603375954,
+      "loss": 4.0068,
+      "step": 2363
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008175220447956281,
+      "loss": 3.9608,
+      "step": 2364
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008173618746679406,
+      "loss": 4.0054,
+      "step": 2365
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008172016499820671,
+      "loss": 4.0963,
+      "step": 2366
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.000817041370765553,
+      "loss": 3.9055,
+      "step": 2367
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008168810370459518,
+      "loss": 3.9599,
+      "step": 2368
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008167206488508267,
+      "loss": 3.973,
+      "step": 2369
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008165602062077502,
+      "loss": 3.9488,
+      "step": 2370
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008163997091443045,
+      "loss": 3.8622,
+      "step": 2371
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008162391576880808,
+      "loss": 4.0369,
+      "step": 2372
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008160785518666796,
+      "loss": 3.8432,
+      "step": 2373
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008159178917077111,
+      "loss": 4.0145,
+      "step": 2374
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008157571772387946,
+      "loss": 3.8793,
+      "step": 2375
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008155964084875587,
+      "loss": 3.9395,
+      "step": 2376
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008154355854816416,
+      "loss": 4.1246,
+      "step": 2377
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008152747082486904,
+      "loss": 3.9749,
+      "step": 2378
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008151137768163619,
+      "loss": 3.8984,
+      "step": 2379
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.000814952791212322,
+      "loss": 3.9859,
+      "step": 2380
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008147917514642462,
+      "loss": 3.9451,
+      "step": 2381
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008146306575998188,
+      "loss": 4.1009,
+      "step": 2382
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008144695096467341,
+      "loss": 4.0218,
+      "step": 2383
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008143083076326947,
+      "loss": 3.927,
+      "step": 2384
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008141470515854135,
+      "loss": 3.8512,
+      "step": 2385
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008139857415326124,
+      "loss": 3.9483,
+      "step": 2386
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008138243775020223,
+      "loss": 3.9267,
+      "step": 2387
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008136629595213834,
+      "loss": 3.8416,
+      "step": 2388
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008135014876184454,
+      "loss": 3.9995,
+      "step": 2389
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008133399618209674,
+      "loss": 3.9715,
+      "step": 2390
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008131783821567174,
+      "loss": 3.9329,
+      "step": 2391
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008130167486534729,
+      "loss": 3.9482,
+      "step": 2392
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008128550613390205,
+      "loss": 3.9944,
+      "step": 2393
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008126933202411559,
+      "loss": 3.9459,
+      "step": 2394
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008125315253876849,
+      "loss": 3.8735,
+      "step": 2395
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008123696768064212,
+      "loss": 3.8474,
+      "step": 2396
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008122077745251888,
+      "loss": 3.9614,
+      "step": 2397
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008120458185718205,
+      "loss": 3.9934,
+      "step": 2398
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008118838089741585,
+      "loss": 3.8315,
+      "step": 2399
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008117217457600541,
+      "loss": 3.8888,
+      "step": 2400
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008115596289573678,
+      "loss": 3.9715,
+      "step": 2401
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008113974585939694,
+      "loss": 3.9451,
+      "step": 2402
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008112352346977377,
+      "loss": 3.928,
+      "step": 2403
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008110729572965612,
+      "loss": 3.8904,
+      "step": 2404
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008109106264183369,
+      "loss": 3.9607,
+      "step": 2405
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008107482420909718,
+      "loss": 3.8602,
+      "step": 2406
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008105858043423811,
+      "loss": 4.0617,
+      "step": 2407
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008104233132004901,
+      "loss": 4.0862,
+      "step": 2408
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008102607686932328,
+      "loss": 3.9999,
+      "step": 2409
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008100981708485526,
+      "loss": 3.8126,
+      "step": 2410
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008099355196944018,
+      "loss": 3.9648,
+      "step": 2411
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.000809772815258742,
+      "loss": 4.0521,
+      "step": 2412
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008096100575695442,
+      "loss": 3.9682,
+      "step": 2413
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008094472466547882,
+      "loss": 4.0446,
+      "step": 2414
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008092843825424629,
+      "loss": 4.038,
+      "step": 2415
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008091214652605669,
+      "loss": 4.0325,
+      "step": 2416
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008089584948371074,
+      "loss": 4.042,
+      "step": 2417
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008087954713001007,
+      "loss": 3.917,
+      "step": 2418
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008086323946775727,
+      "loss": 3.9254,
+      "step": 2419
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008084692649975582,
+      "loss": 3.78,
+      "step": 2420
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008083060822881008,
+      "loss": 4.0023,
+      "step": 2421
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008081428465772538,
+      "loss": 3.9686,
+      "step": 2422
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008079795578930791,
+      "loss": 3.9004,
+      "step": 2423
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.000807816216263648,
+      "loss": 4.0286,
+      "step": 2424
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008076528217170408,
+      "loss": 3.9461,
+      "step": 2425
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008074893742813469,
+      "loss": 3.9381,
+      "step": 2426
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008073258739846649,
+      "loss": 3.94,
+      "step": 2427
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008071623208551023,
+      "loss": 3.8753,
+      "step": 2428
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008069987149207759,
+      "loss": 4.0205,
+      "step": 2429
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008068350562098112,
+      "loss": 3.9044,
+      "step": 2430
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008066713447503433,
+      "loss": 3.9553,
+      "step": 2431
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.000806507580570516,
+      "loss": 3.9525,
+      "step": 2432
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008063437636984823,
+      "loss": 3.8759,
+      "step": 2433
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008061798941624041,
+      "loss": 3.9092,
+      "step": 2434
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008060159719904525,
+      "loss": 3.8505,
+      "step": 2435
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008058519972108079,
+      "loss": 4.0154,
+      "step": 2436
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008056879698516591,
+      "loss": 3.9851,
+      "step": 2437
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008055238899412045,
+      "loss": 3.9791,
+      "step": 2438
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008053597575076512,
+      "loss": 3.8705,
+      "step": 2439
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008051955725792155,
+      "loss": 3.8342,
+      "step": 2440
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008050313351841228,
+      "loss": 3.9756,
+      "step": 2441
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008048670453506074,
+      "loss": 3.9592,
+      "step": 2442
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008047027031069126,
+      "loss": 3.814,
+      "step": 2443
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008045383084812906,
+      "loss": 4.0419,
+      "step": 2444
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008043738615020029,
+      "loss": 3.9555,
+      "step": 2445
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008042093621973198,
+      "loss": 3.9416,
+      "step": 2446
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008040448105955208,
+      "loss": 4.0738,
+      "step": 2447
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008038802067248939,
+      "loss": 3.9399,
+      "step": 2448
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008037155506137366,
+      "loss": 3.9157,
+      "step": 2449
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008035508422903553,
+      "loss": 4.0807,
+      "step": 2450
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.000803386081783065,
+      "loss": 3.9781,
+      "step": 2451
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008032212691201901,
+      "loss": 3.8852,
+      "step": 2452
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008030564043300639,
+      "loss": 4.0372,
+      "step": 2453
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008028914874410283,
+      "loss": 3.9945,
+      "step": 2454
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008027265184814348,
+      "loss": 3.8681,
+      "step": 2455
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008025614974796431,
+      "loss": 3.9674,
+      "step": 2456
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008023964244640224,
+      "loss": 3.8838,
+      "step": 2457
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008022312994629508,
+      "loss": 3.8455,
+      "step": 2458
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008020661225048149,
+      "loss": 4.0864,
+      "step": 2459
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008019008936180108,
+      "loss": 4.0008,
+      "step": 2460
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0008017356128309431,
+      "loss": 4.0164,
+      "step": 2461
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0008015702801720255,
+      "loss": 3.8691,
+      "step": 2462
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0008014048956696807,
+      "loss": 3.8373,
+      "step": 2463
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0008012394593523403,
+      "loss": 3.9386,
+      "step": 2464
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0008010739712484443,
+      "loss": 3.9852,
+      "step": 2465
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0008009084313864423,
+      "loss": 3.9308,
+      "step": 2466
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0008007428397947925,
+      "loss": 3.9772,
+      "step": 2467
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0008005771965019622,
+      "loss": 3.9227,
+      "step": 2468
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0008004115015364272,
+      "loss": 4.1179,
+      "step": 2469
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0008002457549266724,
+      "loss": 3.9367,
+      "step": 2470
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0008000799567011914,
+      "loss": 3.9116,
+      "step": 2471
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007999141068884872,
+      "loss": 3.9445,
+      "step": 2472
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007997482055170712,
+      "loss": 3.9457,
+      "step": 2473
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007995822526154636,
+      "loss": 3.9186,
+      "step": 2474
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007994162482121936,
+      "loss": 4.02,
+      "step": 2475
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007992501923357997,
+      "loss": 3.808,
+      "step": 2476
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007990840850148282,
+      "loss": 3.9722,
+      "step": 2477
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007989179262778355,
+      "loss": 3.8308,
+      "step": 2478
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007987517161533857,
+      "loss": 3.8275,
+      "step": 2479
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007985854546700525,
+      "loss": 4.0212,
+      "step": 2480
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007984191418564182,
+      "loss": 3.8898,
+      "step": 2481
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007982527777410737,
+      "loss": 3.9639,
+      "step": 2482
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007980863623526193,
+      "loss": 4.1155,
+      "step": 2483
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007979198957196634,
+      "loss": 3.9016,
+      "step": 2484
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007977533778708236,
+      "loss": 3.9984,
+      "step": 2485
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007975868088347263,
+      "loss": 3.9873,
+      "step": 2486
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007974201886400066,
+      "loss": 4.0728,
+      "step": 2487
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007972535173153086,
+      "loss": 3.9669,
+      "step": 2488
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007970867948892848,
+      "loss": 4.0217,
+      "step": 2489
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.000796920021390597,
+      "loss": 3.8585,
+      "step": 2490
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007967531968479151,
+      "loss": 3.9939,
+      "step": 2491
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007965863212899185,
+      "loss": 4.0331,
+      "step": 2492
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007964193947452949,
+      "loss": 3.924,
+      "step": 2493
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007962524172427411,
+      "loss": 3.8901,
+      "step": 2494
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007960853888109622,
+      "loss": 3.8953,
+      "step": 2495
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007959183094786725,
+      "loss": 4.0585,
+      "step": 2496
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007957511792745949,
+      "loss": 3.8729,
+      "step": 2497
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007955839982274607,
+      "loss": 3.9054,
+      "step": 2498
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007954167663660108,
+      "loss": 3.9503,
+      "step": 2499
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007952494837189937,
+      "loss": 3.9497,
+      "step": 2500
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007950821503151678,
+      "loss": 4.0946,
+      "step": 2501
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007949147661832992,
+      "loss": 4.1051,
+      "step": 2502
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007947473313521634,
+      "loss": 4.1566,
+      "step": 2503
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007945798458505442,
+      "loss": 3.9379,
+      "step": 2504
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007944123097072347,
+      "loss": 4.097,
+      "step": 2505
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007942447229510359,
+      "loss": 3.8674,
+      "step": 2506
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.000794077085610758,
+      "loss": 4.0345,
+      "step": 2507
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00079390939771522,
+      "loss": 3.9253,
+      "step": 2508
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.000793741659293249,
+      "loss": 3.9391,
+      "step": 2509
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007935738703736817,
+      "loss": 3.9788,
+      "step": 2510
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007934060309853626,
+      "loss": 3.8692,
+      "step": 2511
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007932381411571453,
+      "loss": 3.8721,
+      "step": 2512
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007930702009178918,
+      "loss": 4.0073,
+      "step": 2513
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007929022102964734,
+      "loss": 3.9265,
+      "step": 2514
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007927341693217695,
+      "loss": 4.1179,
+      "step": 2515
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007925660780226683,
+      "loss": 3.8617,
+      "step": 2516
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007923979364280663,
+      "loss": 3.8559,
+      "step": 2517
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007922297445668694,
+      "loss": 3.8867,
+      "step": 2518
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007920615024679916,
+      "loss": 3.9142,
+      "step": 2519
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007918932101603555,
+      "loss": 4.0161,
+      "step": 2520
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007917248676728929,
+      "loss": 3.929,
+      "step": 2521
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007915564750345435,
+      "loss": 4.0751,
+      "step": 2522
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007913880322742559,
+      "loss": 3.9881,
+      "step": 2523
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007912195394209877,
+      "loss": 3.8573,
+      "step": 2524
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007910509965037045,
+      "loss": 3.9446,
+      "step": 2525
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007908824035513808,
+      "loss": 3.9527,
+      "step": 2526
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007907137605929997,
+      "loss": 3.9022,
+      "step": 2527
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.000790545067657553,
+      "loss": 3.9222,
+      "step": 2528
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007903763247740407,
+      "loss": 3.9206,
+      "step": 2529
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007902075319714721,
+      "loss": 3.9952,
+      "step": 2530
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007900386892788642,
+      "loss": 4.0448,
+      "step": 2531
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007898697967252433,
+      "loss": 3.9872,
+      "step": 2532
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007897008543396438,
+      "loss": 4.032,
+      "step": 2533
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.000789531862151109,
+      "loss": 3.9062,
+      "step": 2534
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007893628201886907,
+      "loss": 4.0232,
+      "step": 2535
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007891937284814488,
+      "loss": 3.8137,
+      "step": 2536
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007890245870584522,
+      "loss": 3.8593,
+      "step": 2537
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007888553959487787,
+      "loss": 4.0089,
+      "step": 2538
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007886861551815138,
+      "loss": 3.8739,
+      "step": 2539
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007885168647857524,
+      "loss": 3.9724,
+      "step": 2540
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007883475247905969,
+      "loss": 3.9062,
+      "step": 2541
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007881781352251591,
+      "loss": 3.9424,
+      "step": 2542
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007880086961185589,
+      "loss": 4.0368,
+      "step": 2543
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007878392074999253,
+      "loss": 3.9084,
+      "step": 2544
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007876696693983946,
+      "loss": 4.0196,
+      "step": 2545
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.000787500081843113,
+      "loss": 3.8561,
+      "step": 2546
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007873304448632344,
+      "loss": 3.852,
+      "step": 2547
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007871607584879211,
+      "loss": 3.8928,
+      "step": 2548
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007869910227463446,
+      "loss": 4.0079,
+      "step": 2549
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.000786821237667684,
+      "loss": 4.0662,
+      "step": 2550
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007866514032811276,
+      "loss": 3.826,
+      "step": 2551
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007864815196158718,
+      "loss": 3.9198,
+      "step": 2552
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007863115867011216,
+      "loss": 3.8503,
+      "step": 2553
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007861416045660905,
+      "loss": 4.1262,
+      "step": 2554
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007859715732400004,
+      "loss": 3.8803,
+      "step": 2555
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007858014927520814,
+      "loss": 3.9676,
+      "step": 2556
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007856313631315726,
+      "loss": 3.9496,
+      "step": 2557
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007854611844077213,
+      "loss": 4.1251,
+      "step": 2558
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007852909566097828,
+      "loss": 3.9822,
+      "step": 2559
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007851206797670216,
+      "loss": 3.8148,
+      "step": 2560
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007849503539087101,
+      "loss": 3.8778,
+      "step": 2561
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007847799790641295,
+      "loss": 3.906,
+      "step": 2562
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007846095552625688,
+      "loss": 3.9651,
+      "step": 2563
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007844390825333263,
+      "loss": 4.0091,
+      "step": 2564
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007842685609057077,
+      "loss": 3.9301,
+      "step": 2565
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007840979904090282,
+      "loss": 3.8521,
+      "step": 2566
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007839273710726107,
+      "loss": 3.8874,
+      "step": 2567
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007837567029257864,
+      "loss": 3.9428,
+      "step": 2568
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007835859859978951,
+      "loss": 3.8644,
+      "step": 2569
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007834152203182854,
+      "loss": 4.0804,
+      "step": 2570
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007832444059163137,
+      "loss": 4.0252,
+      "step": 2571
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007830735428213451,
+      "loss": 3.8233,
+      "step": 2572
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007829026310627527,
+      "loss": 4.049,
+      "step": 2573
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007827316706699181,
+      "loss": 3.8957,
+      "step": 2574
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007825606616722318,
+      "loss": 4.1049,
+      "step": 2575
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007823896040990921,
+      "loss": 4.0016,
+      "step": 2576
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007822184979799057,
+      "loss": 3.9764,
+      "step": 2577
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007820473433440876,
+      "loss": 3.9226,
+      "step": 2578
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007818761402210615,
+      "loss": 4.0353,
+      "step": 2579
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007817048886402592,
+      "loss": 3.9156,
+      "step": 2580
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007815335886311206,
+      "loss": 3.9536,
+      "step": 2581
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007813622402230943,
+      "loss": 4.0216,
+      "step": 2582
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007811908434456372,
+      "loss": 3.8143,
+      "step": 2583
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007810193983282141,
+      "loss": 3.9198,
+      "step": 2584
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007808479049002987,
+      "loss": 4.0264,
+      "step": 2585
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007806763631913725,
+      "loss": 3.8785,
+      "step": 2586
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007805047732309256,
+      "loss": 3.8329,
+      "step": 2587
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007803331350484563,
+      "loss": 3.8827,
+      "step": 2588
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007801614486734711,
+      "loss": 3.8696,
+      "step": 2589
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007799897141354853,
+      "loss": 3.9347,
+      "step": 2590
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007798179314640214,
+      "loss": 3.942,
+      "step": 2591
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.000779646100688611,
+      "loss": 3.8611,
+      "step": 2592
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007794742218387943,
+      "loss": 4.0615,
+      "step": 2593
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007793022949441189,
+      "loss": 3.877,
+      "step": 2594
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.000779130320034141,
+      "loss": 4.0263,
+      "step": 2595
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.000778958297138425,
+      "loss": 3.938,
+      "step": 2596
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007787862262865438,
+      "loss": 3.8292,
+      "step": 2597
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007786141075080785,
+      "loss": 3.9721,
+      "step": 2598
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.000778441940832618,
+      "loss": 4.0352,
+      "step": 2599
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00077826972628976,
+      "loss": 3.9366,
+      "step": 2600
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00077809746390911,
+      "loss": 4.0174,
+      "step": 2601
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007779251537202818,
+      "loss": 3.9301,
+      "step": 2602
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007777527957528981,
+      "loss": 3.9726,
+      "step": 2603
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007775803900365885,
+      "loss": 3.9508,
+      "step": 2604
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007774079366009919,
+      "loss": 4.006,
+      "step": 2605
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007772354354757549,
+      "loss": 3.9195,
+      "step": 2606
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007770628866905329,
+      "loss": 4.0013,
+      "step": 2607
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007768902902749884,
+      "loss": 3.8948,
+      "step": 2608
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007767176462587931,
+      "loss": 4.0399,
+      "step": 2609
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007765449546716266,
+      "loss": 3.8275,
+      "step": 2610
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007763722155431761,
+      "loss": 3.9485,
+      "step": 2611
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.000776199428903138,
+      "loss": 3.8771,
+      "step": 2612
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.000776026594781216,
+      "loss": 3.9325,
+      "step": 2613
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007758537132071224,
+      "loss": 3.8466,
+      "step": 2614
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007756807842105774,
+      "loss": 3.9265,
+      "step": 2615
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007755078078213098,
+      "loss": 3.9325,
+      "step": 2616
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.000775334784069056,
+      "loss": 3.9369,
+      "step": 2617
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007751617129835609,
+      "loss": 4.0043,
+      "step": 2618
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007749885945945774,
+      "loss": 3.9963,
+      "step": 2619
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007748154289318665,
+      "loss": 4.1009,
+      "step": 2620
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007746422160251975,
+      "loss": 3.9146,
+      "step": 2621
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007744689559043475,
+      "loss": 4.0355,
+      "step": 2622
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007742956485991022,
+      "loss": 3.9689,
+      "step": 2623
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.000774122294139255,
+      "loss": 3.8316,
+      "step": 2624
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007739488925546075,
+      "loss": 4.0469,
+      "step": 2625
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007737754438749693,
+      "loss": 3.8585,
+      "step": 2626
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007736019481301588,
+      "loss": 4.0109,
+      "step": 2627
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007734284053500013,
+      "loss": 3.9439,
+      "step": 2628
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.000773254815564331,
+      "loss": 4.0457,
+      "step": 2629
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007730811788029902,
+      "loss": 3.901,
+      "step": 2630
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007729074950958289,
+      "loss": 4.0468,
+      "step": 2631
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007727337644727055,
+      "loss": 4.0186,
+      "step": 2632
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.000772559986963486,
+      "loss": 3.9528,
+      "step": 2633
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.000772386162598045,
+      "loss": 4.0246,
+      "step": 2634
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007722122914062649,
+      "loss": 3.9131,
+      "step": 2635
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007720383734180361,
+      "loss": 3.9869,
+      "step": 2636
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007718644086632573,
+      "loss": 4.0174,
+      "step": 2637
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.000771690397171835,
+      "loss": 4.0199,
+      "step": 2638
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007715163389736834,
+      "loss": 3.9274,
+      "step": 2639
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007713422340987257,
+      "loss": 3.8286,
+      "step": 2640
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007711680825768921,
+      "loss": 3.9249,
+      "step": 2641
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007709938844381217,
+      "loss": 4.0234,
+      "step": 2642
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007708196397123607,
+      "loss": 3.9284,
+      "step": 2643
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007706453484295641,
+      "loss": 4.0229,
+      "step": 2644
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007704710106196946,
+      "loss": 3.913,
+      "step": 2645
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007702966263127227,
+      "loss": 3.9223,
+      "step": 2646
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007701221955386273,
+      "loss": 3.9729,
+      "step": 2647
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007699477183273948,
+      "loss": 3.9954,
+      "step": 2648
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00076977319470902,
+      "loss": 3.9481,
+      "step": 2649
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007695986247135059,
+      "loss": 3.9613,
+      "step": 2650
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007694240083708624,
+      "loss": 4.0356,
+      "step": 2651
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007692493457111085,
+      "loss": 4.0191,
+      "step": 2652
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007690746367642707,
+      "loss": 3.9393,
+      "step": 2653
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007688998815603836,
+      "loss": 3.919,
+      "step": 2654
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007687250801294892,
+      "loss": 4.0731,
+      "step": 2655
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007685502325016385,
+      "loss": 3.9903,
+      "step": 2656
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007683753387068893,
+      "loss": 4.1125,
+      "step": 2657
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007682003987753083,
+      "loss": 3.9995,
+      "step": 2658
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007680254127369695,
+      "loss": 3.9809,
+      "step": 2659
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007678503806219549,
+      "loss": 3.9334,
+      "step": 2660
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007676753024603546,
+      "loss": 3.8932,
+      "step": 2661
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007675001782822668,
+      "loss": 4.01,
+      "step": 2662
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007673250081177972,
+      "loss": 4.0128,
+      "step": 2663
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007671497919970597,
+      "loss": 3.918,
+      "step": 2664
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007669745299501758,
+      "loss": 3.8448,
+      "step": 2665
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007667992220072751,
+      "loss": 3.867,
+      "step": 2666
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007666238681984952,
+      "loss": 4.0197,
+      "step": 2667
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007664484685539814,
+      "loss": 3.968,
+      "step": 2668
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007662730231038869,
+      "loss": 3.8664,
+      "step": 2669
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007660975318783729,
+      "loss": 3.9624,
+      "step": 2670
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007659219949076081,
+      "loss": 4.0364,
+      "step": 2671
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007657464122217697,
+      "loss": 3.915,
+      "step": 2672
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007655707838510423,
+      "loss": 3.8902,
+      "step": 2673
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007653951098256183,
+      "loss": 3.8587,
+      "step": 2674
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007652193901756983,
+      "loss": 3.8565,
+      "step": 2675
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007650436249314903,
+      "loss": 3.8916,
+      "step": 2676
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007648678141232106,
+      "loss": 4.0568,
+      "step": 2677
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.000764691957781083,
+      "loss": 4.0688,
+      "step": 2678
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007645160559353393,
+      "loss": 4.1089,
+      "step": 2679
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.000764340108616219,
+      "loss": 4.047,
+      "step": 2680
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007641641158539697,
+      "loss": 3.8708,
+      "step": 2681
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007639880776788464,
+      "loss": 3.9654,
+      "step": 2682
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007638119941211119,
+      "loss": 3.979,
+      "step": 2683
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007636358652110375,
+      "loss": 4.0487,
+      "step": 2684
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007634596909789013,
+      "loss": 3.8555,
+      "step": 2685
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007632834714549901,
+      "loss": 3.8608,
+      "step": 2686
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007631072066695978,
+      "loss": 4.12,
+      "step": 2687
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007629308966530268,
+      "loss": 3.7861,
+      "step": 2688
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007627545414355861,
+      "loss": 3.986,
+      "step": 2689
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007625781410475938,
+      "loss": 4.0421,
+      "step": 2690
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.000762401695519375,
+      "loss": 3.9217,
+      "step": 2691
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007622252048812628,
+      "loss": 3.8682,
+      "step": 2692
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007620486691635977,
+      "loss": 3.9289,
+      "step": 2693
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007618720883967285,
+      "loss": 3.895,
+      "step": 2694
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007616954626110115,
+      "loss": 3.8673,
+      "step": 2695
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007615187918368106,
+      "loss": 4.0853,
+      "step": 2696
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007613420761044976,
+      "loss": 3.8913,
+      "step": 2697
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.000761165315444452,
+      "loss": 3.9622,
+      "step": 2698
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007609885098870609,
+      "loss": 3.9602,
+      "step": 2699
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007608116594627195,
+      "loss": 4.0418,
+      "step": 2700
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007606347642018301,
+      "loss": 4.0106,
+      "step": 2701
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007604578241348032,
+      "loss": 3.9865,
+      "step": 2702
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007602808392920569,
+      "loss": 3.9374,
+      "step": 2703
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007601038097040169,
+      "loss": 4.0358,
+      "step": 2704
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007599267354011166,
+      "loss": 3.9675,
+      "step": 2705
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007597496164137972,
+      "loss": 3.9687,
+      "step": 2706
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007595724527725074,
+      "loss": 3.9743,
+      "step": 2707
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007593952445077035,
+      "loss": 4.066,
+      "step": 2708
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007592179916498503,
+      "loss": 3.8465,
+      "step": 2709
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007590406942294191,
+      "loss": 3.9563,
+      "step": 2710
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007588633522768895,
+      "loss": 3.9767,
+      "step": 2711
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007586859658227486,
+      "loss": 3.8771,
+      "step": 2712
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007585085348974912,
+      "loss": 4.0146,
+      "step": 2713
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007583310595316199,
+      "loss": 4.0591,
+      "step": 2714
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007581535397556445,
+      "loss": 3.9208,
+      "step": 2715
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.000757975975600083,
+      "loss": 3.824,
+      "step": 2716
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007577983670954603,
+      "loss": 3.7023,
+      "step": 2717
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007576207142723099,
+      "loss": 3.8708,
+      "step": 2718
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.000757443017161172,
+      "loss": 3.7574,
+      "step": 2719
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007572652757925951,
+      "loss": 3.9586,
+      "step": 2720
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007570874901971347,
+      "loss": 3.9177,
+      "step": 2721
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007569096604053543,
+      "loss": 4.1819,
+      "step": 2722
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007567317864478247,
+      "loss": 3.8153,
+      "step": 2723
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007565538683551251,
+      "loss": 3.8915,
+      "step": 2724
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007563759061578411,
+      "loss": 3.9975,
+      "step": 2725
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007561978998865666,
+      "loss": 3.9096,
+      "step": 2726
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007560198495719032,
+      "loss": 3.9216,
+      "step": 2727
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007558417552444596,
+      "loss": 3.9271,
+      "step": 2728
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007556636169348523,
+      "loss": 3.9415,
+      "step": 2729
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007554854346737052,
+      "loss": 3.8944,
+      "step": 2730
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00075530720849165,
+      "loss": 4.1597,
+      "step": 2731
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007551289384193261,
+      "loss": 3.9031,
+      "step": 2732
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007549506244873799,
+      "loss": 3.9459,
+      "step": 2733
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007547722667264656,
+      "loss": 4.0071,
+      "step": 2734
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007545938651672453,
+      "loss": 3.9804,
+      "step": 2735
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007544154198403879,
+      "loss": 3.9621,
+      "step": 2736
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007542369307765704,
+      "loss": 4.1172,
+      "step": 2737
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007540583980064771,
+      "loss": 4.0183,
+      "step": 2738
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007538798215608,
+      "loss": 3.9363,
+      "step": 2739
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007537012014702382,
+      "loss": 3.9533,
+      "step": 2740
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007535225377654989,
+      "loss": 3.9671,
+      "step": 2741
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007533438304772962,
+      "loss": 3.962,
+      "step": 2742
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007531650796363521,
+      "loss": 3.9541,
+      "step": 2743
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007529862852733957,
+      "loss": 4.0831,
+      "step": 2744
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.000752807447419164,
+      "loss": 3.9927,
+      "step": 2745
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007526285661044014,
+      "loss": 3.8377,
+      "step": 2746
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007524496413598595,
+      "loss": 3.9365,
+      "step": 2747
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007522706732162976,
+      "loss": 4.0687,
+      "step": 2748
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007520916617044825,
+      "loss": 4.1612,
+      "step": 2749
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007519126068551882,
+      "loss": 3.9841,
+      "step": 2750
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007517335086991964,
+      "loss": 4.0157,
+      "step": 2751
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.000751554367267296,
+      "loss": 4.1203,
+      "step": 2752
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007513751825902837,
+      "loss": 3.8247,
+      "step": 2753
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007511959546989632,
+      "loss": 3.8367,
+      "step": 2754
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.000751016683624146,
+      "loss": 4.0784,
+      "step": 2755
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007508373693966509,
+      "loss": 3.9402,
+      "step": 2756
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007506580120473041,
+      "loss": 3.9777,
+      "step": 2757
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007504786116069391,
+      "loss": 4.0049,
+      "step": 2758
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007502991681063969,
+      "loss": 3.9889,
+      "step": 2759
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007501196815765261,
+      "loss": 4.016,
+      "step": 2760
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007499401520481824,
+      "loss": 3.8748,
+      "step": 2761
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007497605795522288,
+      "loss": 3.9614,
+      "step": 2762
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007495809641195364,
+      "loss": 3.8381,
+      "step": 2763
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007494013057809827,
+      "loss": 3.9545,
+      "step": 2764
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007492216045674532,
+      "loss": 4.049,
+      "step": 2765
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007490418605098408,
+      "loss": 3.89,
+      "step": 2766
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007488620736390454,
+      "loss": 3.972,
+      "step": 2767
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007486822439859744,
+      "loss": 3.883,
+      "step": 2768
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007485023715815426,
+      "loss": 3.944,
+      "step": 2769
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007483224564566724,
+      "loss": 4.1068,
+      "step": 2770
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007481424986422932,
+      "loss": 3.905,
+      "step": 2771
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007479624981693416,
+      "loss": 3.958,
+      "step": 2772
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.000747782455068762,
+      "loss": 3.7514,
+      "step": 2773
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007476023693715058,
+      "loss": 3.8436,
+      "step": 2774
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.000747422241108532,
+      "loss": 4.0183,
+      "step": 2775
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007472420703108064,
+      "loss": 4.0263,
+      "step": 2776
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007470618570093026,
+      "loss": 4.0492,
+      "step": 2777
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007468816012350016,
+      "loss": 3.9063,
+      "step": 2778
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007467013030188911,
+      "loss": 3.93,
+      "step": 2779
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007465209623919668,
+      "loss": 4.0185,
+      "step": 2780
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007463405793852309,
+      "loss": 3.894,
+      "step": 2781
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007461601540296937,
+      "loss": 3.9189,
+      "step": 2782
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007459796863563723,
+      "loss": 4.0698,
+      "step": 2783
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007457991763962911,
+      "loss": 3.9365,
+      "step": 2784
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007456186241804821,
+      "loss": 3.996,
+      "step": 2785
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.000745438029739984,
+      "loss": 3.8728,
+      "step": 2786
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007452573931058432,
+      "loss": 3.8889,
+      "step": 2787
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007450767143091133,
+      "loss": 3.9743,
+      "step": 2788
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007448959933808551,
+      "loss": 3.981,
+      "step": 2789
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007447152303521365,
+      "loss": 3.9528,
+      "step": 2790
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007445344252540327,
+      "loss": 3.8861,
+      "step": 2791
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007443535781176265,
+      "loss": 3.9998,
+      "step": 2792
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007441726889740075,
+      "loss": 3.8559,
+      "step": 2793
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007439917578542723,
+      "loss": 3.8446,
+      "step": 2794
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007438107847895256,
+      "loss": 3.8832,
+      "step": 2795
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007436297698108782,
+      "loss": 3.9212,
+      "step": 2796
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007434487129494493,
+      "loss": 3.9337,
+      "step": 2797
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007432676142363642,
+      "loss": 4.0386,
+      "step": 2798
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007430864737027561,
+      "loss": 3.9915,
+      "step": 2799
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007429052913797648,
+      "loss": 3.8397,
+      "step": 2800
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.000742724067298538,
+      "loss": 3.9381,
+      "step": 2801
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007425428014902303,
+      "loss": 4.0076,
+      "step": 2802
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007423614939860032,
+      "loss": 3.8616,
+      "step": 2803
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007421801448170256,
+      "loss": 3.9106,
+      "step": 2804
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007419987540144733,
+      "loss": 3.8872,
+      "step": 2805
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007418173216095299,
+      "loss": 3.9314,
+      "step": 2806
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007416358476333855,
+      "loss": 3.9446,
+      "step": 2807
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007414543321172376,
+      "loss": 3.8693,
+      "step": 2808
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007412727750922908,
+      "loss": 3.8908,
+      "step": 2809
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007410911765897567,
+      "loss": 3.917,
+      "step": 2810
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007409095366408547,
+      "loss": 3.8978,
+      "step": 2811
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007407278552768103,
+      "loss": 3.9251,
+      "step": 2812
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007405461325288569,
+      "loss": 3.922,
+      "step": 2813
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007403643684282346,
+      "loss": 4.0085,
+      "step": 2814
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.000740182563006191,
+      "loss": 3.981,
+      "step": 2815
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007400007162939802,
+      "loss": 4.1201,
+      "step": 2816
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.000739818828322864,
+      "loss": 4.0245,
+      "step": 2817
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.000739636899124111,
+      "loss": 3.9831,
+      "step": 2818
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007394549287289969,
+      "loss": 3.9635,
+      "step": 2819
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007392729171688046,
+      "loss": 4.0071,
+      "step": 2820
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.000739090864474824,
+      "loss": 3.9406,
+      "step": 2821
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007389087706783518,
+      "loss": 3.911,
+      "step": 2822
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007387266358106924,
+      "loss": 3.8929,
+      "step": 2823
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007385444599031567,
+      "loss": 4.0022,
+      "step": 2824
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.000738362242987063,
+      "loss": 3.8721,
+      "step": 2825
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007381799850937363,
+      "loss": 4.1717,
+      "step": 2826
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007379976862545091,
+      "loss": 3.9192,
+      "step": 2827
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007378153465007202,
+      "loss": 3.9912,
+      "step": 2828
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007376329658637166,
+      "loss": 3.962,
+      "step": 2829
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007374505443748511,
+      "loss": 3.849,
+      "step": 2830
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007372680820654843,
+      "loss": 3.9483,
+      "step": 2831
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007370855789669837,
+      "loss": 4.0,
+      "step": 2832
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007369030351107232,
+      "loss": 3.9274,
+      "step": 2833
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007367204505280847,
+      "loss": 3.8994,
+      "step": 2834
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007365378252504566,
+      "loss": 3.9494,
+      "step": 2835
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007363551593092342,
+      "loss": 4.027,
+      "step": 2836
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007361724527358194,
+      "loss": 3.8949,
+      "step": 2837
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007359897055616225,
+      "loss": 3.9819,
+      "step": 2838
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007358069178180592,
+      "loss": 3.9744,
+      "step": 2839
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.000735624089536553,
+      "loss": 3.7839,
+      "step": 2840
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007354412207485342,
+      "loss": 4.0256,
+      "step": 2841
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007352583114854402,
+      "loss": 3.8588,
+      "step": 2842
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.000735075361778715,
+      "loss": 4.069,
+      "step": 2843
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007348923716598099,
+      "loss": 3.8536,
+      "step": 2844
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007347093411601831,
+      "loss": 4.0294,
+      "step": 2845
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007345262703112996,
+      "loss": 3.8712,
+      "step": 2846
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007343431591446312,
+      "loss": 4.0378,
+      "step": 2847
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007341600076916571,
+      "loss": 3.8067,
+      "step": 2848
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007339768159838631,
+      "loss": 4.0766,
+      "step": 2849
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007337935840527419,
+      "loss": 4.0938,
+      "step": 2850
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.000733610311929793,
+      "loss": 4.0232,
+      "step": 2851
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007334269996465235,
+      "loss": 3.9412,
+      "step": 2852
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007332436472344467,
+      "loss": 3.8283,
+      "step": 2853
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007330602547250827,
+      "loss": 4.0888,
+      "step": 2854
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007328768221499594,
+      "loss": 3.9541,
+      "step": 2855
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007326933495406103,
+      "loss": 3.9421,
+      "step": 2856
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.000732509836928577,
+      "loss": 3.8713,
+      "step": 2857
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007323262843454071,
+      "loss": 3.8733,
+      "step": 2858
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007321426918226557,
+      "loss": 3.9279,
+      "step": 2859
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007319590593918843,
+      "loss": 4.0441,
+      "step": 2860
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007317753870846614,
+      "loss": 3.954,
+      "step": 2861
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007315916749325625,
+      "loss": 3.9701,
+      "step": 2862
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007314079229671699,
+      "loss": 3.8251,
+      "step": 2863
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007312241312200727,
+      "loss": 3.979,
+      "step": 2864
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007310402997228667,
+      "loss": 4.0208,
+      "step": 2865
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007308564285071546,
+      "loss": 3.9634,
+      "step": 2866
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007306725176045464,
+      "loss": 3.966,
+      "step": 2867
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007304885670466582,
+      "loss": 3.8018,
+      "step": 2868
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.000730304576865113,
+      "loss": 3.856,
+      "step": 2869
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007301205470915412,
+      "loss": 3.8818,
+      "step": 2870
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007299364777575796,
+      "loss": 4.0132,
+      "step": 2871
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.000729752368894872,
+      "loss": 3.9784,
+      "step": 2872
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007295682205350684,
+      "loss": 3.86,
+      "step": 2873
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007293840327098264,
+      "loss": 3.9564,
+      "step": 2874
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007291998054508101,
+      "loss": 3.8791,
+      "step": 2875
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007290155387896899,
+      "loss": 3.9675,
+      "step": 2876
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007288312327581438,
+      "loss": 4.0191,
+      "step": 2877
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007286468873878559,
+      "loss": 3.9576,
+      "step": 2878
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007284625027105172,
+      "loss": 3.9486,
+      "step": 2879
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007282780787578258,
+      "loss": 3.8042,
+      "step": 2880
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007280936155614863,
+      "loss": 3.9296,
+      "step": 2881
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007279091131532101,
+      "loss": 3.8787,
+      "step": 2882
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.000727724571564715,
+      "loss": 4.0015,
+      "step": 2883
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.000727539990827726,
+      "loss": 3.9905,
+      "step": 2884
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007273553709739749,
+      "loss": 3.9172,
+      "step": 2885
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007271707120351996,
+      "loss": 3.9144,
+      "step": 2886
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007269860140431454,
+      "loss": 3.9828,
+      "step": 2887
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.000726801277029564,
+      "loss": 3.8576,
+      "step": 2888
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007266165010262138,
+      "loss": 3.9951,
+      "step": 2889
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007264316860648598,
+      "loss": 3.9076,
+      "step": 2890
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.000726246832177274,
+      "loss": 3.7768,
+      "step": 2891
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.000726061939395235,
+      "loss": 4.0216,
+      "step": 2892
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007258770077505275,
+      "loss": 3.9606,
+      "step": 2893
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.000725692037274944,
+      "loss": 3.8562,
+      "step": 2894
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007255070280002829,
+      "loss": 3.8743,
+      "step": 2895
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007253219799583495,
+      "loss": 3.9356,
+      "step": 2896
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007251368931809553,
+      "loss": 3.9326,
+      "step": 2897
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007249517676999192,
+      "loss": 4.0043,
+      "step": 2898
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007247666035470666,
+      "loss": 3.8997,
+      "step": 2899
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007245814007542289,
+      "loss": 3.8268,
+      "step": 2900
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.000724396159353245,
+      "loss": 3.9594,
+      "step": 2901
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007242108793759596,
+      "loss": 3.8883,
+      "step": 2902
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007240255608542251,
+      "loss": 4.0315,
+      "step": 2903
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007238402038198994,
+      "loss": 3.9831,
+      "step": 2904
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007236548083048478,
+      "loss": 4.0002,
+      "step": 2905
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007234693743409417,
+      "loss": 3.9186,
+      "step": 2906
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007232839019600594,
+      "loss": 4.0453,
+      "step": 2907
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007230983911940861,
+      "loss": 3.8145,
+      "step": 2908
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007229128420749127,
+      "loss": 3.9391,
+      "step": 2909
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007227272546344377,
+      "loss": 3.9321,
+      "step": 2910
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007225416289045654,
+      "loss": 4.05,
+      "step": 2911
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007223559649172073,
+      "loss": 3.7593,
+      "step": 2912
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.000722170262704281,
+      "loss": 4.0181,
+      "step": 2913
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007219845222977109,
+      "loss": 3.9664,
+      "step": 2914
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.000721798743729428,
+      "loss": 4.0862,
+      "step": 2915
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007216129270313697,
+      "loss": 3.822,
+      "step": 2916
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007214270722354801,
+      "loss": 3.8137,
+      "step": 2917
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00072124117937371,
+      "loss": 3.9044,
+      "step": 2918
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.000721055248478016,
+      "loss": 3.9167,
+      "step": 2919
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007208692795803622,
+      "loss": 4.0174,
+      "step": 2920
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007206832727127186,
+      "loss": 3.9454,
+      "step": 2921
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007204972279070622,
+      "loss": 3.9818,
+      "step": 2922
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.000720311145195376,
+      "loss": 3.8916,
+      "step": 2923
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00072012502460965,
+      "loss": 3.9142,
+      "step": 2924
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007199388661818802,
+      "loss": 4.0309,
+      "step": 2925
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007197526699440697,
+      "loss": 3.7436,
+      "step": 2926
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007195664359282275,
+      "loss": 3.9494,
+      "step": 2927
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007193801641663697,
+      "loss": 3.809,
+      "step": 2928
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007191938546905183,
+      "loss": 3.814,
+      "step": 2929
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007190075075327021,
+      "loss": 3.9954,
+      "step": 2930
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007188211227249565,
+      "loss": 3.9173,
+      "step": 2931
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007186347002993232,
+      "loss": 3.9688,
+      "step": 2932
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007184482402878501,
+      "loss": 3.9736,
+      "step": 2933
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007182617427225922,
+      "loss": 4.0778,
+      "step": 2934
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007180752076356102,
+      "loss": 3.9672,
+      "step": 2935
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007178886350589721,
+      "loss": 3.9573,
+      "step": 2936
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007177020250247514,
+      "loss": 4.1131,
+      "step": 2937
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007175153775650289,
+      "loss": 3.9929,
+      "step": 2938
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007173286927118914,
+      "loss": 4.0273,
+      "step": 2939
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.000717141970497432,
+      "loss": 4.0659,
+      "step": 2940
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007169552109537507,
+      "loss": 3.8409,
+      "step": 2941
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007167684141129536,
+      "loss": 3.9766,
+      "step": 2942
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007165815800071529,
+      "loss": 4.0145,
+      "step": 2943
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007163947086684676,
+      "loss": 3.9368,
+      "step": 2944
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007162078001290235,
+      "loss": 3.9295,
+      "step": 2945
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007160208544209521,
+      "loss": 3.9666,
+      "step": 2946
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007158338715763913,
+      "loss": 3.9072,
+      "step": 2947
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007156468516274859,
+      "loss": 3.9076,
+      "step": 2948
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007154597946063867,
+      "loss": 3.8815,
+      "step": 2949
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.000715272700545251,
+      "loss": 3.9679,
+      "step": 2950
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007150855694762424,
+      "loss": 3.9801,
+      "step": 2951
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.000714898401431531,
+      "loss": 3.888,
+      "step": 2952
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.000714711196443293,
+      "loss": 3.9187,
+      "step": 2953
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007145239545437112,
+      "loss": 3.9433,
+      "step": 2954
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007143366757649746,
+      "loss": 3.8329,
+      "step": 2955
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007141493601392787,
+      "loss": 3.969,
+      "step": 2956
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007139620076988251,
+      "loss": 3.8614,
+      "step": 2957
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007137746184758218,
+      "loss": 3.9409,
+      "step": 2958
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007135871925024835,
+      "loss": 3.8799,
+      "step": 2959
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007133997298110307,
+      "loss": 3.9502,
+      "step": 2960
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007132122304336903,
+      "loss": 4.0177,
+      "step": 2961
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007130246944026956,
+      "loss": 3.8268,
+      "step": 2962
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007128371217502867,
+      "loss": 3.8417,
+      "step": 2963
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.000712649512508709,
+      "loss": 3.855,
+      "step": 2964
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007124618667102147,
+      "loss": 3.8781,
+      "step": 2965
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007122741843870626,
+      "loss": 3.8559,
+      "step": 2966
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007120864655715171,
+      "loss": 3.8953,
+      "step": 2967
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007118987102958498,
+      "loss": 4.0174,
+      "step": 2968
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007117109185923373,
+      "loss": 3.8568,
+      "step": 2969
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007115230904932639,
+      "loss": 4.0664,
+      "step": 2970
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007113352260309188,
+      "loss": 3.9415,
+      "step": 2971
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007111473252375983,
+      "loss": 3.8634,
+      "step": 2972
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007109593881456048,
+      "loss": 3.8911,
+      "step": 2973
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.000710771414787247,
+      "loss": 3.9991,
+      "step": 2974
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007105834051948395,
+      "loss": 3.8774,
+      "step": 2975
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007103953594007029,
+      "loss": 3.8739,
+      "step": 2976
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007102072774371654,
+      "loss": 3.8846,
+      "step": 2977
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007100191593365599,
+      "loss": 4.1175,
+      "step": 2978
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.000709831005131226,
+      "loss": 4.0011,
+      "step": 2979
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00070964281485351,
+      "loss": 3.8847,
+      "step": 2980
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007094545885357635,
+      "loss": 3.9505,
+      "step": 2981
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007092663262103451,
+      "loss": 3.8738,
+      "step": 2982
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007090780279096192,
+      "loss": 3.8755,
+      "step": 2983
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007088896936659566,
+      "loss": 3.9362,
+      "step": 2984
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007087013235117338,
+      "loss": 3.9685,
+      "step": 2985
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007085129174793342,
+      "loss": 3.8109,
+      "step": 2986
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007083244756011465,
+      "loss": 3.8835,
+      "step": 2987
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007081359979095666,
+      "loss": 3.8011,
+      "step": 2988
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007079474844369957,
+      "loss": 4.0022,
+      "step": 2989
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007077589352158415,
+      "loss": 4.0906,
+      "step": 2990
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007075703502785177,
+      "loss": 3.8325,
+      "step": 2991
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007073817296574444,
+      "loss": 3.9195,
+      "step": 2992
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007071930733850476,
+      "loss": 4.0081,
+      "step": 2993
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007070043814937594,
+      "loss": 3.8181,
+      "step": 2994
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007068156540160182,
+      "loss": 3.8707,
+      "step": 2995
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007066268909842687,
+      "loss": 3.9449,
+      "step": 2996
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.000706438092430961,
+      "loss": 4.0292,
+      "step": 2997
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.000706249258388552,
+      "loss": 3.906,
+      "step": 2998
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007060603888895045,
+      "loss": 3.9801,
+      "step": 2999
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007058714839662874,
+      "loss": 3.8779,
+      "step": 3000
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007056825436513753,
+      "loss": 3.8605,
+      "step": 3001
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007054935679772496,
+      "loss": 3.845,
+      "step": 3002
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007053045569763973,
+      "loss": 3.9991,
+      "step": 3003
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007051155106813114,
+      "loss": 3.9653,
+      "step": 3004
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007049264291244914,
+      "loss": 3.8827,
+      "step": 3005
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007047373123384426,
+      "loss": 4.0024,
+      "step": 3006
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007045481603556763,
+      "loss": 3.9305,
+      "step": 3007
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0007043589732087098,
+      "loss": 3.9698,
+      "step": 3008
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0007041697509300667,
+      "loss": 4.0041,
+      "step": 3009
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0007039804935522765,
+      "loss": 4.0581,
+      "step": 3010
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0007037912011078749,
+      "loss": 3.9547,
+      "step": 3011
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.000703601873629403,
+      "loss": 3.9822,
+      "step": 3012
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0007034125111494087,
+      "loss": 3.9808,
+      "step": 3013
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0007032231137004457,
+      "loss": 3.8924,
+      "step": 3014
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0007030336813150733,
+      "loss": 3.8835,
+      "step": 3015
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0007028442140258575,
+      "loss": 3.9797,
+      "step": 3016
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0007026547118653696,
+      "loss": 3.8529,
+      "step": 3017
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0007024651748661875,
+      "loss": 3.9157,
+      "step": 3018
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0007022756030608946,
+      "loss": 4.0004,
+      "step": 3019
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0007020859964820805,
+      "loss": 3.8643,
+      "step": 3020
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.000701896355162341,
+      "loss": 3.9325,
+      "step": 3021
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0007017066791342772,
+      "loss": 3.9331,
+      "step": 3022
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0007015169684304969,
+      "loss": 3.8927,
+      "step": 3023
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0007013272230836138,
+      "loss": 3.9844,
+      "step": 3024
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.000701137443126247,
+      "loss": 4.0698,
+      "step": 3025
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0007009476285910218,
+      "loss": 3.9107,
+      "step": 3026
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0007007577795105696,
+      "loss": 3.9791,
+      "step": 3027
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0007005678959175279,
+      "loss": 4.032,
+      "step": 3028
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0007003779778445397,
+      "loss": 3.9358,
+      "step": 3029
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0007001880253242541,
+      "loss": 3.8833,
+      "step": 3030
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0006999980383893261,
+      "loss": 3.8952,
+      "step": 3031
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0006998080170724167,
+      "loss": 4.0159,
+      "step": 3032
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0006996179614061929,
+      "loss": 3.9064,
+      "step": 3033
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0006994278714233273,
+      "loss": 3.852,
+      "step": 3034
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0006992377471564987,
+      "loss": 3.9862,
+      "step": 3035
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0006990475886383915,
+      "loss": 3.8768,
+      "step": 3036
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0006988573959016963,
+      "loss": 3.9425,
+      "step": 3037
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0006986671689791094,
+      "loss": 3.8077,
+      "step": 3038
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0006984769079033331,
+      "loss": 4.0236,
+      "step": 3039
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0006982866127070753,
+      "loss": 3.9022,
+      "step": 3040
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0006980962834230499,
+      "loss": 4.018,
+      "step": 3041
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0006979059200839771,
+      "loss": 3.9192,
+      "step": 3042
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0006977155227225822,
+      "loss": 3.8955,
+      "step": 3043
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0006975250913715968,
+      "loss": 3.8758,
+      "step": 3044
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0006973346260637583,
+      "loss": 3.9734,
+      "step": 3045
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0006971441268318099,
+      "loss": 3.8476,
+      "step": 3046
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0006969535937085005,
+      "loss": 3.9612,
+      "step": 3047
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0006967630267265853,
+      "loss": 4.0517,
+      "step": 3048
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0006965724259188246,
+      "loss": 3.9644,
+      "step": 3049
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0006963817913179849,
+      "loss": 3.8619,
+      "step": 3050
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0006961911229568387,
+      "loss": 3.9431,
+      "step": 3051
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0006960004208681642,
+      "loss": 4.009,
+      "step": 3052
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.000695809685084745,
+      "loss": 3.9361,
+      "step": 3053
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0006956189156393708,
+      "loss": 3.8585,
+      "step": 3054
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0006954281125648373,
+      "loss": 3.9502,
+      "step": 3055
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0006952372758939457,
+      "loss": 3.9545,
+      "step": 3056
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0006950464056595029,
+      "loss": 3.8505,
+      "step": 3057
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0006948555018943219,
+      "loss": 3.9866,
+      "step": 3058
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0006946645646312211,
+      "loss": 3.8518,
+      "step": 3059
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0006944735939030249,
+      "loss": 4.0232,
+      "step": 3060
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0006942825897425633,
+      "loss": 3.9493,
+      "step": 3061
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0006940915521826723,
+      "loss": 3.9649,
+      "step": 3062
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0006939004812561933,
+      "loss": 3.9406,
+      "step": 3063
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0006937093769959736,
+      "loss": 4.0292,
+      "step": 3064
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0006935182394348664,
+      "loss": 3.8984,
+      "step": 3065
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0006933270686057307,
+      "loss": 3.8967,
+      "step": 3066
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0006931358645414304,
+      "loss": 3.9774,
+      "step": 3067
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0006929446272748359,
+      "loss": 3.806,
+      "step": 3068
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0006927533568388231,
+      "loss": 4.0142,
+      "step": 3069
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.000692562053266274,
+      "loss": 3.9867,
+      "step": 3070
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0006923707165900754,
+      "loss": 3.9472,
+      "step": 3071
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0006921793468431203,
+      "loss": 3.8831,
+      "step": 3072
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0006919879440583077,
+      "loss": 3.7709,
+      "step": 3073
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0006917965082685418,
+      "loss": 3.9524,
+      "step": 3074
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0006916050395067327,
+      "loss": 3.9563,
+      "step": 3075
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0006914135378057959,
+      "loss": 4.0244,
+      "step": 3076
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0006912220031986531,
+      "loss": 3.9709,
+      "step": 3077
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0006910304357182307,
+      "loss": 3.9777,
+      "step": 3078
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0006908388353974622,
+      "loss": 3.8777,
+      "step": 3079
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0006906472022692854,
+      "loss": 3.8536,
+      "step": 3080
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0006904555363666442,
+      "loss": 3.9172,
+      "step": 3081
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0006902638377224886,
+      "loss": 4.0229,
+      "step": 3082
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0006900721063697732,
+      "loss": 3.9817,
+      "step": 3083
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0006898803423414595,
+      "loss": 3.8891,
+      "step": 3084
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0006896885456705137,
+      "loss": 3.8953,
+      "step": 3085
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006894967163899076,
+      "loss": 4.0383,
+      "step": 3086
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006893048545326192,
+      "loss": 4.0575,
+      "step": 3087
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006891129601316318,
+      "loss": 4.0069,
+      "step": 3088
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006889210332199343,
+      "loss": 4.1062,
+      "step": 3089
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006887290738305207,
+      "loss": 3.9602,
+      "step": 3090
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006885370819963917,
+      "loss": 3.8287,
+      "step": 3091
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006883450577505523,
+      "loss": 3.9055,
+      "step": 3092
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006881530011260142,
+      "loss": 4.0137,
+      "step": 3093
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006879609121557937,
+      "loss": 3.8841,
+      "step": 3094
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006877687908729137,
+      "loss": 4.0371,
+      "step": 3095
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006875766373104016,
+      "loss": 3.954,
+      "step": 3096
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006873844515012908,
+      "loss": 3.8101,
+      "step": 3097
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006871922334786205,
+      "loss": 3.8314,
+      "step": 3098
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006869999832754355,
+      "loss": 4.0715,
+      "step": 3099
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006868077009247852,
+      "loss": 4.0037,
+      "step": 3100
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006866153864597253,
+      "loss": 3.9794,
+      "step": 3101
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006864230399133173,
+      "loss": 3.985,
+      "step": 3102
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006862306613186275,
+      "loss": 4.0688,
+      "step": 3103
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006860382507087279,
+      "loss": 3.9172,
+      "step": 3104
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006858458081166964,
+      "loss": 3.9657,
+      "step": 3105
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006856533335756158,
+      "loss": 3.8884,
+      "step": 3106
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006854608271185752,
+      "loss": 3.9679,
+      "step": 3107
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006852682887786681,
+      "loss": 4.0507,
+      "step": 3108
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006850757185889945,
+      "loss": 4.0684,
+      "step": 3109
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006848831165826591,
+      "loss": 3.9947,
+      "step": 3110
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006846904827927728,
+      "loss": 3.94,
+      "step": 3111
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006844978172524513,
+      "loss": 3.9983,
+      "step": 3112
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006843051199948161,
+      "loss": 3.9461,
+      "step": 3113
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006841123910529943,
+      "loss": 4.0106,
+      "step": 3114
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006839196304601179,
+      "loss": 3.9378,
+      "step": 3115
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006837268382493248,
+      "loss": 3.9659,
+      "step": 3116
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006835340144537584,
+      "loss": 3.9847,
+      "step": 3117
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006833411591065673,
+      "loss": 4.0472,
+      "step": 3118
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006831482722409052,
+      "loss": 3.9729,
+      "step": 3119
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006829553538899321,
+      "loss": 3.8567,
+      "step": 3120
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006827624040868125,
+      "loss": 4.0094,
+      "step": 3121
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006825694228647169,
+      "loss": 4.0958,
+      "step": 3122
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006823764102568211,
+      "loss": 3.9721,
+      "step": 3123
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.000682183366296306,
+      "loss": 3.9311,
+      "step": 3124
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006819902910163581,
+      "loss": 3.8925,
+      "step": 3125
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006817971844501695,
+      "loss": 4.055,
+      "step": 3126
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006816040466309372,
+      "loss": 4.0012,
+      "step": 3127
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006814108775918641,
+      "loss": 3.8523,
+      "step": 3128
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006812176773661579,
+      "loss": 3.8263,
+      "step": 3129
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006810244459870322,
+      "loss": 3.9175,
+      "step": 3130
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006808311834877056,
+      "loss": 3.8999,
+      "step": 3131
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006806378899014023,
+      "loss": 3.9269,
+      "step": 3132
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006804445652613514,
+      "loss": 3.9168,
+      "step": 3133
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006802512096007878,
+      "loss": 3.9165,
+      "step": 3134
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006800578229529518,
+      "loss": 3.9776,
+      "step": 3135
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006798644053510886,
+      "loss": 4.0186,
+      "step": 3136
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006796709568284488,
+      "loss": 3.9644,
+      "step": 3137
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006794774774182887,
+      "loss": 4.009,
+      "step": 3138
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006792839671538695,
+      "loss": 4.0108,
+      "step": 3139
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006790904260684581,
+      "loss": 3.9259,
+      "step": 3140
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006788968541953261,
+      "loss": 3.9487,
+      "step": 3141
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006787032515677509,
+      "loss": 3.9863,
+      "step": 3142
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006785096182190151,
+      "loss": 3.9926,
+      "step": 3143
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006783159541824065,
+      "loss": 4.0834,
+      "step": 3144
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006781222594912181,
+      "loss": 3.9334,
+      "step": 3145
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006779285341787484,
+      "loss": 3.9849,
+      "step": 3146
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006777347782783008,
+      "loss": 3.7085,
+      "step": 3147
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006775409918231843,
+      "loss": 3.988,
+      "step": 3148
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006773471748467133,
+      "loss": 3.8623,
+      "step": 3149
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.000677153327382207,
+      "loss": 3.8078,
+      "step": 3150
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006769594494629897,
+      "loss": 3.9534,
+      "step": 3151
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006767655411223916,
+      "loss": 3.8626,
+      "step": 3152
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006765716023937479,
+      "loss": 3.9841,
+      "step": 3153
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006763776333103988,
+      "loss": 3.877,
+      "step": 3154
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006761836339056896,
+      "loss": 3.9383,
+      "step": 3155
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006759896042129713,
+      "loss": 3.8934,
+      "step": 3156
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006757955442655998,
+      "loss": 3.8541,
+      "step": 3157
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006756014540969362,
+      "loss": 4.0093,
+      "step": 3158
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006754073337403469,
+      "loss": 4.0088,
+      "step": 3159
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006752131832292035,
+      "loss": 3.9448,
+      "step": 3160
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006750190025968827,
+      "loss": 4.0304,
+      "step": 3161
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006748247918767662,
+      "loss": 3.9646,
+      "step": 3162
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006746305511022414,
+      "loss": 3.9241,
+      "step": 3163
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006744362803067002,
+      "loss": 3.8869,
+      "step": 3164
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006742419795235403,
+      "loss": 3.9317,
+      "step": 3165
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.000674047648786164,
+      "loss": 3.86,
+      "step": 3166
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006738532881279793,
+      "loss": 3.9524,
+      "step": 3167
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.000673658897582399,
+      "loss": 3.914,
+      "step": 3168
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006734644771828407,
+      "loss": 3.9952,
+      "step": 3169
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006732700269627279,
+      "loss": 3.9562,
+      "step": 3170
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006730755469554888,
+      "loss": 3.9234,
+      "step": 3171
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006728810371945566,
+      "loss": 3.9585,
+      "step": 3172
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00067268649771337,
+      "loss": 3.9406,
+      "step": 3173
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006724919285453725,
+      "loss": 4.0072,
+      "step": 3174
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006722973297240127,
+      "loss": 3.9919,
+      "step": 3175
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006721027012827444,
+      "loss": 3.8968,
+      "step": 3176
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006719080432550269,
+      "loss": 3.9127,
+      "step": 3177
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006717133556743235,
+      "loss": 3.9208,
+      "step": 3178
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006715186385741037,
+      "loss": 3.9606,
+      "step": 3179
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006713238919878415,
+      "loss": 3.9851,
+      "step": 3180
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006711291159490161,
+      "loss": 3.9745,
+      "step": 3181
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006709343104911119,
+      "loss": 3.8687,
+      "step": 3182
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006707394756476179,
+      "loss": 3.8784,
+      "step": 3183
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006705446114520288,
+      "loss": 3.9608,
+      "step": 3184
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006703497179378438,
+      "loss": 3.9158,
+      "step": 3185
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006701547951385675,
+      "loss": 3.8899,
+      "step": 3186
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006699598430877092,
+      "loss": 3.9578,
+      "step": 3187
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006697648618187836,
+      "loss": 3.935,
+      "step": 3188
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006695698513653101,
+      "loss": 3.8995,
+      "step": 3189
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006693748117608133,
+      "loss": 3.8054,
+      "step": 3190
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.000669179743038823,
+      "loss": 3.8713,
+      "step": 3191
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006689846452328735,
+      "loss": 3.8843,
+      "step": 3192
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006687895183765043,
+      "loss": 3.8407,
+      "step": 3193
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006685943625032603,
+      "loss": 4.0122,
+      "step": 3194
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006683991776466906,
+      "loss": 3.9589,
+      "step": 3195
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006682039638403503,
+      "loss": 3.8447,
+      "step": 3196
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006680087211177985,
+      "loss": 3.8361,
+      "step": 3197
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006678134495125998,
+      "loss": 3.9118,
+      "step": 3198
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006676181490583238,
+      "loss": 3.8091,
+      "step": 3199
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006674228197885446,
+      "loss": 4.0066,
+      "step": 3200
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006672274617368418,
+      "loss": 3.83,
+      "step": 3201
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006670320749367997,
+      "loss": 3.8777,
+      "step": 3202
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006668366594220075,
+      "loss": 3.929,
+      "step": 3203
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006666412152260595,
+      "loss": 4.0516,
+      "step": 3204
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006664457423825546,
+      "loss": 3.9838,
+      "step": 3205
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.000666250240925097,
+      "loss": 3.9162,
+      "step": 3206
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006660547108872959,
+      "loss": 3.9095,
+      "step": 3207
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006658591523027643,
+      "loss": 3.932,
+      "step": 3208
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006656635652051222,
+      "loss": 3.8472,
+      "step": 3209
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006654679496279926,
+      "loss": 3.9356,
+      "step": 3210
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.000665272305605004,
+      "loss": 3.9219,
+      "step": 3211
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006650766331697902,
+      "loss": 3.8946,
+      "step": 3212
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006648809323559895,
+      "loss": 3.8731,
+      "step": 3213
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006646852031972449,
+      "loss": 3.8174,
+      "step": 3214
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006644894457272047,
+      "loss": 3.8704,
+      "step": 3215
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006642936599795219,
+      "loss": 3.9779,
+      "step": 3216
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006640978459878543,
+      "loss": 3.8606,
+      "step": 3217
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006639020037858646,
+      "loss": 3.8202,
+      "step": 3218
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006637061334072203,
+      "loss": 3.9171,
+      "step": 3219
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006635102348855939,
+      "loss": 3.9309,
+      "step": 3220
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006633143082546624,
+      "loss": 3.8392,
+      "step": 3221
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006631183535481081,
+      "loss": 3.9481,
+      "step": 3222
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006629223707996177,
+      "loss": 3.9318,
+      "step": 3223
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006627263600428832,
+      "loss": 3.9795,
+      "step": 3224
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006625303213116008,
+      "loss": 3.9136,
+      "step": 3225
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006623342546394719,
+      "loss": 3.8398,
+      "step": 3226
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006621381600602028,
+      "loss": 4.09,
+      "step": 3227
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006619420376075042,
+      "loss": 3.8181,
+      "step": 3228
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006617458873150921,
+      "loss": 4.1098,
+      "step": 3229
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006615497092166868,
+      "loss": 4.0933,
+      "step": 3230
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006613535033460135,
+      "loss": 3.8643,
+      "step": 3231
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006611572697368026,
+      "loss": 4.0028,
+      "step": 3232
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006609610084227886,
+      "loss": 3.8544,
+      "step": 3233
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006607647194377113,
+      "loss": 3.9487,
+      "step": 3234
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006605684028153149,
+      "loss": 3.9318,
+      "step": 3235
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006603720585893486,
+      "loss": 4.0548,
+      "step": 3236
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006601756867935664,
+      "loss": 3.9309,
+      "step": 3237
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006599792874617266,
+      "loss": 3.8638,
+      "step": 3238
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006597828606275928,
+      "loss": 3.9187,
+      "step": 3239
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006595864063249326,
+      "loss": 3.9941,
+      "step": 3240
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006593899245875192,
+      "loss": 3.778,
+      "step": 3241
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00065919341544913,
+      "loss": 3.9597,
+      "step": 3242
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006589968789435472,
+      "loss": 3.8852,
+      "step": 3243
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006588003151045576,
+      "loss": 3.9332,
+      "step": 3244
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006586037239659528,
+      "loss": 3.9085,
+      "step": 3245
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006584071055615293,
+      "loss": 3.9093,
+      "step": 3246
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006582104599250878,
+      "loss": 3.9889,
+      "step": 3247
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006580137870904341,
+      "loss": 3.9957,
+      "step": 3248
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006578170870913787,
+      "loss": 4.049,
+      "step": 3249
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006576203599617363,
+      "loss": 3.925,
+      "step": 3250
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006574236057353268,
+      "loss": 3.9024,
+      "step": 3251
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006572268244459745,
+      "loss": 3.8465,
+      "step": 3252
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006570300161275084,
+      "loss": 4.0236,
+      "step": 3253
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006568331808137619,
+      "loss": 3.8662,
+      "step": 3254
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006566363185385735,
+      "loss": 3.9889,
+      "step": 3255
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006564394293357861,
+      "loss": 3.8585,
+      "step": 3256
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006562425132392472,
+      "loss": 4.0484,
+      "step": 3257
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006560455702828089,
+      "loss": 4.0181,
+      "step": 3258
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006558486005003282,
+      "loss": 3.9665,
+      "step": 3259
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006556516039256661,
+      "loss": 3.9914,
+      "step": 3260
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.000655454580592689,
+      "loss": 4.0511,
+      "step": 3261
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006552575305352671,
+      "loss": 3.9683,
+      "step": 3262
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006550604537872759,
+      "loss": 3.8411,
+      "step": 3263
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006548633503825952,
+      "loss": 3.8547,
+      "step": 3264
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006546662203551092,
+      "loss": 3.9988,
+      "step": 3265
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006544690637387068,
+      "loss": 3.9696,
+      "step": 3266
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006542718805672816,
+      "loss": 3.8459,
+      "step": 3267
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006540746708747317,
+      "loss": 3.8873,
+      "step": 3268
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006538774346949597,
+      "loss": 3.8987,
+      "step": 3269
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006536801720618727,
+      "loss": 3.8783,
+      "step": 3270
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006534828830093827,
+      "loss": 3.8855,
+      "step": 3271
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006532855675714054,
+      "loss": 3.7634,
+      "step": 3272
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006530882257818622,
+      "loss": 3.9587,
+      "step": 3273
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006528908576746784,
+      "loss": 4.108,
+      "step": 3274
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006526934632837835,
+      "loss": 4.0171,
+      "step": 3275
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006524960426431122,
+      "loss": 3.8514,
+      "step": 3276
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006522985957866032,
+      "loss": 3.9451,
+      "step": 3277
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006521011227482,
+      "loss": 4.0662,
+      "step": 3278
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006519036235618505,
+      "loss": 3.9263,
+      "step": 3279
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006517060982615071,
+      "loss": 4.0454,
+      "step": 3280
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006515085468811268,
+      "loss": 3.9994,
+      "step": 3281
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006513109694546706,
+      "loss": 3.9285,
+      "step": 3282
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006511133660161047,
+      "loss": 3.8945,
+      "step": 3283
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006509157365993994,
+      "loss": 3.7812,
+      "step": 3284
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006507180812385295,
+      "loss": 4.0251,
+      "step": 3285
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.000650520399967474,
+      "loss": 3.8198,
+      "step": 3286
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006503226928202167,
+      "loss": 3.8973,
+      "step": 3287
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.000650124959830746,
+      "loss": 3.9739,
+      "step": 3288
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006499272010330543,
+      "loss": 3.9671,
+      "step": 3289
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006497294164611385,
+      "loss": 3.8776,
+      "step": 3290
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006495316061490002,
+      "loss": 3.92,
+      "step": 3291
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006493337701306454,
+      "loss": 3.9239,
+      "step": 3292
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006491359084400843,
+      "loss": 4.044,
+      "step": 3293
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006489380211113315,
+      "loss": 4.0512,
+      "step": 3294
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006487401081784062,
+      "loss": 3.9135,
+      "step": 3295
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.000648542169675332,
+      "loss": 3.9777,
+      "step": 3296
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006483442056361368,
+      "loss": 3.9724,
+      "step": 3297
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.000648146216094853,
+      "loss": 3.7974,
+      "step": 3298
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006479482010855172,
+      "loss": 3.9543,
+      "step": 3299
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006477501606421702,
+      "loss": 3.9769,
+      "step": 3300
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.000647552094798858,
+      "loss": 3.9932,
+      "step": 3301
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00064735400358963,
+      "loss": 4.0031,
+      "step": 3302
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006471558870485407,
+      "loss": 3.7783,
+      "step": 3303
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006469577452096482,
+      "loss": 3.8795,
+      "step": 3304
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006467595781070157,
+      "loss": 3.9137,
+      "step": 3305
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006465613857747104,
+      "loss": 3.801,
+      "step": 3306
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006463631682468037,
+      "loss": 3.9418,
+      "step": 3307
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006461649255573716,
+      "loss": 4.0682,
+      "step": 3308
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006459666577404943,
+      "loss": 3.9524,
+      "step": 3309
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006457683648302565,
+      "loss": 3.8532,
+      "step": 3310
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006455700468607469,
+      "loss": 3.9021,
+      "step": 3311
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006453717038660584,
+      "loss": 4.0016,
+      "step": 3312
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006451733358802888,
+      "loss": 3.8222,
+      "step": 3313
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006449749429375398,
+      "loss": 3.945,
+      "step": 3314
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006447765250719174,
+      "loss": 3.8546,
+      "step": 3315
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006445780823175321,
+      "loss": 3.9348,
+      "step": 3316
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006443796147084981,
+      "loss": 4.0914,
+      "step": 3317
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006441811222789347,
+      "loss": 4.0214,
+      "step": 3318
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006439826050629648,
+      "loss": 3.8324,
+      "step": 3319
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006437840630947161,
+      "loss": 4.0038,
+      "step": 3320
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006435854964083198,
+      "loss": 3.8939,
+      "step": 3321
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006433869050379121,
+      "loss": 3.9092,
+      "step": 3322
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006431882890176333,
+      "loss": 4.0129,
+      "step": 3323
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006429896483816276,
+      "loss": 4.0635,
+      "step": 3324
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006427909831640437,
+      "loss": 4.0381,
+      "step": 3325
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006425922933990343,
+      "loss": 4.0574,
+      "step": 3326
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006423935791207568,
+      "loss": 4.0496,
+      "step": 3327
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006421948403633721,
+      "loss": 3.8077,
+      "step": 3328
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.000641996077161046,
+      "loss": 3.961,
+      "step": 3329
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.000641797289547948,
+      "loss": 3.845,
+      "step": 3330
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006415984775582521,
+      "loss": 3.8687,
+      "step": 3331
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006413996412261363,
+      "loss": 4.0036,
+      "step": 3332
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006412007805857828,
+      "loss": 3.8034,
+      "step": 3333
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006410018956713783,
+      "loss": 3.7867,
+      "step": 3334
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006408029865171134,
+      "loss": 3.8536,
+      "step": 3335
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006406040531571828,
+      "loss": 3.9999,
+      "step": 3336
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006404050956257852,
+      "loss": 3.9975,
+      "step": 3337
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006402061139571243,
+      "loss": 3.9474,
+      "step": 3338
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006400071081854068,
+      "loss": 4.0321,
+      "step": 3339
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006398080783448442,
+      "loss": 3.9022,
+      "step": 3340
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006396090244696523,
+      "loss": 3.8236,
+      "step": 3341
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006394099465940505,
+      "loss": 3.8028,
+      "step": 3342
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006392108447522629,
+      "loss": 3.8801,
+      "step": 3343
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.000639011718978517,
+      "loss": 3.8435,
+      "step": 3344
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006388125693070451,
+      "loss": 3.9811,
+      "step": 3345
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006386133957720832,
+      "loss": 3.8889,
+      "step": 3346
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006384141984078719,
+      "loss": 3.811,
+      "step": 3347
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.000638214977248655,
+      "loss": 3.8602,
+      "step": 3348
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006380157323286812,
+      "loss": 4.0131,
+      "step": 3349
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006378164636822032,
+      "loss": 3.9442,
+      "step": 3350
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006376171713434771,
+      "loss": 3.9701,
+      "step": 3351
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006374178553467639,
+      "loss": 4.0133,
+      "step": 3352
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006372185157263286,
+      "loss": 3.9079,
+      "step": 3353
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006370191525164394,
+      "loss": 3.9659,
+      "step": 3354
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006368197657513695,
+      "loss": 3.8941,
+      "step": 3355
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006366203554653957,
+      "loss": 3.8972,
+      "step": 3356
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.000636420921692799,
+      "loss": 3.9394,
+      "step": 3357
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006362214644678641,
+      "loss": 3.9659,
+      "step": 3358
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006360219838248806,
+      "loss": 3.9441,
+      "step": 3359
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006358224797981409,
+      "loss": 3.8233,
+      "step": 3360
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006356229524219425,
+      "loss": 4.0306,
+      "step": 3361
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006354234017305862,
+      "loss": 3.95,
+      "step": 3362
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006352238277583773,
+      "loss": 3.8565,
+      "step": 3363
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006350242305396247,
+      "loss": 3.9287,
+      "step": 3364
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006348246101086413,
+      "loss": 3.8856,
+      "step": 3365
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006346249664997448,
+      "loss": 3.875,
+      "step": 3366
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006344252997472556,
+      "loss": 4.0013,
+      "step": 3367
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006342256098854992,
+      "loss": 3.8554,
+      "step": 3368
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006340258969488039,
+      "loss": 3.9394,
+      "step": 3369
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006338261609715037,
+      "loss": 3.9579,
+      "step": 3370
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006336264019879348,
+      "loss": 3.8728,
+      "step": 3371
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006334266200324381,
+      "loss": 3.9559,
+      "step": 3372
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006332268151393589,
+      "loss": 3.9944,
+      "step": 3373
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006330269873430454,
+      "loss": 3.929,
+      "step": 3374
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006328271366778509,
+      "loss": 3.9578,
+      "step": 3375
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006326272631781314,
+      "loss": 3.8514,
+      "step": 3376
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.000632427366878248,
+      "loss": 3.879,
+      "step": 3377
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.000632227447812565,
+      "loss": 3.8205,
+      "step": 3378
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006320275060154507,
+      "loss": 3.977,
+      "step": 3379
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006318275415212776,
+      "loss": 3.93,
+      "step": 3380
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006316275543644218,
+      "loss": 3.9457,
+      "step": 3381
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006314275445792637,
+      "loss": 3.9089,
+      "step": 3382
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006312275122001866,
+      "loss": 3.9981,
+      "step": 3383
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006310274572615791,
+      "loss": 3.8959,
+      "step": 3384
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006308273797978328,
+      "loss": 3.9123,
+      "step": 3385
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.000630627279843343,
+      "loss": 3.7528,
+      "step": 3386
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006304271574325095,
+      "loss": 3.786,
+      "step": 3387
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006302270125997356,
+      "loss": 3.9535,
+      "step": 3388
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006300268453794287,
+      "loss": 3.986,
+      "step": 3389
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006298266558059995,
+      "loss": 3.8389,
+      "step": 3390
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.000629626443913863,
+      "loss": 4.0424,
+      "step": 3391
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006294262097374382,
+      "loss": 3.8366,
+      "step": 3392
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006292259533111474,
+      "loss": 3.9426,
+      "step": 3393
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006290256746694172,
+      "loss": 3.9338,
+      "step": 3394
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006288253738466777,
+      "loss": 3.8177,
+      "step": 3395
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.000628625050877363,
+      "loss": 4.0048,
+      "step": 3396
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006284247057959107,
+      "loss": 4.0451,
+      "step": 3397
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006282243386367628,
+      "loss": 3.8896,
+      "step": 3398
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006280239494343647,
+      "loss": 3.8402,
+      "step": 3399
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006278235382231654,
+      "loss": 3.884,
+      "step": 3400
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.000627623105037618,
+      "loss": 3.8771,
+      "step": 3401
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006274226499121792,
+      "loss": 3.9514,
+      "step": 3402
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006272221728813099,
+      "loss": 3.8425,
+      "step": 3403
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006270216739794741,
+      "loss": 3.8205,
+      "step": 3404
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00062682115324114,
+      "loss": 3.8975,
+      "step": 3405
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006266206107007793,
+      "loss": 3.9205,
+      "step": 3406
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006264200463928679,
+      "loss": 3.8565,
+      "step": 3407
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006262194603518847,
+      "loss": 3.9894,
+      "step": 3408
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006260188526123133,
+      "loss": 3.9514,
+      "step": 3409
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00062581822320864,
+      "loss": 4.0071,
+      "step": 3410
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006256175721753556,
+      "loss": 3.7922,
+      "step": 3411
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006254168995469541,
+      "loss": 3.9107,
+      "step": 3412
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006252162053579337,
+      "loss": 3.9161,
+      "step": 3413
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006250154896427962,
+      "loss": 4.0238,
+      "step": 3414
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006248147524360464,
+      "loss": 4.0654,
+      "step": 3415
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006246139937721939,
+      "loss": 3.7897,
+      "step": 3416
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006244132136857512,
+      "loss": 3.8057,
+      "step": 3417
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006242124122112347,
+      "loss": 3.9351,
+      "step": 3418
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006240115893831644,
+      "loss": 3.9303,
+      "step": 3419
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006238107452360643,
+      "loss": 3.9974,
+      "step": 3420
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006236098798044618,
+      "loss": 3.8639,
+      "step": 3421
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.000623408993122888,
+      "loss": 3.8725,
+      "step": 3422
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006232080852258776,
+      "loss": 3.9294,
+      "step": 3423
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.000623007156147969,
+      "loss": 3.9685,
+      "step": 3424
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.000622806205923704,
+      "loss": 3.8078,
+      "step": 3425
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006226052345876288,
+      "loss": 3.9136,
+      "step": 3426
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006224042421742924,
+      "loss": 3.9304,
+      "step": 3427
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006222032287182477,
+      "loss": 3.9655,
+      "step": 3428
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006220021942540512,
+      "loss": 3.9442,
+      "step": 3429
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.000621801138816263,
+      "loss": 3.7237,
+      "step": 3430
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006216000624394472,
+      "loss": 3.882,
+      "step": 3431
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.000621398965158171,
+      "loss": 4.0017,
+      "step": 3432
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006211978470070052,
+      "loss": 3.9118,
+      "step": 3433
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006209967080205244,
+      "loss": 3.9963,
+      "step": 3434
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.000620795548233307,
+      "loss": 3.7942,
+      "step": 3435
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006205943676799344,
+      "loss": 3.9704,
+      "step": 3436
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006203931663949918,
+      "loss": 3.7477,
+      "step": 3437
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006201919444130683,
+      "loss": 3.8733,
+      "step": 3438
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006199907017687561,
+      "loss": 3.9319,
+      "step": 3439
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006197894384966512,
+      "loss": 3.8742,
+      "step": 3440
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006195881546313532,
+      "loss": 4.0379,
+      "step": 3441
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006193868502074651,
+      "loss": 3.964,
+      "step": 3442
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006191855252595933,
+      "loss": 3.9792,
+      "step": 3443
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006189841798223479,
+      "loss": 3.7933,
+      "step": 3444
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006187828139303427,
+      "loss": 3.7991,
+      "step": 3445
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006185814276181948,
+      "loss": 3.843,
+      "step": 3446
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006183800209205246,
+      "loss": 3.9207,
+      "step": 3447
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006181785938719565,
+      "loss": 3.886,
+      "step": 3448
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006179771465071181,
+      "loss": 4.0509,
+      "step": 3449
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006177756788606405,
+      "loss": 3.8216,
+      "step": 3450
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006175741909671584,
+      "loss": 3.9577,
+      "step": 3451
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006173726828613097,
+      "loss": 3.9211,
+      "step": 3452
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006171711545777363,
+      "loss": 3.8929,
+      "step": 3453
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.000616969606151083,
+      "loss": 3.8846,
+      "step": 3454
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006167680376159982,
+      "loss": 3.9828,
+      "step": 3455
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006165664490071342,
+      "loss": 4.046,
+      "step": 3456
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006163648403591462,
+      "loss": 3.8391,
+      "step": 3457
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006161632117066931,
+      "loss": 3.9758,
+      "step": 3458
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006159615630844374,
+      "loss": 3.9368,
+      "step": 3459
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006157598945270446,
+      "loss": 3.8647,
+      "step": 3460
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006155582060691837,
+      "loss": 3.8945,
+      "step": 3461
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006153564977455277,
+      "loss": 3.7823,
+      "step": 3462
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006151547695907525,
+      "loss": 4.0902,
+      "step": 3463
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006149530216395374,
+      "loss": 3.8651,
+      "step": 3464
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006147512539265651,
+      "loss": 4.0047,
+      "step": 3465
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.000614549466486522,
+      "loss": 3.8184,
+      "step": 3466
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006143476593540976,
+      "loss": 4.0856,
+      "step": 3467
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.000614145832563985,
+      "loss": 3.9146,
+      "step": 3468
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006139439861508803,
+      "loss": 3.8637,
+      "step": 3469
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006137421201494836,
+      "loss": 4.0032,
+      "step": 3470
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006135402345944978,
+      "loss": 3.9375,
+      "step": 3471
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006133383295206294,
+      "loss": 3.9222,
+      "step": 3472
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006131364049625882,
+      "loss": 3.8849,
+      "step": 3473
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006129344609550876,
+      "loss": 3.9427,
+      "step": 3474
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006127324975328436,
+      "loss": 4.1218,
+      "step": 3475
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006125305147305764,
+      "loss": 3.8249,
+      "step": 3476
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0006123285125830093,
+      "loss": 3.9439,
+      "step": 3477
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0006121264911248687,
+      "loss": 3.8469,
+      "step": 3478
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0006119244503908843,
+      "loss": 3.8912,
+      "step": 3479
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0006117223904157893,
+      "loss": 3.9756,
+      "step": 3480
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0006115203112343203,
+      "loss": 3.8022,
+      "step": 3481
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0006113182128812169,
+      "loss": 3.9904,
+      "step": 3482
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0006111160953912221,
+      "loss": 3.7779,
+      "step": 3483
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0006109139587990825,
+      "loss": 3.7388,
+      "step": 3484
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0006107118031395475,
+      "loss": 3.7997,
+      "step": 3485
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0006105096284473701,
+      "loss": 4.0073,
+      "step": 3486
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0006103074347573063,
+      "loss": 3.9264,
+      "step": 3487
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0006101052221041157,
+      "loss": 4.0455,
+      "step": 3488
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0006099029905225611,
+      "loss": 3.9325,
+      "step": 3489
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.000609700740047408,
+      "loss": 3.7907,
+      "step": 3490
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0006094984707134262,
+      "loss": 3.9211,
+      "step": 3491
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.000609296182555388,
+      "loss": 3.7421,
+      "step": 3492
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0006090938756080688,
+      "loss": 3.8872,
+      "step": 3493
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0006088915499062475,
+      "loss": 3.8069,
+      "step": 3494
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0006086892054847065,
+      "loss": 3.8366,
+      "step": 3495
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0006084868423782312,
+      "loss": 3.9349,
+      "step": 3496
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0006082844606216098,
+      "loss": 4.1325,
+      "step": 3497
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0006080820602496344,
+      "loss": 3.8952,
+      "step": 3498
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0006078796412970997,
+      "loss": 3.7804,
+      "step": 3499
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0006076772037988042,
+      "loss": 3.9555,
+      "step": 3500
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0006074747477895489,
+      "loss": 3.9921,
+      "step": 3501
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0006072722733041387,
+      "loss": 3.9385,
+      "step": 3502
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.000607069780377381,
+      "loss": 3.9301,
+      "step": 3503
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0006068672690440867,
+      "loss": 3.8748,
+      "step": 3504
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00060666473933907,
+      "loss": 3.8861,
+      "step": 3505
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0006064621912971481,
+      "loss": 3.9573,
+      "step": 3506
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0006062596249531414,
+      "loss": 3.9772,
+      "step": 3507
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0006060570403418731,
+      "loss": 4.0992,
+      "step": 3508
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00060585443749817,
+      "loss": 3.8976,
+      "step": 3509
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0006056518164568622,
+      "loss": 3.8646,
+      "step": 3510
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0006054491772527821,
+      "loss": 3.8708,
+      "step": 3511
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0006052465199207661,
+      "loss": 3.8281,
+      "step": 3512
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0006050438444956531,
+      "loss": 3.8171,
+      "step": 3513
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0006048411510122855,
+      "loss": 4.0206,
+      "step": 3514
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0006046384395055086,
+      "loss": 3.9069,
+      "step": 3515
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.000604435710010171,
+      "loss": 3.8859,
+      "step": 3516
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0006042329625611239,
+      "loss": 4.0243,
+      "step": 3517
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0006040301971932222,
+      "loss": 3.9743,
+      "step": 3518
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0006038274139413237,
+      "loss": 4.0669,
+      "step": 3519
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0006036246128402891,
+      "loss": 3.8842,
+      "step": 3520
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0006034217939249822,
+      "loss": 3.9955,
+      "step": 3521
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00060321895723027,
+      "loss": 3.8833,
+      "step": 3522
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0006030161027910223,
+      "loss": 3.8731,
+      "step": 3523
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0006028132306421124,
+      "loss": 3.9207,
+      "step": 3524
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0006026103408184162,
+      "loss": 3.888,
+      "step": 3525
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0006024074333548127,
+      "loss": 4.0201,
+      "step": 3526
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0006022045082861843,
+      "loss": 3.9978,
+      "step": 3527
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0006020015656474161,
+      "loss": 3.9381,
+      "step": 3528
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0006017986054733962,
+      "loss": 3.8223,
+      "step": 3529
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.000601595627799016,
+      "loss": 3.835,
+      "step": 3530
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0006013926326591695,
+      "loss": 3.9247,
+      "step": 3531
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0006011896200887539,
+      "loss": 3.959,
+      "step": 3532
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0006009865901226697,
+      "loss": 3.8777,
+      "step": 3533
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0006007835427958198,
+      "loss": 4.0626,
+      "step": 3534
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0006005804781431106,
+      "loss": 3.9179,
+      "step": 3535
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0006003773961994511,
+      "loss": 3.7778,
+      "step": 3536
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0006001742969997534,
+      "loss": 3.9451,
+      "step": 3537
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0005999711805789329,
+      "loss": 3.8928,
+      "step": 3538
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0005997680469719076,
+      "loss": 3.8558,
+      "step": 3539
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0005995648962135983,
+      "loss": 3.9486,
+      "step": 3540
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0005993617283389289,
+      "loss": 3.9708,
+      "step": 3541
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0005991585433828267,
+      "loss": 3.8389,
+      "step": 3542
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0005989553413802214,
+      "loss": 3.8391,
+      "step": 3543
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0005987521223660456,
+      "loss": 3.8178,
+      "step": 3544
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0005985488863752351,
+      "loss": 3.8927,
+      "step": 3545
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0005983456334427286,
+      "loss": 3.8335,
+      "step": 3546
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0005981423636034676,
+      "loss": 3.873,
+      "step": 3547
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0005979390768923967,
+      "loss": 3.96,
+      "step": 3548
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.000597735773344463,
+      "loss": 3.8656,
+      "step": 3549
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0005975324529946165,
+      "loss": 4.0149,
+      "step": 3550
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0005973291158778109,
+      "loss": 3.927,
+      "step": 3551
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.000597125762029002,
+      "loss": 3.9593,
+      "step": 3552
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0005969223914831486,
+      "loss": 3.8263,
+      "step": 3553
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0005967190042752123,
+      "loss": 3.8239,
+      "step": 3554
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.000596515600440158,
+      "loss": 3.9573,
+      "step": 3555
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005963121800129532,
+      "loss": 3.9558,
+      "step": 3556
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005961087430285681,
+      "loss": 3.9271,
+      "step": 3557
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005959052895219757,
+      "loss": 3.9034,
+      "step": 3558
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005957018195281523,
+      "loss": 3.7905,
+      "step": 3559
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005954983330820766,
+      "loss": 3.961,
+      "step": 3560
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005952948302187306,
+      "loss": 3.9077,
+      "step": 3561
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005950913109730982,
+      "loss": 3.8446,
+      "step": 3562
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005948877753801673,
+      "loss": 4.0095,
+      "step": 3563
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005946842234749274,
+      "loss": 3.8398,
+      "step": 3564
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005944806552923721,
+      "loss": 3.9769,
+      "step": 3565
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005942770708674969,
+      "loss": 4.0046,
+      "step": 3566
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005940734702353002,
+      "loss": 3.9013,
+      "step": 3567
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005938698534307833,
+      "loss": 3.8608,
+      "step": 3568
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005936662204889504,
+      "loss": 3.9517,
+      "step": 3569
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005934625714448084,
+      "loss": 3.8386,
+      "step": 3570
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005932589063333668,
+      "loss": 3.9145,
+      "step": 3571
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.000593055225189638,
+      "loss": 4.0058,
+      "step": 3572
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005928515280486372,
+      "loss": 3.958,
+      "step": 3573
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005926478149453823,
+      "loss": 3.9461,
+      "step": 3574
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.000592444085914894,
+      "loss": 3.9033,
+      "step": 3575
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005922403409921956,
+      "loss": 3.9612,
+      "step": 3576
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005920365802123133,
+      "loss": 3.9739,
+      "step": 3577
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005918328036102758,
+      "loss": 3.8738,
+      "step": 3578
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.000591629011221115,
+      "loss": 3.9034,
+      "step": 3579
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005914252030798645,
+      "loss": 3.772,
+      "step": 3580
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.000591221379221562,
+      "loss": 3.843,
+      "step": 3581
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005910175396812467,
+      "loss": 3.8313,
+      "step": 3582
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005908136844939611,
+      "loss": 3.8348,
+      "step": 3583
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005906098136947507,
+      "loss": 3.8717,
+      "step": 3584
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005904059273186626,
+      "loss": 3.8847,
+      "step": 3585
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005902020254007475,
+      "loss": 3.9271,
+      "step": 3586
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005899981079760585,
+      "loss": 3.9272,
+      "step": 3587
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005897941750796516,
+      "loss": 3.9035,
+      "step": 3588
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.000589590226746585,
+      "loss": 3.9087,
+      "step": 3589
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005893862630119197,
+      "loss": 4.0524,
+      "step": 3590
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005891822839107195,
+      "loss": 3.9766,
+      "step": 3591
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005889782894780509,
+      "loss": 3.9299,
+      "step": 3592
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005887742797489828,
+      "loss": 3.9504,
+      "step": 3593
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005885702547585867,
+      "loss": 3.926,
+      "step": 3594
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005883662145419373,
+      "loss": 3.9085,
+      "step": 3595
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005881621591341109,
+      "loss": 3.7023,
+      "step": 3596
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005879580885701874,
+      "loss": 3.8876,
+      "step": 3597
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005877540028852488,
+      "loss": 3.8563,
+      "step": 3598
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005875499021143799,
+      "loss": 4.0416,
+      "step": 3599
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005873457862926677,
+      "loss": 3.9299,
+      "step": 3600
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005871416554552021,
+      "loss": 3.9029,
+      "step": 3601
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.000586937509637076,
+      "loss": 3.8004,
+      "step": 3602
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005867333488733839,
+      "loss": 3.9028,
+      "step": 3603
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005865291731992238,
+      "loss": 3.8283,
+      "step": 3604
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005863249826496955,
+      "loss": 4.0576,
+      "step": 3605
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.000586120777259902,
+      "loss": 3.8592,
+      "step": 3606
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005859165570649486,
+      "loss": 3.8976,
+      "step": 3607
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005857123220999428,
+      "loss": 3.894,
+      "step": 3608
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005855080723999954,
+      "loss": 3.9832,
+      "step": 3609
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005853038080002188,
+      "loss": 3.8766,
+      "step": 3610
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005850995289357289,
+      "loss": 3.8916,
+      "step": 3611
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005848952352416433,
+      "loss": 3.8414,
+      "step": 3612
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005846909269530827,
+      "loss": 3.8691,
+      "step": 3613
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005844866041051698,
+      "loss": 3.9495,
+      "step": 3614
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005842822667330302,
+      "loss": 4.0384,
+      "step": 3615
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005840779148717919,
+      "loss": 3.8913,
+      "step": 3616
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005838735485565855,
+      "loss": 3.911,
+      "step": 3617
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005836691678225434,
+      "loss": 3.7868,
+      "step": 3618
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005834647727048016,
+      "loss": 3.8957,
+      "step": 3619
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005832603632384977,
+      "loss": 3.9316,
+      "step": 3620
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005830559394587724,
+      "loss": 3.8111,
+      "step": 3621
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005828515014007678,
+      "loss": 4.0528,
+      "step": 3622
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005826470490996298,
+      "loss": 3.8948,
+      "step": 3623
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.000582442582590506,
+      "loss": 3.9049,
+      "step": 3624
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005822381019085465,
+      "loss": 3.8362,
+      "step": 3625
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005820336070889037,
+      "loss": 3.8874,
+      "step": 3626
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005818290981667332,
+      "loss": 3.8238,
+      "step": 3627
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005816245751771917,
+      "loss": 3.8396,
+      "step": 3628
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005814200381554397,
+      "loss": 3.9694,
+      "step": 3629
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005812154871366389,
+      "loss": 3.8059,
+      "step": 3630
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005810109221559548,
+      "loss": 3.9389,
+      "step": 3631
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005808063432485537,
+      "loss": 3.9122,
+      "step": 3632
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005806017504496054,
+      "loss": 3.901,
+      "step": 3633
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005803971437942819,
+      "loss": 3.9066,
+      "step": 3634
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005801925233177574,
+      "loss": 3.8445,
+      "step": 3635
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005799878890552083,
+      "loss": 3.8451,
+      "step": 3636
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.000579783241041814,
+      "loss": 3.8142,
+      "step": 3637
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005795785793127554,
+      "loss": 3.9125,
+      "step": 3638
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005793739039032165,
+      "loss": 4.0593,
+      "step": 3639
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005791692148483834,
+      "loss": 3.9366,
+      "step": 3640
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005789645121834445,
+      "loss": 3.9092,
+      "step": 3641
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005787597959435904,
+      "loss": 3.9507,
+      "step": 3642
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005785550661640144,
+      "loss": 3.9073,
+      "step": 3643
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005783503228799119,
+      "loss": 3.9741,
+      "step": 3644
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005781455661264805,
+      "loss": 3.9287,
+      "step": 3645
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005779407959389204,
+      "loss": 3.8961,
+      "step": 3646
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005777360123524338,
+      "loss": 3.8488,
+      "step": 3647
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005775312154022255,
+      "loss": 3.9736,
+      "step": 3648
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005773264051235025,
+      "loss": 3.9063,
+      "step": 3649
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005771215815514741,
+      "loss": 3.863,
+      "step": 3650
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005769167447213517,
+      "loss": 3.8444,
+      "step": 3651
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005767118946683491,
+      "loss": 3.9887,
+      "step": 3652
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005765070314276826,
+      "loss": 3.8125,
+      "step": 3653
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005763021550345703,
+      "loss": 4.0085,
+      "step": 3654
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005760972655242332,
+      "loss": 3.9998,
+      "step": 3655
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005758923629318937,
+      "loss": 3.8843,
+      "step": 3656
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005756874472927774,
+      "loss": 3.879,
+      "step": 3657
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005754825186421113,
+      "loss": 3.9743,
+      "step": 3658
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005752775770151254,
+      "loss": 3.9267,
+      "step": 3659
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005750726224470513,
+      "loss": 4.0108,
+      "step": 3660
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005748676549731229,
+      "loss": 3.9401,
+      "step": 3661
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005746626746285767,
+      "loss": 3.9277,
+      "step": 3662
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005744576814486512,
+      "loss": 4.0102,
+      "step": 3663
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005742526754685871,
+      "loss": 3.9143,
+      "step": 3664
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005740476567236272,
+      "loss": 3.9748,
+      "step": 3665
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005738426252490168,
+      "loss": 3.9477,
+      "step": 3666
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.000573637581080003,
+      "loss": 3.8741,
+      "step": 3667
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005734325242518354,
+      "loss": 3.7242,
+      "step": 3668
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005732274547997655,
+      "loss": 3.9114,
+      "step": 3669
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005730223727590471,
+      "loss": 3.9185,
+      "step": 3670
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005728172781649365,
+      "loss": 4.0041,
+      "step": 3671
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005726121710526913,
+      "loss": 3.9608,
+      "step": 3672
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005724070514575724,
+      "loss": 3.8428,
+      "step": 3673
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005722019194148419,
+      "loss": 3.8462,
+      "step": 3674
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005719967749597643,
+      "loss": 3.9017,
+      "step": 3675
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005717916181276065,
+      "loss": 3.8792,
+      "step": 3676
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005715864489536372,
+      "loss": 4.0651,
+      "step": 3677
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005713812674731276,
+      "loss": 3.9173,
+      "step": 3678
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005711760737213504,
+      "loss": 3.8881,
+      "step": 3679
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.000570970867733581,
+      "loss": 4.0202,
+      "step": 3680
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005707656495450968,
+      "loss": 3.8118,
+      "step": 3681
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.000570560419191177,
+      "loss": 4.0582,
+      "step": 3682
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.000570355176707103,
+      "loss": 3.9965,
+      "step": 3683
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005701499221281586,
+      "loss": 3.9146,
+      "step": 3684
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005699446554896293,
+      "loss": 4.036,
+      "step": 3685
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005697393768268031,
+      "loss": 3.9713,
+      "step": 3686
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005695340861749693,
+      "loss": 3.8151,
+      "step": 3687
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00056932878356942,
+      "loss": 3.7956,
+      "step": 3688
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005691234690454491,
+      "loss": 4.0088,
+      "step": 3689
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005689181426383524,
+      "loss": 3.8901,
+      "step": 3690
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005687128043834282,
+      "loss": 3.8776,
+      "step": 3691
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005685074543159763,
+      "loss": 3.9917,
+      "step": 3692
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005683020924712988,
+      "loss": 3.9097,
+      "step": 3693
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005680967188846996,
+      "loss": 4.0017,
+      "step": 3694
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005678913335914851,
+      "loss": 4.0358,
+      "step": 3695
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005676859366269634,
+      "loss": 3.8433,
+      "step": 3696
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005674805280264445,
+      "loss": 3.9548,
+      "step": 3697
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005672751078252403,
+      "loss": 3.842,
+      "step": 3698
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005670696760586653,
+      "loss": 3.8971,
+      "step": 3699
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005668642327620356,
+      "loss": 3.8075,
+      "step": 3700
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.000566658777970669,
+      "loss": 3.7844,
+      "step": 3701
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005664533117198855,
+      "loss": 3.7637,
+      "step": 3702
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005662478340450076,
+      "loss": 3.7319,
+      "step": 3703
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005660423449813588,
+      "loss": 4.0318,
+      "step": 3704
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005658368445642654,
+      "loss": 3.9096,
+      "step": 3705
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005656313328290553,
+      "loss": 3.9262,
+      "step": 3706
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.000565425809811058,
+      "loss": 3.8712,
+      "step": 3707
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005652202755456055,
+      "loss": 4.1197,
+      "step": 3708
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005650147300680317,
+      "loss": 3.8543,
+      "step": 3709
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.000564809173413672,
+      "loss": 3.727,
+      "step": 3710
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.000564603605617864,
+      "loss": 4.0498,
+      "step": 3711
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005643980267159473,
+      "loss": 3.9682,
+      "step": 3712
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005641924367432633,
+      "loss": 4.0535,
+      "step": 3713
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005639868357351553,
+      "loss": 3.7912,
+      "step": 3714
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005637812237269682,
+      "loss": 4.0648,
+      "step": 3715
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005635756007540494,
+      "loss": 4.0113,
+      "step": 3716
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.000563369966851748,
+      "loss": 3.8446,
+      "step": 3717
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005631643220554146,
+      "loss": 3.8702,
+      "step": 3718
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005629586664004017,
+      "loss": 3.8983,
+      "step": 3719
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005627529999220644,
+      "loss": 3.867,
+      "step": 3720
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005625473226557588,
+      "loss": 3.7925,
+      "step": 3721
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005623416346368434,
+      "loss": 3.9562,
+      "step": 3722
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005621359359006783,
+      "loss": 3.8349,
+      "step": 3723
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005619302264826255,
+      "loss": 3.8914,
+      "step": 3724
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005617245064180488,
+      "loss": 3.9983,
+      "step": 3725
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005615187757423137,
+      "loss": 3.8799,
+      "step": 3726
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005613130344907881,
+      "loss": 3.8679,
+      "step": 3727
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005611072826988409,
+      "loss": 3.9029,
+      "step": 3728
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005609015204018435,
+      "loss": 3.9113,
+      "step": 3729
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005606957476351687,
+      "loss": 3.8973,
+      "step": 3730
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005604899644341911,
+      "loss": 3.7816,
+      "step": 3731
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005602841708342876,
+      "loss": 3.9081,
+      "step": 3732
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.000560078366870836,
+      "loss": 4.052,
+      "step": 3733
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005598725525792168,
+      "loss": 3.9681,
+      "step": 3734
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005596667279948117,
+      "loss": 3.9318,
+      "step": 3735
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005594608931530043,
+      "loss": 3.8548,
+      "step": 3736
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00055925504808918,
+      "loss": 4.0186,
+      "step": 3737
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005590491928387262,
+      "loss": 3.8966,
+      "step": 3738
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005588433274370313,
+      "loss": 3.8736,
+      "step": 3739
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005586374519194864,
+      "loss": 3.8596,
+      "step": 3740
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005584315663214836,
+      "loss": 3.8934,
+      "step": 3741
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005582256706784173,
+      "loss": 3.9798,
+      "step": 3742
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005580197650256832,
+      "loss": 3.894,
+      "step": 3743
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005578138493986788,
+      "loss": 3.7736,
+      "step": 3744
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005576079238328034,
+      "loss": 3.9497,
+      "step": 3745
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005574019883634583,
+      "loss": 4.0746,
+      "step": 3746
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005571960430260457,
+      "loss": 3.9025,
+      "step": 3747
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005569900878559704,
+      "loss": 4.0189,
+      "step": 3748
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005567841228886381,
+      "loss": 3.8962,
+      "step": 3749
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.000556578148159457,
+      "loss": 3.9839,
+      "step": 3750
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005563721637038362,
+      "loss": 4.0031,
+      "step": 3751
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.000556166169557187,
+      "loss": 3.9362,
+      "step": 3752
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005559601657549219,
+      "loss": 3.8777,
+      "step": 3753
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005557541523324558,
+      "loss": 3.8604,
+      "step": 3754
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005555481293252044,
+      "loss": 3.9561,
+      "step": 3755
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005553420967685857,
+      "loss": 4.0394,
+      "step": 3756
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005551360546980191,
+      "loss": 3.9276,
+      "step": 3757
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005549300031489252,
+      "loss": 3.9177,
+      "step": 3758
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005547239421567271,
+      "loss": 3.9065,
+      "step": 3759
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005545178717568491,
+      "loss": 3.9884,
+      "step": 3760
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005543117919847167,
+      "loss": 4.0439,
+      "step": 3761
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005541057028757576,
+      "loss": 3.9689,
+      "step": 3762
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005538996044654011,
+      "loss": 3.8975,
+      "step": 3763
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005536934967890777,
+      "loss": 4.0033,
+      "step": 3764
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005534873798822196,
+      "loss": 3.9073,
+      "step": 3765
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005532812537802611,
+      "loss": 3.8532,
+      "step": 3766
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005530751185186372,
+      "loss": 3.9198,
+      "step": 3767
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.000552868974132785,
+      "loss": 3.9009,
+      "step": 3768
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005526628206581433,
+      "loss": 3.9991,
+      "step": 3769
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005524566581301525,
+      "loss": 3.8324,
+      "step": 3770
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005522504865842537,
+      "loss": 3.8357,
+      "step": 3771
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005520443060558906,
+      "loss": 4.0094,
+      "step": 3772
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.000551838116580508,
+      "loss": 4.0607,
+      "step": 3773
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005516319181935521,
+      "loss": 3.7995,
+      "step": 3774
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005514257109304711,
+      "loss": 3.7824,
+      "step": 3775
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005512194948267139,
+      "loss": 3.9229,
+      "step": 3776
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.000551013269917732,
+      "loss": 3.9284,
+      "step": 3777
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005508070362389776,
+      "loss": 4.0119,
+      "step": 3778
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005506007938259045,
+      "loss": 3.8854,
+      "step": 3779
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005503945427139684,
+      "loss": 3.8255,
+      "step": 3780
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005501882829386263,
+      "loss": 3.9638,
+      "step": 3781
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005499820145353363,
+      "loss": 3.8392,
+      "step": 3782
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005497757375395588,
+      "loss": 3.8447,
+      "step": 3783
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005495694519867549,
+      "loss": 3.9854,
+      "step": 3784
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005493631579123876,
+      "loss": 4.0736,
+      "step": 3785
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.000549156855351921,
+      "loss": 3.9168,
+      "step": 3786
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005489505443408211,
+      "loss": 3.9922,
+      "step": 3787
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005487442249145553,
+      "loss": 3.8802,
+      "step": 3788
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005485378971085921,
+      "loss": 4.0033,
+      "step": 3789
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005483315609584016,
+      "loss": 3.7642,
+      "step": 3790
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005481252164994554,
+      "loss": 3.7745,
+      "step": 3791
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005479188637672269,
+      "loss": 4.0279,
+      "step": 3792
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00054771250279719,
+      "loss": 3.9641,
+      "step": 3793
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005475061336248207,
+      "loss": 3.955,
+      "step": 3794
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005472997562855962,
+      "loss": 3.7835,
+      "step": 3795
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005470933708149955,
+      "loss": 3.8716,
+      "step": 3796
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005468869772484982,
+      "loss": 3.9218,
+      "step": 3797
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.000546680575621586,
+      "loss": 3.8801,
+      "step": 3798
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005464741659697419,
+      "loss": 3.8984,
+      "step": 3799
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005462677483284496,
+      "loss": 3.8973,
+      "step": 3800
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.000546061322733195,
+      "loss": 3.9468,
+      "step": 3801
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005458548892194651,
+      "loss": 3.8065,
+      "step": 3802
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005456484478227482,
+      "loss": 3.9076,
+      "step": 3803
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005454419985785339,
+      "loss": 3.8687,
+      "step": 3804
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005452355415223131,
+      "loss": 3.8871,
+      "step": 3805
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005450290766895786,
+      "loss": 3.9271,
+      "step": 3806
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005448226041158236,
+      "loss": 3.7918,
+      "step": 3807
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005446161238365433,
+      "loss": 3.8878,
+      "step": 3808
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005444096358872341,
+      "loss": 3.9085,
+      "step": 3809
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005442031403033937,
+      "loss": 3.7314,
+      "step": 3810
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005439966371205209,
+      "loss": 3.7816,
+      "step": 3811
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005437901263741163,
+      "loss": 3.8919,
+      "step": 3812
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005435836080996812,
+      "loss": 3.9223,
+      "step": 3813
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005433770823327186,
+      "loss": 4.0539,
+      "step": 3814
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005431705491087326,
+      "loss": 3.9304,
+      "step": 3815
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.000542964008463229,
+      "loss": 3.8411,
+      "step": 3816
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005427574604317142,
+      "loss": 3.7518,
+      "step": 3817
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.000542550905049696,
+      "loss": 3.827,
+      "step": 3818
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005423443423526841,
+      "loss": 3.9153,
+      "step": 3819
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.000542137772376189,
+      "loss": 3.9919,
+      "step": 3820
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005419311951557221,
+      "loss": 3.9843,
+      "step": 3821
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005417246107267968,
+      "loss": 3.9518,
+      "step": 3822
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005415180191249271,
+      "loss": 3.8053,
+      "step": 3823
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005413114203856286,
+      "loss": 3.84,
+      "step": 3824
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005411048145444182,
+      "loss": 3.97,
+      "step": 3825
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005408982016368133,
+      "loss": 3.9422,
+      "step": 3826
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005406915816983336,
+      "loss": 3.9064,
+      "step": 3827
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005404849547644994,
+      "loss": 3.9065,
+      "step": 3828
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005402783208708319,
+      "loss": 3.8322,
+      "step": 3829
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005400716800528542,
+      "loss": 3.8891,
+      "step": 3830
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005398650323460903,
+      "loss": 3.8257,
+      "step": 3831
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.000539658377786065,
+      "loss": 3.8959,
+      "step": 3832
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005394517164083047,
+      "loss": 3.9723,
+      "step": 3833
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005392450482483372,
+      "loss": 3.8651,
+      "step": 3834
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005390383733416909,
+      "loss": 3.7823,
+      "step": 3835
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005388316917238955,
+      "loss": 3.9116,
+      "step": 3836
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005386250034304823,
+      "loss": 4.0195,
+      "step": 3837
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005384183084969831,
+      "loss": 3.9175,
+      "step": 3838
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005382116069589315,
+      "loss": 3.9455,
+      "step": 3839
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005380048988518613,
+      "loss": 3.7935,
+      "step": 3840
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005377981842113087,
+      "loss": 4.0413,
+      "step": 3841
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005375914630728099,
+      "loss": 3.9957,
+      "step": 3842
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005373847354719029,
+      "loss": 3.9176,
+      "step": 3843
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005371780014441264,
+      "loss": 3.8118,
+      "step": 3844
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005369712610250206,
+      "loss": 3.8806,
+      "step": 3845
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005367645142501262,
+      "loss": 3.9624,
+      "step": 3846
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005365577611549855,
+      "loss": 3.909,
+      "step": 3847
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005363510017751422,
+      "loss": 3.8143,
+      "step": 3848
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005361442361461402,
+      "loss": 3.8814,
+      "step": 3849
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005359374643035248,
+      "loss": 4.006,
+      "step": 3850
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005357306862828427,
+      "loss": 3.9836,
+      "step": 3851
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005355239021196416,
+      "loss": 3.9329,
+      "step": 3852
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005353171118494698,
+      "loss": 3.8428,
+      "step": 3853
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005351103155078768,
+      "loss": 3.8356,
+      "step": 3854
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005349035131304138,
+      "loss": 3.715,
+      "step": 3855
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005346967047526321,
+      "loss": 4.043,
+      "step": 3856
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005344898904100848,
+      "loss": 3.89,
+      "step": 3857
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005342830701383254,
+      "loss": 3.8264,
+      "step": 3858
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005340762439729088,
+      "loss": 3.8352,
+      "step": 3859
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005338694119493908,
+      "loss": 3.8902,
+      "step": 3860
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005336625741033283,
+      "loss": 3.8323,
+      "step": 3861
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005334557304702789,
+      "loss": 3.778,
+      "step": 3862
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005332488810858017,
+      "loss": 3.9127,
+      "step": 3863
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005330420259854564,
+      "loss": 3.9078,
+      "step": 3864
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005328351652048037,
+      "loss": 4.0012,
+      "step": 3865
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005326282987794056,
+      "loss": 3.9876,
+      "step": 3866
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005324214267448245,
+      "loss": 3.7988,
+      "step": 3867
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005322145491366243,
+      "loss": 3.7736,
+      "step": 3868
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005320076659903698,
+      "loss": 3.8577,
+      "step": 3869
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005318007773416263,
+      "loss": 3.8034,
+      "step": 3870
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005315938832259606,
+      "loss": 3.7969,
+      "step": 3871
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00053138698367894,
+      "loss": 3.85,
+      "step": 3872
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005311800787361332,
+      "loss": 3.9418,
+      "step": 3873
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005309731684331093,
+      "loss": 3.8881,
+      "step": 3874
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005307662528054387,
+      "loss": 3.9713,
+      "step": 3875
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005305593318886924,
+      "loss": 3.9869,
+      "step": 3876
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005303524057184429,
+      "loss": 3.9229,
+      "step": 3877
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005301454743302629,
+      "loss": 3.8943,
+      "step": 3878
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005299385377597265,
+      "loss": 3.9116,
+      "step": 3879
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005297315960424081,
+      "loss": 3.8654,
+      "step": 3880
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005295246492138839,
+      "loss": 3.866,
+      "step": 3881
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005293176973097302,
+      "loss": 3.9009,
+      "step": 3882
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005291107403655242,
+      "loss": 3.9963,
+      "step": 3883
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005289037784168448,
+      "loss": 3.9654,
+      "step": 3884
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005286968114992706,
+      "loss": 3.7978,
+      "step": 3885
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005284898396483819,
+      "loss": 3.9451,
+      "step": 3886
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005282828628997595,
+      "loss": 4.0281,
+      "step": 3887
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.000528075881288985,
+      "loss": 3.871,
+      "step": 3888
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005278688948516412,
+      "loss": 3.7775,
+      "step": 3889
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005276619036233111,
+      "loss": 3.876,
+      "step": 3890
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005274549076395792,
+      "loss": 3.9757,
+      "step": 3891
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005272479069360302,
+      "loss": 3.8861,
+      "step": 3892
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005270409015482503,
+      "loss": 3.9644,
+      "step": 3893
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005268338915118258,
+      "loss": 3.9466,
+      "step": 3894
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005266268768623444,
+      "loss": 3.9331,
+      "step": 3895
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005264198576353942,
+      "loss": 3.9661,
+      "step": 3896
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005262128338665641,
+      "loss": 3.9432,
+      "step": 3897
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.000526005805591444,
+      "loss": 3.8595,
+      "step": 3898
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005257987728456244,
+      "loss": 3.8703,
+      "step": 3899
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005255917356646968,
+      "loss": 3.7306,
+      "step": 3900
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005253846940842528,
+      "loss": 4.0347,
+      "step": 3901
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.000525177648139886,
+      "loss": 3.9178,
+      "step": 3902
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005249705978671895,
+      "loss": 3.862,
+      "step": 3903
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005247635433017576,
+      "loss": 3.9276,
+      "step": 3904
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005245564844791857,
+      "loss": 3.8585,
+      "step": 3905
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005243494214350693,
+      "loss": 3.7838,
+      "step": 3906
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005241423542050051,
+      "loss": 3.8312,
+      "step": 3907
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005239352828245904,
+      "loss": 3.915,
+      "step": 3908
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.000523728207329423,
+      "loss": 3.9609,
+      "step": 3909
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005235211277551018,
+      "loss": 3.8358,
+      "step": 3910
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.000523314044137226,
+      "loss": 3.9173,
+      "step": 3911
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005231069565113957,
+      "loss": 3.8282,
+      "step": 3912
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005228998649132118,
+      "loss": 3.9138,
+      "step": 3913
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005226927693782754,
+      "loss": 3.9521,
+      "step": 3914
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005224856699421889,
+      "loss": 4.0709,
+      "step": 3915
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005222785666405553,
+      "loss": 3.7833,
+      "step": 3916
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005220714595089777,
+      "loss": 3.7192,
+      "step": 3917
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005218643485830605,
+      "loss": 3.9083,
+      "step": 3918
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.000521657233898408,
+      "loss": 3.8745,
+      "step": 3919
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005214501154906261,
+      "loss": 3.8134,
+      "step": 3920
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005212429933953206,
+      "loss": 3.8932,
+      "step": 3921
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005210358676480983,
+      "loss": 3.981,
+      "step": 3922
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005208287382845666,
+      "loss": 3.8469,
+      "step": 3923
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005206216053403333,
+      "loss": 3.9106,
+      "step": 3924
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005204144688510068,
+      "loss": 3.899,
+      "step": 3925
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005202073288521964,
+      "loss": 3.8754,
+      "step": 3926
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005200001853795123,
+      "loss": 3.9348,
+      "step": 3927
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005197930384685642,
+      "loss": 3.6914,
+      "step": 3928
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005195858881549632,
+      "loss": 4.0395,
+      "step": 3929
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.000519378734474321,
+      "loss": 3.8281,
+      "step": 3930
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005191715774622499,
+      "loss": 4.0183,
+      "step": 3931
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005189644171543624,
+      "loss": 3.8344,
+      "step": 3932
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005187572535862715,
+      "loss": 3.8783,
+      "step": 3933
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005185500867935913,
+      "loss": 3.7861,
+      "step": 3934
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005183429168119362,
+      "loss": 3.887,
+      "step": 3935
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.000518135743676921,
+      "loss": 3.9781,
+      "step": 3936
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005179285674241614,
+      "loss": 3.8872,
+      "step": 3937
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005177213880892729,
+      "loss": 3.9565,
+      "step": 3938
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005175142057078727,
+      "loss": 3.8976,
+      "step": 3939
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005173070203155772,
+      "loss": 3.9587,
+      "step": 3940
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005170998319480043,
+      "loss": 3.8283,
+      "step": 3941
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005168926406407722,
+      "loss": 3.9323,
+      "step": 3942
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005166854464294993,
+      "loss": 3.8787,
+      "step": 3943
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005164782493498046,
+      "loss": 3.9081,
+      "step": 3944
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.000516271049437308,
+      "loss": 3.7857,
+      "step": 3945
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005160638467276293,
+      "loss": 3.7778,
+      "step": 3946
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.000515856641256389,
+      "loss": 4.0112,
+      "step": 3947
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005156494330592083,
+      "loss": 3.9404,
+      "step": 3948
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005154422221717086,
+      "loss": 3.9418,
+      "step": 3949
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005152350086295121,
+      "loss": 3.8137,
+      "step": 3950
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005150277924682408,
+      "loss": 3.8669,
+      "step": 3951
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005148205737235177,
+      "loss": 3.8655,
+      "step": 3952
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005146133524309664,
+      "loss": 3.976,
+      "step": 3953
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005144061286262102,
+      "loss": 3.8781,
+      "step": 3954
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005141989023448735,
+      "loss": 3.8753,
+      "step": 3955
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.000513991673622581,
+      "loss": 3.8856,
+      "step": 3956
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005137844424949575,
+      "loss": 3.9493,
+      "step": 3957
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005135772089976285,
+      "loss": 3.8588,
+      "step": 3958
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00051336997316622,
+      "loss": 3.9717,
+      "step": 3959
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005131627350363581,
+      "loss": 3.9345,
+      "step": 3960
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005129554946436694,
+      "loss": 3.7395,
+      "step": 3961
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005127482520237808,
+      "loss": 3.8605,
+      "step": 3962
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005125410072123202,
+      "loss": 4.0569,
+      "step": 3963
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.000512333760244915,
+      "loss": 4.1224,
+      "step": 3964
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005121265111571933,
+      "loss": 3.8937,
+      "step": 3965
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005119192599847837,
+      "loss": 3.9187,
+      "step": 3966
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005117120067633153,
+      "loss": 4.1086,
+      "step": 3967
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005115047515284173,
+      "loss": 3.9605,
+      "step": 3968
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005112974943157189,
+      "loss": 3.7355,
+      "step": 3969
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005110902351608503,
+      "loss": 3.6517,
+      "step": 3970
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005108829740994417,
+      "loss": 3.816,
+      "step": 3971
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005106757111671235,
+      "loss": 3.7954,
+      "step": 3972
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005104684463995271,
+      "loss": 3.9245,
+      "step": 3973
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005102611798322833,
+      "loss": 3.9571,
+      "step": 3974
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005100539115010236,
+      "loss": 3.7652,
+      "step": 3975
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00050984664144138,
+      "loss": 3.8895,
+      "step": 3976
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005096393696889847,
+      "loss": 3.8128,
+      "step": 3977
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005094320962794701,
+      "loss": 3.8795,
+      "step": 3978
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005092248212484686,
+      "loss": 3.9907,
+      "step": 3979
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005090175446316133,
+      "loss": 3.9784,
+      "step": 3980
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005088102664645378,
+      "loss": 3.8942,
+      "step": 3981
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005086029867828752,
+      "loss": 4.059,
+      "step": 3982
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005083957056222594,
+      "loss": 3.8759,
+      "step": 3983
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005081884230183247,
+      "loss": 3.7305,
+      "step": 3984
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005079811390067048,
+      "loss": 3.8146,
+      "step": 3985
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.000507773853623035,
+      "loss": 3.9074,
+      "step": 3986
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005075665669029494,
+      "loss": 3.8685,
+      "step": 3987
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005073592788820832,
+      "loss": 3.8906,
+      "step": 3988
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005071519895960717,
+      "loss": 3.808,
+      "step": 3989
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005069446990805501,
+      "loss": 3.7923,
+      "step": 3990
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005067374073711544,
+      "loss": 3.9411,
+      "step": 3991
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005065301145035204,
+      "loss": 3.8131,
+      "step": 3992
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005063228205132839,
+      "loss": 3.8072,
+      "step": 3993
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.000506115525436081,
+      "loss": 3.9342,
+      "step": 3994
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005059082293075487,
+      "loss": 3.8381,
+      "step": 3995
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005057009321633233,
+      "loss": 3.9585,
+      "step": 3996
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005054936340390414,
+      "loss": 3.872,
+      "step": 3997
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005052863349703402,
+      "loss": 3.893,
+      "step": 3998
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005050790349928566,
+      "loss": 4.0721,
+      "step": 3999
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005048717341422282,
+      "loss": 4.1289,
+      "step": 4000
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005046644324540921,
+      "loss": 3.8245,
+      "step": 4001
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.000504457129964086,
+      "loss": 3.8816,
+      "step": 4002
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005042498267078476,
+      "loss": 3.9085,
+      "step": 4003
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005040425227210148,
+      "loss": 3.8316,
+      "step": 4004
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005038352180392254,
+      "loss": 3.914,
+      "step": 4005
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005036279126981176,
+      "loss": 3.8388,
+      "step": 4006
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005034206067333296,
+      "loss": 3.8378,
+      "step": 4007
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005032133001804993,
+      "loss": 3.7134,
+      "step": 4008
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005030059930752658,
+      "loss": 3.7376,
+      "step": 4009
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005027986854532672,
+      "loss": 3.9257,
+      "step": 4010
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005025913773501421,
+      "loss": 3.9438,
+      "step": 4011
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005023840688015291,
+      "loss": 3.8766,
+      "step": 4012
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005021767598430671,
+      "loss": 4.0192,
+      "step": 4013
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005019694505103948,
+      "loss": 3.9073,
+      "step": 4014
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005017621408391513,
+      "loss": 3.9461,
+      "step": 4015
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005015548308649752,
+      "loss": 3.8953,
+      "step": 4016
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005013475206235056,
+      "loss": 3.9344,
+      "step": 4017
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005011402101503818,
+      "loss": 3.7404,
+      "step": 4018
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005009328994812425,
+      "loss": 3.7979,
+      "step": 4019
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.000500725588651727,
+      "loss": 3.8204,
+      "step": 4020
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005005182776974745,
+      "loss": 3.9249,
+      "step": 4021
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005003109666541238,
+      "loss": 3.9374,
+      "step": 4022
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005001036555573145,
+      "loss": 3.8377,
+      "step": 4023
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004998963444426855,
+      "loss": 3.8577,
+      "step": 4024
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004996890333458762,
+      "loss": 3.8732,
+      "step": 4025
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004994817223025256,
+      "loss": 3.8914,
+      "step": 4026
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.000499274411348273,
+      "loss": 4.0682,
+      "step": 4027
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004990671005187576,
+      "loss": 3.8628,
+      "step": 4028
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004988597898496183,
+      "loss": 3.8429,
+      "step": 4029
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004986524793764945,
+      "loss": 3.8329,
+      "step": 4030
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004984451691350249,
+      "loss": 3.8681,
+      "step": 4031
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004982378591608488,
+      "loss": 3.9807,
+      "step": 4032
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004980305494896052,
+      "loss": 3.9117,
+      "step": 4033
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004978232401569329,
+      "loss": 3.8648,
+      "step": 4034
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.000497615931198471,
+      "loss": 3.9387,
+      "step": 4035
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004974086226498581,
+      "loss": 3.9405,
+      "step": 4036
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004972013145467329,
+      "loss": 3.9716,
+      "step": 4037
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004969940069247343,
+      "loss": 3.7577,
+      "step": 4038
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004967866998195008,
+      "loss": 3.9171,
+      "step": 4039
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004965793932666707,
+      "loss": 3.8566,
+      "step": 4040
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004963720873018827,
+      "loss": 3.9884,
+      "step": 4041
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004961647819607749,
+      "loss": 3.9311,
+      "step": 4042
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004959574772789853,
+      "loss": 3.9391,
+      "step": 4043
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004957501732921524,
+      "loss": 3.8388,
+      "step": 4044
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.000495542870035914,
+      "loss": 3.8727,
+      "step": 4045
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.000495335567545908,
+      "loss": 3.9088,
+      "step": 4046
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004951282658577718,
+      "loss": 3.8634,
+      "step": 4047
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004949209650071435,
+      "loss": 3.9459,
+      "step": 4048
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004947136650296601,
+      "loss": 3.9376,
+      "step": 4049
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004945063659609587,
+      "loss": 3.8216,
+      "step": 4050
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004942990678366769,
+      "loss": 3.8385,
+      "step": 4051
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004940917706924513,
+      "loss": 3.8854,
+      "step": 4052
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.000493884474563919,
+      "loss": 3.9489,
+      "step": 4053
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004936771794867163,
+      "loss": 3.9093,
+      "step": 4054
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004934698854964797,
+      "loss": 3.9156,
+      "step": 4055
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004932625926288456,
+      "loss": 3.9882,
+      "step": 4056
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004930553009194499,
+      "loss": 3.9222,
+      "step": 4057
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004928480104039284,
+      "loss": 3.9229,
+      "step": 4058
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.000492640721117917,
+      "loss": 3.7465,
+      "step": 4059
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004924334330970509,
+      "loss": 3.8288,
+      "step": 4060
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004922261463769651,
+      "loss": 4.0805,
+      "step": 4061
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004920188609932951,
+      "loss": 3.7507,
+      "step": 4062
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004918115769816754,
+      "loss": 3.9307,
+      "step": 4063
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004916042943777406,
+      "loss": 3.8887,
+      "step": 4064
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004913970132171248,
+      "loss": 3.901,
+      "step": 4065
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004911897335354624,
+      "loss": 3.9172,
+      "step": 4066
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004909824553683867,
+      "loss": 3.9428,
+      "step": 4067
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004907751787515315,
+      "loss": 3.8056,
+      "step": 4068
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004905679037205302,
+      "loss": 3.8175,
+      "step": 4069
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004903606303110152,
+      "loss": 3.9192,
+      "step": 4070
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004901533585586199,
+      "loss": 3.7548,
+      "step": 4071
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004899460884989764,
+      "loss": 3.8786,
+      "step": 4072
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004897388201677168,
+      "loss": 3.9547,
+      "step": 4073
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.000489531553600473,
+      "loss": 3.9571,
+      "step": 4074
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004893242888328765,
+      "loss": 3.9407,
+      "step": 4075
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004891170259005585,
+      "loss": 3.9855,
+      "step": 4076
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00048890976483915,
+      "loss": 3.8176,
+      "step": 4077
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004887025056842814,
+      "loss": 3.9484,
+      "step": 4078
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.000488495248471583,
+      "loss": 3.8467,
+      "step": 4079
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004882879932366846,
+      "loss": 3.814,
+      "step": 4080
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004880807400152162,
+      "loss": 3.9265,
+      "step": 4081
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004878734888428068,
+      "loss": 3.9036,
+      "step": 4082
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004876662397550851,
+      "loss": 3.8565,
+      "step": 4083
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00048745899278767983,
+      "loss": 3.9557,
+      "step": 4084
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00048725174797621917,
+      "loss": 3.9472,
+      "step": 4085
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00048704450535633074,
+      "loss": 3.8568,
+      "step": 4086
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00048683726496364207,
+      "loss": 3.8116,
+      "step": 4087
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00048663002683378015,
+      "loss": 3.8808,
+      "step": 4088
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00048642279100237143,
+      "loss": 3.8606,
+      "step": 4089
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00048621555750504253,
+      "loss": 3.7125,
+      "step": 4090
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.000486008326377419,
+      "loss": 3.7619,
+      "step": 4091
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004858010976551265,
+      "loss": 4.0137,
+      "step": 4092
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004855938713737899,
+      "loss": 4.007,
+      "step": 4093
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00048538664756903375,
+      "loss": 3.9666,
+      "step": 4094
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004851794262764823,
+      "loss": 3.8739,
+      "step": 4095
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00048497220753175943,
+      "loss": 3.8632,
+      "step": 4096
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004847649913704881,
+      "loss": 3.8958,
+      "step": 4097
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00048455777782829124,
+      "loss": 3.7157,
+      "step": 4098
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00048435056694079166,
+      "loss": 3.9272,
+      "step": 4099
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.000484143358743611,
+      "loss": 3.8576,
+      "step": 4100
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004839361532723708,
+      "loss": 3.8468,
+      "step": 4101
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0004837289505626921,
+      "loss": 4.0277,
+      "step": 4102
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0004835217506501954,
+      "loss": 4.0658,
+      "step": 4103
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0004833145535705008,
+      "loss": 4.027,
+      "step": 4104
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0004831073593592279,
+      "loss": 3.8462,
+      "step": 4105
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0004829001680519958,
+      "loss": 3.9086,
+      "step": 4106
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0004826929796844228,
+      "loss": 3.767,
+      "step": 4107
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00048248579429212735,
+      "loss": 3.8168,
+      "step": 4108
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00048227861191072705,
+      "loss": 3.8822,
+      "step": 4109
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00048207143257583874,
+      "loss": 3.9645,
+      "step": 4110
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00048186425632307906,
+      "loss": 4.0516,
+      "step": 4111
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00048165708318806383,
+      "loss": 3.9422,
+      "step": 4112
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00048144991320640877,
+      "loss": 3.9241,
+      "step": 4113
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00048124274641372873,
+      "loss": 3.913,
+      "step": 4114
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00048103558284563785,
+      "loss": 3.8803,
+      "step": 4115
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00048082842253775027,
+      "loss": 3.8283,
+      "step": 4116
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0004806212655256789,
+      "loss": 3.8819,
+      "step": 4117
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00048041411184503684,
+      "loss": 3.8559,
+      "step": 4118
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00048020696153143596,
+      "loss": 3.8394,
+      "step": 4119
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0004799998146204879,
+      "loss": 3.865,
+      "step": 4120
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0004797926711478036,
+      "loss": 3.8469,
+      "step": 4121
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00047958553114899326,
+      "loss": 3.7559,
+      "step": 4122
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0004793783946596669,
+      "loss": 3.8344,
+      "step": 4123
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0004791712617154336,
+      "loss": 3.7205,
+      "step": 4124
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00047896413235190175,
+      "loss": 3.5884,
+      "step": 4125
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0004787570066046793,
+      "loss": 3.8801,
+      "step": 4126
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0004785498845093739,
+      "loss": 3.9283,
+      "step": 4127
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.000478342766101592,
+      "loss": 3.9334,
+      "step": 4128
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0004781356514169396,
+      "loss": 3.8964,
+      "step": 4129
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0004779285404910223,
+      "loss": 3.9094,
+      "step": 4130
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0004777214333594448,
+      "loss": 3.9319,
+      "step": 4131
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0004775143300578111,
+      "loss": 4.0249,
+      "step": 4132
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0004773072306217247,
+      "loss": 4.008,
+      "step": 4133
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0004771001350867885,
+      "loss": 3.9011,
+      "step": 4134
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00047689304348860436,
+      "loss": 4.0373,
+      "step": 4135
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00047668595586277415,
+      "loss": 3.9733,
+      "step": 4136
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0004764788722448983,
+      "loss": 3.9545,
+      "step": 4137
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0004762717926705771,
+      "loss": 3.857,
+      "step": 4138
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00047606471717540977,
+      "loss": 3.8219,
+      "step": 4139
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00047585764579499495,
+      "loss": 3.8533,
+      "step": 4140
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0004756505785649308,
+      "loss": 3.7727,
+      "step": 4141
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00047544351552081455,
+      "loss": 3.9709,
+      "step": 4142
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0004752364566982425,
+      "loss": 3.6858,
+      "step": 4143
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0004750294021328105,
+      "loss": 3.7887,
+      "step": 4144
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00047482235186011403,
+      "loss": 3.9263,
+      "step": 4145
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0004746153059157471,
+      "loss": 3.8744,
+      "step": 4146
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0004744082643353033,
+      "loss": 3.9093,
+      "step": 4147
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00047420122715437565,
+      "loss": 3.9715,
+      "step": 4148
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00047399419440855614,
+      "loss": 3.887,
+      "step": 4149
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.000473787166133436,
+      "loss": 3.802,
+      "step": 4150
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.000473580142364606,
+      "loss": 3.9889,
+      "step": 4151
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00047337312313765575,
+      "loss": 3.9982,
+      "step": 4152
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00047316610848817435,
+      "loss": 3.9149,
+      "step": 4153
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0004729590984517498,
+      "loss": 3.8117,
+      "step": 4154
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0004727520930639698,
+      "loss": 3.9084,
+      "step": 4155
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00047254509236042096,
+      "loss": 3.8674,
+      "step": 4156
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00047233809637668906,
+      "loss": 3.8821,
+      "step": 4157
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.000472131105148359,
+      "loss": 3.8416,
+      "step": 4158
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0004719241187110151,
+      "loss": 3.9313,
+      "step": 4159
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00047171713710024074,
+      "loss": 3.8691,
+      "step": 4160
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0004715101603516182,
+      "loss": 3.8568,
+      "step": 4161
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0004713031885007296,
+      "loss": 3.7705,
+      "step": 4162
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00047109622158315526,
+      "loss": 3.9413,
+      "step": 4163
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00047088925963447575,
+      "loss": 3.9744,
+      "step": 4164
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0004706823026902699,
+      "loss": 4.0111,
+      "step": 4165
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0004704753507861162,
+      "loss": 3.8024,
+      "step": 4166
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00047026840395759196,
+      "loss": 3.7229,
+      "step": 4167
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0004700614622402737,
+      "loss": 3.8927,
+      "step": 4168
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00046985452566973725,
+      "loss": 3.9217,
+      "step": 4169
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0004696475942815573,
+      "loss": 4.0396,
+      "step": 4170
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00046944066811130776,
+      "loss": 3.8644,
+      "step": 4171
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00046923374719456143,
+      "loss": 3.8835,
+      "step": 4172
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00046902683156689075,
+      "loss": 3.8627,
+      "step": 4173
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0004688199212638669,
+      "loss": 3.8363,
+      "step": 4174
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00046861301632106,
+      "loss": 3.8314,
+      "step": 4175
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00046840611677403943,
+      "loss": 4.0085,
+      "step": 4176
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00046819922265837376,
+      "loss": 3.8738,
+      "step": 4177
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00046799233400963037,
+      "loss": 3.9245,
+      "step": 4178
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0004677854508633757,
+      "loss": 3.9956,
+      "step": 4179
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0004675785732551756,
+      "loss": 3.8423,
+      "step": 4180
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00046737170122059446,
+      "loss": 3.9207,
+      "step": 4181
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0004671648347951963,
+      "loss": 3.8454,
+      "step": 4182
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0004669579740145436,
+      "loss": 3.9181,
+      "step": 4183
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0004667511189141983,
+      "loss": 3.8181,
+      "step": 4184
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0004665442695297212,
+      "loss": 3.9071,
+      "step": 4185
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0004663374258966718,
+      "loss": 3.8387,
+      "step": 4186
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00046613058805060933,
+      "loss": 3.9134,
+      "step": 4187
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0004659237560270914,
+      "loss": 3.9451,
+      "step": 4188
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0004657169298616747,
+      "loss": 3.9465,
+      "step": 4189
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00046551010958991536,
+      "loss": 3.9592,
+      "step": 4190
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0004653032952473678,
+      "loss": 3.9385,
+      "step": 4191
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0004650964868695863,
+      "loss": 4.0228,
+      "step": 4192
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0004648896844921232,
+      "loss": 4.0412,
+      "step": 4193
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0004646828881505304,
+      "loss": 3.791,
+      "step": 4194
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0004644760978803586,
+      "loss": 3.8477,
+      "step": 4195
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0004642693137171574,
+      "loss": 3.7695,
+      "step": 4196
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0004640625356964753,
+      "loss": 3.7721,
+      "step": 4197
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00046385576385386004,
+      "loss": 3.9817,
+      "step": 4198
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.000463648998224858,
+      "loss": 3.7057,
+      "step": 4199
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00046344223884501435,
+      "loss": 3.6904,
+      "step": 4200
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0004632354857498738,
+      "loss": 3.992,
+      "step": 4201
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00046302873897497956,
+      "loss": 3.8486,
+      "step": 4202
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00046282199855587365,
+      "loss": 3.997,
+      "step": 4203
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00046261526452809717,
+      "loss": 3.8672,
+      "step": 4204
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00046240853692719014,
+      "loss": 3.8099,
+      "step": 4205
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00046220181578869146,
+      "loss": 3.8252,
+      "step": 4206
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0004619951011481387,
+      "loss": 3.9073,
+      "step": 4207
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0004617883930410688,
+      "loss": 3.9256,
+      "step": 4208
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0004615816915030169,
+      "loss": 3.7099,
+      "step": 4209
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00046137499656951774,
+      "loss": 3.8504,
+      "step": 4210
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00046116830827610454,
+      "loss": 3.9942,
+      "step": 4211
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0004609616266583092,
+      "loss": 3.9606,
+      "step": 4212
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0004607549517516629,
+      "loss": 4.009,
+      "step": 4213
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0004605482835916954,
+      "loss": 3.7863,
+      "step": 4214
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0004603416222139352,
+      "loss": 3.9043,
+      "step": 4215
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0004601349676539099,
+      "loss": 4.0073,
+      "step": 4216
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00045992831994714594,
+      "loss": 3.8142,
+      "step": 4217
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00045972167912916806,
+      "loss": 3.8416,
+      "step": 4218
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0004595150452355006,
+      "loss": 3.9223,
+      "step": 4219
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0004593084183016664,
+      "loss": 3.9607,
+      "step": 4220
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00045910179836318666,
+      "loss": 3.8736,
+      "step": 4221
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00045889518545558195,
+      "loss": 3.8999,
+      "step": 4222
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00045868857961437143,
+      "loss": 3.7729,
+      "step": 4223
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00045848198087507295,
+      "loss": 4.0436,
+      "step": 4224
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00045827538927320326,
+      "loss": 3.9161,
+      "step": 4225
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.000458068804844278,
+      "loss": 3.9718,
+      "step": 4226
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00045786222762381125,
+      "loss": 3.8347,
+      "step": 4227
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0004576556576473158,
+      "loss": 3.8455,
+      "step": 4228
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00045744909495030393,
+      "loss": 3.8239,
+      "step": 4229
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0004572425395682859,
+      "loss": 3.9617,
+      "step": 4230
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00045703599153677104,
+      "loss": 3.7814,
+      "step": 4231
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00045682945089126737,
+      "loss": 3.8527,
+      "step": 4232
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0004566229176672814,
+      "loss": 3.7732,
+      "step": 4233
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0004564163919003189,
+      "loss": 3.8675,
+      "step": 4234
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00045620987362588394,
+      "loss": 3.8949,
+      "step": 4235
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0004560033628794792,
+      "loss": 3.7563,
+      "step": 4236
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00045579685969660634,
+      "loss": 3.9355,
+      "step": 4237
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00045559036411276595,
+      "loss": 3.8351,
+      "step": 4238
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00045538387616345685,
+      "loss": 4.0087,
+      "step": 4239
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0004551773958841765,
+      "loss": 3.8582,
+      "step": 4240
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00045497092331042157,
+      "loss": 4.0709,
+      "step": 4241
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00045476445847768694,
+      "loss": 3.8141,
+      "step": 4242
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00045455800142146617,
+      "loss": 3.8027,
+      "step": 4243
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0004543515521772519,
+      "loss": 3.9095,
+      "step": 4244
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00045414511078053506,
+      "loss": 3.8707,
+      "step": 4245
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.000453938677266805,
+      "loss": 3.9266,
+      "step": 4246
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0004537322516715505,
+      "loss": 3.9002,
+      "step": 4247
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00045352583403025825,
+      "loss": 3.8196,
+      "step": 4248
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.000453319424378414,
+      "loss": 3.8556,
+      "step": 4249
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0004531130227515019,
+      "loss": 3.898,
+      "step": 4250
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00045290662918500465,
+      "loss": 4.0081,
+      "step": 4251
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00045270024371440385,
+      "loss": 3.9518,
+      "step": 4252
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00045249386637517954,
+      "loss": 4.0143,
+      "step": 4253
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00045228749720281023,
+      "loss": 3.9029,
+      "step": 4254
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0004520811362327731,
+      "loss": 3.8988,
+      "step": 4255
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00045187478350054445,
+      "loss": 3.974,
+      "step": 4256
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0004516684390415985,
+      "loss": 3.8357,
+      "step": 4257
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.000451462102891408,
+      "loss": 3.8677,
+      "step": 4258
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0004512557750854448,
+      "loss": 3.9461,
+      "step": 4259
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.000451049455659179,
+      "loss": 3.9766,
+      "step": 4260
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00045084314464807914,
+      "loss": 3.8791,
+      "step": 4261
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00045063684208761267,
+      "loss": 3.8285,
+      "step": 4262
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00045043054801324533,
+      "loss": 3.895,
+      "step": 4263
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00045022426246044134,
+      "loss": 3.9211,
+      "step": 4264
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0004500179854646637,
+      "loss": 3.9489,
+      "step": 4265
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00044981171706137376,
+      "loss": 3.9474,
+      "step": 4266
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00044960545728603163,
+      "loss": 3.8993,
+      "step": 4267
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0004493992061740956,
+      "loss": 4.1037,
+      "step": 4268
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00044919296376102254,
+      "loss": 3.9747,
+      "step": 4269
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0004489867300822681,
+      "loss": 3.8111,
+      "step": 4270
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0004487805051732862,
+      "loss": 3.8125,
+      "step": 4271
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00044857428906952914,
+      "loss": 3.7292,
+      "step": 4272
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00044836808180644805,
+      "loss": 3.7768,
+      "step": 4273
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00044816188341949205,
+      "loss": 4.0367,
+      "step": 4274
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00044795569394410947,
+      "loss": 3.9333,
+      "step": 4275
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00044774951341574633,
+      "loss": 3.6988,
+      "step": 4276
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0004475433418698477,
+      "loss": 3.975,
+      "step": 4277
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0004473371793418567,
+      "loss": 3.9056,
+      "step": 4278
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.000447131025867215,
+      "loss": 3.9564,
+      "step": 4279
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.000446924881481363,
+      "loss": 3.9129,
+      "step": 4280
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0004467187462197392,
+      "loss": 3.7268,
+      "step": 4281
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00044651262011778046,
+      "loss": 3.9719,
+      "step": 4282
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00044630650321092236,
+      "loss": 3.7867,
+      "step": 4283
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00044610039553459884,
+      "loss": 3.7521,
+      "step": 4284
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0004458942971242423,
+      "loss": 3.8996,
+      "step": 4285
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0004456882080152834,
+      "loss": 3.8951,
+      "step": 4286
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00044548212824315104,
+      "loss": 3.914,
+      "step": 4287
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0004452760578432729,
+      "loss": 3.9235,
+      "step": 4288
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0004450699968510749,
+      "loss": 3.8796,
+      "step": 4289
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00044486394530198116,
+      "loss": 3.9239,
+      "step": 4290
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00044465790323141446,
+      "loss": 4.0355,
+      "step": 4291
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0004444518706747956,
+      "loss": 3.9002,
+      "step": 4292
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0004442458476675443,
+      "loss": 3.7744,
+      "step": 4293
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0004440398342450781,
+      "loss": 3.7849,
+      "step": 4294
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0004438338304428132,
+      "loss": 3.8276,
+      "step": 4295
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.000443627836296164,
+      "loss": 3.8883,
+      "step": 4296
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00044342185184054314,
+      "loss": 3.9607,
+      "step": 4297
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00044321587711136195,
+      "loss": 4.0181,
+      "step": 4298
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00044300991214402986,
+      "loss": 3.842,
+      "step": 4299
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00044280395697395437,
+      "loss": 3.8519,
+      "step": 4300
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0004425980116365419,
+      "loss": 3.9464,
+      "step": 4301
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00044239207616719647,
+      "loss": 3.8858,
+      "step": 4302
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00044218615060132125,
+      "loss": 3.9559,
+      "step": 4303
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0004419802349743169,
+      "loss": 3.8982,
+      "step": 4304
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0004417743293215827,
+      "loss": 3.85,
+      "step": 4305
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0004415684336785164,
+      "loss": 3.8197,
+      "step": 4306
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0004413625480805138,
+      "loss": 3.86,
+      "step": 4307
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0004411566725629688,
+      "loss": 3.752,
+      "step": 4308
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00044095080716127414,
+      "loss": 3.8654,
+      "step": 4309
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0004407449519108202,
+      "loss": 3.7264,
+      "step": 4310
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00044053910684699576,
+      "loss": 3.8896,
+      "step": 4311
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00044033327200518834,
+      "loss": 3.8976,
+      "step": 4312
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00044012744742078327,
+      "loss": 3.8631,
+      "step": 4313
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0004399216331291641,
+      "loss": 3.8934,
+      "step": 4314
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0004397158291657125,
+      "loss": 4.0533,
+      "step": 4315
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0004395100355658089,
+      "loss": 3.9196,
+      "step": 4316
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0004393042523648315,
+      "loss": 3.8569,
+      "step": 4317
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0004390984795981566,
+      "loss": 3.9062,
+      "step": 4318
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0004388927173011592,
+      "loss": 3.8137,
+      "step": 4319
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00043868696550921195,
+      "loss": 3.8687,
+      "step": 4320
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00043848122425768627,
+      "loss": 3.836,
+      "step": 4321
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00043827549358195137,
+      "loss": 3.9219,
+      "step": 4322
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0004380697735173745,
+      "loss": 3.9009,
+      "step": 4323
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00043786406409932176,
+      "loss": 4.0706,
+      "step": 4324
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00043765836536315667,
+      "loss": 3.8462,
+      "step": 4325
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0004374526773442412,
+      "loss": 3.8587,
+      "step": 4326
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00043724700007793574,
+      "loss": 3.8156,
+      "step": 4327
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00043704133359959846,
+      "loss": 4.0009,
+      "step": 4328
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00043683567794458556,
+      "loss": 3.8751,
+      "step": 4329
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.000436630033148252,
+      "loss": 3.9055,
+      "step": 4330
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00043642439924595057,
+      "loss": 3.7482,
+      "step": 4331
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00043621877627303185,
+      "loss": 3.8589,
+      "step": 4332
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00043601316426484485,
+      "loss": 3.9021,
+      "step": 4333
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00043580756325673674,
+      "loss": 3.8855,
+      "step": 4334
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0004356019732840528,
+      "loss": 3.8268,
+      "step": 4335
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.000435396394382136,
+      "loss": 3.845,
+      "step": 4336
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00043519082658632815,
+      "loss": 4.0204,
+      "step": 4337
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0004349852699319685,
+      "loss": 4.025,
+      "step": 4338
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00043477972445439447,
+      "loss": 3.8781,
+      "step": 4339
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00043457419018894203,
+      "loss": 3.9176,
+      "step": 4340
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0004343686671709448,
+      "loss": 3.8041,
+      "step": 4341
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00043416315543573465,
+      "loss": 3.9596,
+      "step": 4342
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0004339576550186413,
+      "loss": 3.9259,
+      "step": 4343
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0004337521659549925,
+      "loss": 3.8285,
+      "step": 4344
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0004335466882801146,
+      "loss": 3.7569,
+      "step": 4345
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0004333412220293313,
+      "loss": 3.9989,
+      "step": 4346
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0004331357672379646,
+      "loss": 3.8643,
+      "step": 4347
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0004329303239413346,
+      "loss": 3.9542,
+      "step": 4348
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00043272489217475963,
+      "loss": 3.8038,
+      "step": 4349
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0004325194719735557,
+      "loss": 3.7584,
+      "step": 4350
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00043231406337303663,
+      "loss": 3.9862,
+      "step": 4351
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00043210866640851485,
+      "loss": 3.8319,
+      "step": 4352
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00043190328111530045,
+      "loss": 3.9264,
+      "step": 4353
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00043169790752870137,
+      "loss": 3.8154,
+      "step": 4354
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00043149254568402385,
+      "loss": 3.8385,
+      "step": 4355
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00043128719561657196,
+      "loss": 3.8476,
+      "step": 4356
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0004310818573616476,
+      "loss": 3.9658,
+      "step": 4357
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0004308765309545509,
+      "loss": 3.8773,
+      "step": 4358
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00043067121643058004,
+      "loss": 3.8657,
+      "step": 4359
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0004304659138250309,
+      "loss": 3.8848,
+      "step": 4360
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00043026062317319704,
+      "loss": 3.9923,
+      "step": 4361
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00043005534451037063,
+      "loss": 3.8374,
+      "step": 4362
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0004298500778718415,
+      "loss": 3.8368,
+      "step": 4363
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0004296448232928971,
+      "loss": 3.818,
+      "step": 4364
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0004294395808088232,
+      "loss": 3.7103,
+      "step": 4365
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00042923435045490323,
+      "loss": 3.9513,
+      "step": 4366
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.000429029132266419,
+      "loss": 3.8971,
+      "step": 4367
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0004288239262786497,
+      "loss": 3.9406,
+      "step": 4368
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0004286187325268725,
+      "loss": 3.7499,
+      "step": 4369
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00042841355104636283,
+      "loss": 4.0023,
+      "step": 4370
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00042820838187239364,
+      "loss": 3.7144,
+      "step": 4371
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00042800322504023573,
+      "loss": 3.8397,
+      "step": 4372
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00042779808058515824,
+      "loss": 3.8127,
+      "step": 4373
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00042759294854242773,
+      "loss": 3.9076,
+      "step": 4374
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00042738782894730874,
+      "loss": 3.942,
+      "step": 4375
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00042718272183506353,
+      "loss": 3.9021,
+      "step": 4376
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00042697762724095285,
+      "loss": 3.7933,
+      "step": 4377
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00042677254520023463,
+      "loss": 3.8219,
+      "step": 4378
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0004265674757481647,
+      "loss": 4.0498,
+      "step": 4379
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00042636241891999706,
+      "loss": 3.9574,
+      "step": 4380
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0004261573747509833,
+      "loss": 3.7353,
+      "step": 4381
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00042595234327637277,
+      "loss": 3.7784,
+      "step": 4382
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.000425747324531413,
+      "loss": 3.8372,
+      "step": 4383
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0004255423185513489,
+      "loss": 3.7998,
+      "step": 4384
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00042533732537142335,
+      "loss": 3.8889,
+      "step": 4385
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00042513234502687723,
+      "loss": 3.8927,
+      "step": 4386
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00042492737755294883,
+      "loss": 3.811,
+      "step": 4387
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00042472242298487466,
+      "loss": 3.8932,
+      "step": 4388
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0004245174813578887,
+      "loss": 3.7098,
+      "step": 4389
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0004243125527072227,
+      "loss": 3.9807,
+      "step": 4390
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00042410763706810634,
+      "loss": 3.9935,
+      "step": 4391
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.000423902734475767,
+      "loss": 3.7892,
+      "step": 4392
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00042369784496542977,
+      "loss": 3.8721,
+      "step": 4393
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0004234929685723174,
+      "loss": 3.8888,
+      "step": 4394
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0004232881053316509,
+      "loss": 4.0186,
+      "step": 4395
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0004230832552786484,
+      "loss": 3.8254,
+      "step": 4396
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00042287841844852595,
+      "loss": 3.9223,
+      "step": 4397
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00042267359487649755,
+      "loss": 4.0668,
+      "step": 4398
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0004224687845977746,
+      "loss": 3.9052,
+      "step": 4399
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00042226398764756625,
+      "loss": 4.0064,
+      "step": 4400
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0004220592040610798,
+      "loss": 3.8434,
+      "step": 4401
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0004218544338735197,
+      "loss": 3.9506,
+      "step": 4402
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0004216496771200881,
+      "loss": 4.0212,
+      "step": 4403
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0004214449338359856,
+      "loss": 3.847,
+      "step": 4404
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00042124020405640955,
+      "loss": 3.7269,
+      "step": 4405
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00042103548781655554,
+      "loss": 3.7848,
+      "step": 4406
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00042083078515161665,
+      "loss": 3.9213,
+      "step": 4407
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00042062609609678353,
+      "loss": 3.9072,
+      "step": 4408
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0004204214206872448,
+      "loss": 3.8739,
+      "step": 4409
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0004202167589581863,
+      "loss": 3.7704,
+      "step": 4410
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0004200121109447918,
+      "loss": 4.0015,
+      "step": 4411
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00041980747668224285,
+      "loss": 3.7918,
+      "step": 4412
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00041960285620571807,
+      "loss": 3.9454,
+      "step": 4413
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00041939824955039464,
+      "loss": 3.8707,
+      "step": 4414
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00041919365675144635,
+      "loss": 3.907,
+      "step": 4415
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0004189890778440454,
+      "loss": 3.8448,
+      "step": 4416
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0004187845128633611,
+      "loss": 3.9061,
+      "step": 4417
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00041857996184456047,
+      "loss": 3.6667,
+      "step": 4418
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00041837542482280846,
+      "loss": 3.8809,
+      "step": 4419
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0004181709018332671,
+      "loss": 3.8344,
+      "step": 4420
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0004179663929110963,
+      "loss": 3.8068,
+      "step": 4421
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0004177618980914536,
+      "loss": 3.8151,
+      "step": 4422
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.000417557417409494,
+      "loss": 3.8035,
+      "step": 4423
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0004173529509003702,
+      "loss": 3.9076,
+      "step": 4424
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00041714849859923227,
+      "loss": 3.9327,
+      "step": 4425
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0004169440605412278,
+      "loss": 3.9496,
+      "step": 4426
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00041673963676150226,
+      "loss": 4.0564,
+      "step": 4427
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0004165352272951985,
+      "loss": 3.8867,
+      "step": 4428
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0004163308321774566,
+      "loss": 3.9043,
+      "step": 4429
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0004161264514434148,
+      "loss": 3.8588,
+      "step": 4430
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.000415922085128208,
+      "loss": 3.8938,
+      "step": 4431
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00041571773326696984,
+      "loss": 3.9155,
+      "step": 4432
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00041551339589483016,
+      "loss": 3.8717,
+      "step": 4433
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0004153090730469174,
+      "loss": 3.8247,
+      "step": 4434
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0004151047647583568,
+      "loss": 3.9624,
+      "step": 4435
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00041490047106427114,
+      "loss": 3.8051,
+      "step": 4436
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00041469619199978124,
+      "loss": 3.8364,
+      "step": 4437
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0004144919276000048,
+      "loss": 3.8973,
+      "step": 4438
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00041428767790005727,
+      "loss": 4.0065,
+      "step": 4439
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0004140834429350515,
+      "loss": 3.9873,
+      "step": 4440
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.000413879222740098,
+      "loss": 3.8636,
+      "step": 4441
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00041367501735030453,
+      "loss": 3.9038,
+      "step": 4442
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0004134708268007764,
+      "loss": 4.0489,
+      "step": 4443
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00041326665112661614,
+      "loss": 3.9093,
+      "step": 4444
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0004130624903629242,
+      "loss": 3.8616,
+      "step": 4445
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00041285834454479796,
+      "loss": 3.7702,
+      "step": 4446
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00041265421370733247,
+      "loss": 3.8927,
+      "step": 4447
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00041245009788562037,
+      "loss": 3.9608,
+      "step": 4448
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0004122459971147514,
+      "loss": 3.9018,
+      "step": 4449
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0004120419114298126,
+      "loss": 3.8216,
+      "step": 4450
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00041183784086588904,
+      "loss": 3.9301,
+      "step": 4451
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00041163378545806286,
+      "loss": 3.8987,
+      "step": 4452
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0004114297452414133,
+      "loss": 3.7799,
+      "step": 4453
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0004112257202510173,
+      "loss": 3.8299,
+      "step": 4454
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0004110217105219492,
+      "loss": 3.7933,
+      "step": 4455
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00041081771608928065,
+      "loss": 3.8818,
+      "step": 4456
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0004106137369880804,
+      "loss": 3.7916,
+      "step": 4457
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0004104097732534152,
+      "loss": 3.8194,
+      "step": 4458
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0004102058249203483,
+      "loss": 4.0614,
+      "step": 4459
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0004100018920239414,
+      "loss": 3.9412,
+      "step": 4460
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00040979797459925253,
+      "loss": 3.9804,
+      "step": 4461
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00040959407268133745,
+      "loss": 3.7897,
+      "step": 4462
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0004093901863052495,
+      "loss": 3.8546,
+      "step": 4463
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0004091863155060389,
+      "loss": 4.0182,
+      "step": 4464
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0004089824603187534,
+      "loss": 3.9132,
+      "step": 4465
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00040877862077843825,
+      "loss": 3.8389,
+      "step": 4466
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00040857479692013566,
+      "loss": 3.9121,
+      "step": 4467
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00040837098877888523,
+      "loss": 3.8885,
+      "step": 4468
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0004081671963897241,
+      "loss": 3.8459,
+      "step": 4469
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0004079634197876867,
+      "loss": 3.7279,
+      "step": 4470
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00040775965900780443,
+      "loss": 3.7756,
+      "step": 4471
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.000407555914085106,
+      "loss": 3.6799,
+      "step": 4472
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0004073521850546177,
+      "loss": 4.1543,
+      "step": 4473
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00040714847195136294,
+      "loss": 3.9356,
+      "step": 4474
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00040694477481036215,
+      "loss": 3.7817,
+      "step": 4475
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00040674109366663344,
+      "loss": 3.9063,
+      "step": 4476
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00040653742855519166,
+      "loss": 3.8386,
+      "step": 4477
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00040633377951104966,
+      "loss": 3.8523,
+      "step": 4478
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00040613014656921677,
+      "loss": 3.8049,
+      "step": 4479
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0004059265297646999,
+      "loss": 3.7869,
+      "step": 4480
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0004057229291325032,
+      "loss": 3.9028,
+      "step": 4481
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00040551934470762797,
+      "loss": 3.8837,
+      "step": 4482
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00040531577652507255,
+      "loss": 3.9765,
+      "step": 4483
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.000405112224619833,
+      "loss": 3.818,
+      "step": 4484
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00040490868902690204,
+      "loss": 3.8414,
+      "step": 4485
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0004047051697812697,
+      "loss": 3.998,
+      "step": 4486
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00040450166691792327,
+      "loss": 3.7585,
+      "step": 4487
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00040429818047184774,
+      "loss": 3.8084,
+      "step": 4488
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00040409471047802436,
+      "loss": 3.8965,
+      "step": 4489
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.000403891256971432,
+      "loss": 3.9161,
+      "step": 4490
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0004036878199870469,
+      "loss": 3.8665,
+      "step": 4491
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00040348439955984204,
+      "loss": 3.97,
+      "step": 4492
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0004032809957247877,
+      "loss": 3.7842,
+      "step": 4493
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0004030776085168516,
+      "loss": 3.6969,
+      "step": 4494
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0004028742379709982,
+      "loss": 3.9087,
+      "step": 4495
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.000402670884122189,
+      "loss": 3.8305,
+      "step": 4496
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00040246754700538353,
+      "loss": 3.8976,
+      "step": 4497
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0004022642266555372,
+      "loss": 4.005,
+      "step": 4498
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00040206092310760347,
+      "loss": 3.8676,
+      "step": 4499
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00040185763639653246,
+      "loss": 3.7823,
+      "step": 4500
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0004016543665572714,
+      "loss": 3.8836,
+      "step": 4501
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.000401451113624765,
+      "loss": 3.945,
+      "step": 4502
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00040124787763395463,
+      "loss": 3.947,
+      "step": 4503
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0004010446586197788,
+      "loss": 3.9802,
+      "step": 4504
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00040084145661717325,
+      "loss": 3.9173,
+      "step": 4505
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.000400638271661071,
+      "loss": 4.052,
+      "step": 4506
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00040043510378640193,
+      "loss": 3.9699,
+      "step": 4507
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0004002319530280925,
+      "loss": 3.8344,
+      "step": 4508
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0004000288194210671,
+      "loss": 3.7515,
+      "step": 4509
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00039982570300024664,
+      "loss": 3.8627,
+      "step": 4510
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00039962260380054903,
+      "loss": 3.9218,
+      "step": 4511
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0003994195218568896,
+      "loss": 3.8174,
+      "step": 4512
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0003992164572041804,
+      "loss": 4.0006,
+      "step": 4513
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00039901340987733036,
+      "loss": 3.9841,
+      "step": 4514
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0003988103799112461,
+      "loss": 3.9642,
+      "step": 4515
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0003986073673408306,
+      "loss": 3.9688,
+      "step": 4516
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0003984043722009841,
+      "loss": 3.8777,
+      "step": 4517
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0003982013945266039,
+      "loss": 4.084,
+      "step": 4518
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.000397998434352584,
+      "loss": 3.8677,
+      "step": 4519
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0003977954917138159,
+      "loss": 3.7944,
+      "step": 4520
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00039759256664518753,
+      "loss": 3.771,
+      "step": 4521
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0003973896591815841,
+      "loss": 3.8413,
+      "step": 4522
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0003971867693578879,
+      "loss": 3.8716,
+      "step": 4523
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00039698389720897774,
+      "loss": 3.8824,
+      "step": 4524
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00039678104276973016,
+      "loss": 3.9863,
+      "step": 4525
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0003965782060750178,
+      "loss": 3.8333,
+      "step": 4526
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00039637538715971095,
+      "loss": 3.7578,
+      "step": 4527
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00039617258605867633,
+      "loss": 3.9559,
+      "step": 4528
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00039596980280677774,
+      "loss": 3.9796,
+      "step": 4529
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0003957670374388762,
+      "loss": 3.9892,
+      "step": 4530
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00039556428998982924,
+      "loss": 3.8102,
+      "step": 4531
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00039536156049449153,
+      "loss": 3.7618,
+      "step": 4532
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0003951588489877145,
+      "loss": 3.9927,
+      "step": 4533
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0003949561555043469,
+      "loss": 3.9238,
+      "step": 4534
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.000394753480079234,
+      "loss": 4.0198,
+      "step": 4535
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.000394550822747218,
+      "loss": 3.8022,
+      "step": 4536
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00039434818354313787,
+      "loss": 3.768,
+      "step": 4537
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00039414556250183,
+      "loss": 3.8404,
+      "step": 4538
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0003939429596581271,
+      "loss": 3.9317,
+      "step": 4539
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00039374037504685876,
+      "loss": 3.647,
+      "step": 4540
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00039353780870285204,
+      "loss": 3.9099,
+      "step": 4541
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00039333526066092993,
+      "loss": 3.8749,
+      "step": 4542
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0003931327309559133,
+      "loss": 3.8803,
+      "step": 4543
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.000392930219622619,
+      "loss": 3.8923,
+      "step": 4544
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0003927277266958614,
+      "loss": 3.912,
+      "step": 4545
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00039252525221045116,
+      "loss": 3.8365,
+      "step": 4546
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00039232279620119586,
+      "loss": 3.9591,
+      "step": 4547
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00039212035870290037,
+      "loss": 3.7804,
+      "step": 4548
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00039191793975036583,
+      "loss": 4.0388,
+      "step": 4549
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0003917155393783903,
+      "loss": 3.7791,
+      "step": 4550
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0003915131576217688,
+      "loss": 3.8223,
+      "step": 4551
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0003913107945152934,
+      "loss": 3.848,
+      "step": 4552
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0003911084500937525,
+      "loss": 3.734,
+      "step": 4553
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00039090612439193127,
+      "loss": 3.8683,
+      "step": 4554
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00039070381744461217,
+      "loss": 3.7542,
+      "step": 4555
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0003905015292865738,
+      "loss": 3.9953,
+      "step": 4556
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00039029925995259204,
+      "loss": 3.8225,
+      "step": 4557
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00039009700947743907,
+      "loss": 3.8923,
+      "step": 4558
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00038989477789588443,
+      "loss": 3.845,
+      "step": 4559
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00038969256524269393,
+      "loss": 3.9011,
+      "step": 4560
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00038949037155263007,
+      "loss": 3.8095,
+      "step": 4561
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00038928819686045253,
+      "loss": 3.814,
+      "step": 4562
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00038908604120091754,
+      "loss": 3.801,
+      "step": 4563
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00038888390460877794,
+      "loss": 3.8519,
+      "step": 4564
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00038868178711878317,
+      "loss": 3.7974,
+      "step": 4565
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0003884796887656798,
+      "loss": 3.8254,
+      "step": 4566
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00038827760958421077,
+      "loss": 3.9009,
+      "step": 4567
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0003880755496091158,
+      "loss": 3.7778,
+      "step": 4568
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00038787350887513144,
+      "loss": 3.9509,
+      "step": 4569
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00038767148741699063,
+      "loss": 3.9103,
+      "step": 4570
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00038746948526942353,
+      "loss": 3.8305,
+      "step": 4571
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0003872675024671564,
+      "loss": 3.8486,
+      "step": 4572
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00038706553904491257,
+      "loss": 3.8903,
+      "step": 4573
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00038686359503741185,
+      "loss": 3.7259,
+      "step": 4574
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0003866616704793706,
+      "loss": 3.8157,
+      "step": 4575
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0003864597654055023,
+      "loss": 3.8098,
+      "step": 4576
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00038625787985051654,
+      "loss": 3.9649,
+      "step": 4577
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0003860560138491199,
+      "loss": 3.8633,
+      "step": 4578
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0003858541674360151,
+      "loss": 3.9513,
+      "step": 4579
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00038565234064590244,
+      "loss": 3.8156,
+      "step": 4580
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00038545053351347815,
+      "loss": 3.721,
+      "step": 4581
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00038524874607343506,
+      "loss": 3.8865,
+      "step": 4582
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00038504697836046274,
+      "loss": 3.8311,
+      "step": 4583
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00038484523040924765,
+      "loss": 3.8587,
+      "step": 4584
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0003846435022544724,
+      "loss": 3.758,
+      "step": 4585
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0003844417939308163,
+      "loss": 3.8338,
+      "step": 4586
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0003842401054729556,
+      "loss": 3.8949,
+      "step": 4587
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0003840384369155626,
+      "loss": 3.7563,
+      "step": 4588
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0003838367882933068,
+      "loss": 3.881,
+      "step": 4589
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0003836351596408537,
+      "loss": 3.8128,
+      "step": 4590
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00038343355099286583,
+      "loss": 3.8917,
+      "step": 4591
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0003832319623840018,
+      "loss": 3.8657,
+      "step": 4592
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00038303039384891715,
+      "loss": 3.9551,
+      "step": 4593
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0003828288454222638,
+      "loss": 3.7521,
+      "step": 4594
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0003826273171386904,
+      "loss": 3.8575,
+      "step": 4595
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0003824258090328418,
+      "loss": 3.9252,
+      "step": 4596
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0003822243211393596,
+      "loss": 3.8036,
+      "step": 4597
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0003820228534928818,
+      "loss": 3.8205,
+      "step": 4598
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0003818214061280435,
+      "loss": 3.8981,
+      "step": 4599
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0003816199790794754,
+      "loss": 3.9739,
+      "step": 4600
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0003814185723818053,
+      "loss": 3.8692,
+      "step": 4601
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0003812171860696574,
+      "loss": 4.0347,
+      "step": 4602
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00038101582017765226,
+      "loss": 3.6661,
+      "step": 4603
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00038081447474040686,
+      "loss": 3.7569,
+      "step": 4604
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00038061314979253506,
+      "loss": 3.8024,
+      "step": 4605
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0003804118453686469,
+      "loss": 3.8458,
+      "step": 4606
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0003802105615033488,
+      "loss": 4.096,
+      "step": 4607
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00038000929823124385,
+      "loss": 3.8396,
+      "step": 4608
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0003798080555869318,
+      "loss": 3.7479,
+      "step": 4609
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00037960683360500826,
+      "loss": 3.8291,
+      "step": 4610
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0003794056323200657,
+      "loss": 3.8948,
+      "step": 4611
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00037920445176669313,
+      "loss": 3.8708,
+      "step": 4612
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00037900329197947563,
+      "loss": 3.8543,
+      "step": 4613
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00037880215299299487,
+      "loss": 3.8685,
+      "step": 4614
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0003786010348418292,
+      "loss": 3.8677,
+      "step": 4615
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0003783999375605527,
+      "loss": 3.7992,
+      "step": 4616
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0003781988611837369,
+      "loss": 4.016,
+      "step": 4617
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.000377997805745949,
+      "loss": 4.0392,
+      "step": 4618
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00037779677128175245,
+      "loss": 3.9072,
+      "step": 4619
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0003775957578257078,
+      "loss": 3.8629,
+      "step": 4620
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0003773947654123713,
+      "loss": 3.8627,
+      "step": 4621
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00037719379407629593,
+      "loss": 3.9237,
+      "step": 4622
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0003769928438520312,
+      "loss": 3.8593,
+      "step": 4623
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0003767919147741226,
+      "loss": 4.0293,
+      "step": 4624
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.000376591006877112,
+      "loss": 3.7803,
+      "step": 4625
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0003763901201955381,
+      "loss": 3.7645,
+      "step": 4626
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00037618925476393567,
+      "loss": 3.8953,
+      "step": 4627
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0003759884106168356,
+      "loss": 3.9692,
+      "step": 4628
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00037578758778876545,
+      "loss": 3.8223,
+      "step": 4629
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.000375586786314249,
+      "loss": 3.9132,
+      "step": 4630
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00037538600622780626,
+      "loss": 3.867,
+      "step": 4631
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0003751852475639536,
+      "loss": 3.8946,
+      "step": 4632
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.000374984510357204,
+      "loss": 3.9964,
+      "step": 4633
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00037478379464206635,
+      "loss": 3.8576,
+      "step": 4634
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00037458310045304584,
+      "loss": 3.9641,
+      "step": 4635
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0003743824278246445,
+      "loss": 3.787,
+      "step": 4636
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00037418177679136,
+      "loss": 3.7187,
+      "step": 4637
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00037398114738768686,
+      "loss": 3.8023,
+      "step": 4638
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0003737805396481153,
+      "loss": 3.8807,
+      "step": 4639
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0003735799536071322,
+      "loss": 3.8113,
+      "step": 4640
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00037337938929922075,
+      "loss": 3.6536,
+      "step": 4641
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00037317884675886024,
+      "loss": 3.8474,
+      "step": 4642
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00037297832602052605,
+      "loss": 3.7853,
+      "step": 4643
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0003727778271186901,
+      "loss": 3.7788,
+      "step": 4644
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00037257735008782073,
+      "loss": 3.8529,
+      "step": 4645
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00037237689496238214,
+      "loss": 3.9439,
+      "step": 4646
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00037217646177683465,
+      "loss": 3.8662,
+      "step": 4647
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00037197605056563544,
+      "loss": 3.9116,
+      "step": 4648
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00037177566136323726,
+      "loss": 3.8827,
+      "step": 4649
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00037157529420408933,
+      "loss": 3.9811,
+      "step": 4650
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0003713749491226372,
+      "loss": 3.9395,
+      "step": 4651
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00037117462615332254,
+      "loss": 3.8874,
+      "step": 4652
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00037097432533058283,
+      "loss": 3.7459,
+      "step": 4653
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00037077404668885264,
+      "loss": 3.8125,
+      "step": 4654
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00037057379026256184,
+      "loss": 3.8508,
+      "step": 4655
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.000370373556086137,
+      "loss": 3.9531,
+      "step": 4656
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0003701733441940007,
+      "loss": 3.8872,
+      "step": 4657
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00036997315462057145,
+      "loss": 3.8336,
+      "step": 4658
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00036977298740026445,
+      "loss": 3.9523,
+      "step": 4659
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0003695728425674906,
+      "loss": 3.5946,
+      "step": 4660
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00036937272015665705,
+      "loss": 3.7138,
+      "step": 4661
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00036917262020216726,
+      "loss": 3.8052,
+      "step": 4662
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00036897254273842085,
+      "loss": 3.8735,
+      "step": 4663
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0003687724877998133,
+      "loss": 3.8913,
+      "step": 4664
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00036857245542073645,
+      "loss": 3.9065,
+      "step": 4665
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00036837244563557814,
+      "loss": 3.8317,
+      "step": 4666
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0003681724584787225,
+      "loss": 3.795,
+      "step": 4667
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00036797249398454933,
+      "loss": 3.8201,
+      "step": 4668
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00036777255218743514,
+      "loss": 3.9547,
+      "step": 4669
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00036757263312175217,
+      "loss": 3.8289,
+      "step": 4670
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0003673727368218687,
+      "loss": 3.9077,
+      "step": 4671
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00036717286332214926,
+      "loss": 3.8518,
+      "step": 4672
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0003669730126569545,
+      "loss": 3.7622,
+      "step": 4673
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00036677318486064126,
+      "loss": 3.8047,
+      "step": 4674
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0003665733799675619,
+      "loss": 3.8704,
+      "step": 4675
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0003663735980120653,
+      "loss": 4.0947,
+      "step": 4676
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00036617383902849644,
+      "loss": 3.8151,
+      "step": 4677
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.000365974103051196,
+      "loss": 3.8574,
+      "step": 4678
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0003657743901145011,
+      "loss": 3.8935,
+      "step": 4679
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00036557470025274454,
+      "loss": 3.8271,
+      "step": 4680
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0003653750335002552,
+      "loss": 4.0128,
+      "step": 4681
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00036517538989135865,
+      "loss": 3.9397,
+      "step": 4682
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0003649757694603754,
+      "loss": 3.9098,
+      "step": 4683
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00036477617224162285,
+      "loss": 4.047,
+      "step": 4684
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0003645765982694139,
+      "loss": 3.8115,
+      "step": 4685
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0003643770475780576,
+      "loss": 3.8608,
+      "step": 4686
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00036417752020185916,
+      "loss": 3.8086,
+      "step": 4687
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0003639780161751196,
+      "loss": 3.8148,
+      "step": 4688
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00036377853553213593,
+      "loss": 3.9226,
+      "step": 4689
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00036357907830720116,
+      "loss": 3.8397,
+      "step": 4690
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0003633796445346044,
+      "loss": 3.8292,
+      "step": 4691
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00036318023424863053,
+      "loss": 3.7986,
+      "step": 4692
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00036298084748356076,
+      "loss": 3.7817,
+      "step": 4693
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0003627814842736715,
+      "loss": 3.8267,
+      "step": 4694
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.000362582144653236,
+      "loss": 3.8635,
+      "step": 4695
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.000362382828656523,
+      "loss": 4.0247,
+      "step": 4696
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.000362183536317797,
+      "loss": 4.0125,
+      "step": 4697
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0003619842676713189,
+      "loss": 3.9404,
+      "step": 4698
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00036178502275134507,
+      "loss": 3.8738,
+      "step": 4699
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0003615858015921283,
+      "loss": 3.772,
+      "step": 4700
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00036138660422791667,
+      "loss": 3.8195,
+      "step": 4701
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.000361187430692955,
+      "loss": 3.9501,
+      "step": 4702
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0003609882810214832,
+      "loss": 3.9118,
+      "step": 4703
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0003607891552477373,
+      "loss": 3.8995,
+      "step": 4704
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00036059005340594957,
+      "loss": 3.9541,
+      "step": 4705
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0003603909755303479,
+      "loss": 3.7793,
+      "step": 4706
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0003601919216551559,
+      "loss": 3.7345,
+      "step": 4707
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0003599928918145934,
+      "loss": 3.8769,
+      "step": 4708
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00035979388604287575,
+      "loss": 3.8507,
+      "step": 4709
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0003595949043742147,
+      "loss": 3.8164,
+      "step": 4710
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00035939594684281735,
+      "loss": 3.7508,
+      "step": 4711
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0003591970134828866,
+      "loss": 3.9264,
+      "step": 4712
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00035899810432862166,
+      "loss": 3.8312,
+      "step": 4713
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0003587992194142173,
+      "loss": 3.8827,
+      "step": 4714
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00035860035877386385,
+      "loss": 3.9248,
+      "step": 4715
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0003584015224417482,
+      "loss": 3.8536,
+      "step": 4716
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0003582027104520523,
+      "loss": 3.9336,
+      "step": 4717
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0003580039228389541,
+      "loss": 3.8015,
+      "step": 4718
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00035780515963662786,
+      "loss": 3.6769,
+      "step": 4719
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0003576064208792433,
+      "loss": 3.9468,
+      "step": 4720
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0003574077066009657,
+      "loss": 3.8132,
+      "step": 4721
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0003572090168359563,
+      "loss": 3.7819,
+      "step": 4722
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0003570103516183724,
+      "loss": 3.8795,
+      "step": 4723
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00035681171098236674,
+      "loss": 3.9986,
+      "step": 4724
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0003566130949620878,
+      "loss": 3.7588,
+      "step": 4725
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0003564145035916803,
+      "loss": 3.9544,
+      "step": 4726
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.000356215936905284,
+      "loss": 3.9889,
+      "step": 4727
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00035601739493703515,
+      "loss": 3.8527,
+      "step": 4728
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00035581887772106536,
+      "loss": 3.9416,
+      "step": 4729
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00035562038529150184,
+      "loss": 3.9103,
+      "step": 4730
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00035542191768246807,
+      "loss": 3.8292,
+      "step": 4731
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0003552234749280827,
+      "loss": 3.7156,
+      "step": 4732
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00035502505706246025,
+      "loss": 3.7892,
+      "step": 4733
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0003548266641197113,
+      "loss": 3.6851,
+      "step": 4734
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0003546282961339418,
+      "loss": 3.7495,
+      "step": 4735
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0003544299531392533,
+      "loss": 3.8864,
+      "step": 4736
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00035423163516974356,
+      "loss": 3.9641,
+      "step": 4737
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0003540333422595057,
+      "loss": 3.786,
+      "step": 4738
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0003538350744426285,
+      "loss": 3.9025,
+      "step": 4739
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0003536368317531963,
+      "loss": 3.9207,
+      "step": 4740
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00035343861422528974,
+      "loss": 3.7675,
+      "step": 4741
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00035324042189298447,
+      "loss": 3.9251,
+      "step": 4742
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00035304225479035193,
+      "loss": 3.8671,
+      "step": 4743
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0003528441129514595,
+      "loss": 3.9312,
+      "step": 4744
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0003526459964103701,
+      "loss": 3.8168,
+      "step": 4745
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.000352447905201142,
+      "loss": 3.7619,
+      "step": 4746
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0003522498393578297,
+      "loss": 4.0195,
+      "step": 4747
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0003520517989144829,
+      "loss": 3.9025,
+      "step": 4748
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0003518537839051471,
+      "loss": 3.8531,
+      "step": 4749
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0003516557943638632,
+      "loss": 3.8946,
+      "step": 4750
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00035145783032466797,
+      "loss": 3.748,
+      "step": 4751
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00035125989182159393,
+      "loss": 3.9494,
+      "step": 4752
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0003510619788886687,
+      "loss": 3.8912,
+      "step": 4753
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0003508640915599159,
+      "loss": 3.9422,
+      "step": 4754
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0003506662298693546,
+      "loss": 3.8176,
+      "step": 4755
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00035046839385099977,
+      "loss": 3.7937,
+      "step": 4756
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0003502705835388616,
+      "loss": 3.8561,
+      "step": 4757
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00035007279896694576,
+      "loss": 3.8388,
+      "step": 4758
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0003498750401692541,
+      "loss": 3.8186,
+      "step": 4759
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00034967730717978335,
+      "loss": 3.951,
+      "step": 4760
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0003494796000325261,
+      "loss": 3.9448,
+      "step": 4761
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0003492819187614707,
+      "loss": 3.8865,
+      "step": 4762
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00034908426340060075,
+      "loss": 3.757,
+      "step": 4763
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00034888663398389524,
+      "loss": 3.8465,
+      "step": 4764
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0003486890305453293,
+      "loss": 3.9315,
+      "step": 4765
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0003484914531188733,
+      "loss": 3.7633,
+      "step": 4766
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00034829390173849297,
+      "loss": 3.8785,
+      "step": 4767
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00034809637643814964,
+      "loss": 3.6849,
+      "step": 4768
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0003478988772518001,
+      "loss": 4.0047,
+      "step": 4769
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.000347701404213397,
+      "loss": 3.8118,
+      "step": 4770
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00034750395735688803,
+      "loss": 3.8504,
+      "step": 4771
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00034730653671621663,
+      "loss": 3.8222,
+      "step": 4772
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0003471091423253216,
+      "loss": 3.8731,
+      "step": 4773
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0003469117742181378,
+      "loss": 3.8001,
+      "step": 4774
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00034671443242859464,
+      "loss": 3.9012,
+      "step": 4775
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0003465171169906175,
+      "loss": 3.7991,
+      "step": 4776
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00034631982793812744,
+      "loss": 3.9359,
+      "step": 4777
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0003461225653050405,
+      "loss": 3.8578,
+      "step": 4778
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0003459253291252684,
+      "loss": 3.7377,
+      "step": 4779
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00034572811943271855,
+      "loss": 3.8629,
+      "step": 4780
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0003455309362612934,
+      "loss": 3.792,
+      "step": 4781
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.000345333779644891,
+      "loss": 3.8472,
+      "step": 4782
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00034513664961740475,
+      "loss": 3.8156,
+      "step": 4783
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00034493954621272403,
+      "loss": 3.8124,
+      "step": 4784
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00034474246946473286,
+      "loss": 3.826,
+      "step": 4785
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0003445454194073111,
+      "loss": 3.9727,
+      "step": 4786
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00034434839607433394,
+      "loss": 3.8727,
+      "step": 4787
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.000344151399499672,
+      "loss": 3.9871,
+      "step": 4788
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00034395442971719124,
+      "loss": 3.9551,
+      "step": 4789
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0003437574867607529,
+      "loss": 3.8068,
+      "step": 4790
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0003435605706642141,
+      "loss": 3.8423,
+      "step": 4791
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00034336368146142647,
+      "loss": 3.95,
+      "step": 4792
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0003431668191862382,
+      "loss": 3.7319,
+      "step": 4793
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0003429699838724917,
+      "loss": 3.7605,
+      "step": 4794
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0003427731755540256,
+      "loss": 3.7729,
+      "step": 4795
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0003425763942646733,
+      "loss": 3.7467,
+      "step": 4796
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00034237964003826373,
+      "loss": 3.8618,
+      "step": 4797
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0003421829129086215,
+      "loss": 3.7863,
+      "step": 4798
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.000341986212909566,
+      "loss": 3.9712,
+      "step": 4799
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00034178954007491237,
+      "loss": 3.9232,
+      "step": 4800
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00034159289443847076,
+      "loss": 3.9838,
+      "step": 4801
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00034139627603404723,
+      "loss": 4.0382,
+      "step": 4802
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0003411996848954425,
+      "loss": 3.8381,
+      "step": 4803
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0003410031210564528,
+      "loss": 3.832,
+      "step": 4804
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00034080658455087,
+      "loss": 3.907,
+      "step": 4805
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0003406100754124809,
+      "loss": 3.8486,
+      "step": 4806
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0003404135936750674,
+      "loss": 3.8471,
+      "step": 4807
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00034021713937240747,
+      "loss": 4.0009,
+      "step": 4808
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00034002071253827357,
+      "loss": 3.8756,
+      "step": 4809
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00033982431320643357,
+      "loss": 4.0469,
+      "step": 4810
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0003396279414106513,
+      "loss": 3.8476,
+      "step": 4811
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.000339431597184685,
+      "loss": 3.7419,
+      "step": 4812
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00033923528056228875,
+      "loss": 3.8786,
+      "step": 4813
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0003390389915772115,
+      "loss": 3.7479,
+      "step": 4814
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00033884273026319747,
+      "loss": 3.6625,
+      "step": 4815
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0003386464966539865,
+      "loss": 3.9657,
+      "step": 4816
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0003384502907833134,
+      "loss": 3.9473,
+      "step": 4817
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.000338254112684908,
+      "loss": 3.8498,
+      "step": 4818
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0003380579623924959,
+      "loss": 3.8512,
+      "step": 4819
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0003378618399397972,
+      "loss": 3.7345,
+      "step": 4820
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0003376657453605281,
+      "loss": 3.9098,
+      "step": 4821
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00033746967868839917,
+      "loss": 3.8495,
+      "step": 4822
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0003372736399571169,
+      "loss": 3.858,
+      "step": 4823
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0003370776292003823,
+      "loss": 3.9784,
+      "step": 4824
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00033688164645189194,
+      "loss": 3.9195,
+      "step": 4825
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0003366856917453377,
+      "loss": 3.9618,
+      "step": 4826
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00033648976511440633,
+      "loss": 3.8728,
+      "step": 4827
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00033629386659277984,
+      "loss": 3.7253,
+      "step": 4828
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0003360979962141355,
+      "loss": 3.861,
+      "step": 4829
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0003359021540121457,
+      "loss": 3.8712,
+      "step": 4830
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0003357063400204781,
+      "loss": 3.8058,
+      "step": 4831
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0003355105542727954,
+      "loss": 3.8831,
+      "step": 4832
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00033531479680275515,
+      "loss": 3.7593,
+      "step": 4833
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00033511906764401066,
+      "loss": 3.8481,
+      "step": 4834
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00033492336683020997,
+      "loss": 3.9213,
+      "step": 4835
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0003347276943949961,
+      "loss": 3.8616,
+      "step": 4836
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00033453205037200763,
+      "loss": 3.8014,
+      "step": 4837
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00033433643479487775,
+      "loss": 3.8957,
+      "step": 4838
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0003341408476972355,
+      "loss": 3.7339,
+      "step": 4839
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00033394528911270425,
+      "loss": 3.7394,
+      "step": 4840
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.000333749759074903,
+      "loss": 3.778,
+      "step": 4841
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0003335542576174455,
+      "loss": 3.8945,
+      "step": 4842
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00033335878477394056,
+      "loss": 3.7713,
+      "step": 4843
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0003331633405779926,
+      "loss": 3.8825,
+      "step": 4844
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0003329679250632004,
+      "loss": 3.8101,
+      "step": 4845
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0003327725382631582,
+      "loss": 3.6855,
+      "step": 4846
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00033257718021145543,
+      "loss": 3.8802,
+      "step": 4847
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00033238185094167625,
+      "loss": 3.9434,
+      "step": 4848
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00033218655048740023,
+      "loss": 3.7834,
+      "step": 4849
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0003319912788822016,
+      "loss": 3.7919,
+      "step": 4850
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00033179603615964974,
+      "loss": 3.8361,
+      "step": 4851
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00033160082235330935,
+      "loss": 3.9216,
+      "step": 4852
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0003314056374967399,
+      "loss": 3.818,
+      "step": 4853
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0003312104816234958,
+      "loss": 3.8731,
+      "step": 4854
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00033101535476712674,
+      "loss": 3.829,
+      "step": 4855
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0003308202569611772,
+      "loss": 3.7294,
+      "step": 4856
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00033062518823918663,
+      "loss": 3.7302,
+      "step": 4857
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00033043014863468983,
+      "loss": 3.8764,
+      "step": 4858
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0003302351381812164,
+      "loss": 3.9106,
+      "step": 4859
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0003300401569122908,
+      "loss": 3.7418,
+      "step": 4860
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00032984520486143256,
+      "loss": 3.8622,
+      "step": 4861
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00032965028206215625,
+      "loss": 3.8874,
+      "step": 4862
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0003294553885479713,
+      "loss": 4.058,
+      "step": 4863
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0003292605243523821,
+      "loss": 3.7859,
+      "step": 4864
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00032906568950888827,
+      "loss": 3.8227,
+      "step": 4865
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00032887088405098377,
+      "loss": 3.9368,
+      "step": 4866
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0003286761080121585,
+      "loss": 3.8903,
+      "step": 4867
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00032848136142589636,
+      "loss": 3.8792,
+      "step": 4868
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00032828664432567653,
+      "loss": 3.7526,
+      "step": 4869
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00032809195674497326,
+      "loss": 3.8861,
+      "step": 4870
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0003278972987172556,
+      "loss": 3.8725,
+      "step": 4871
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00032770267027598736,
+      "loss": 3.7965,
+      "step": 4872
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0003275080714546277,
+      "loss": 3.8362,
+      "step": 4873
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00032731350228663023,
+      "loss": 4.0625,
+      "step": 4874
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0003271189628054434,
+      "loss": 3.846,
+      "step": 4875
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0003269244530445112,
+      "loss": 3.9269,
+      "step": 4876
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00032672997303727217,
+      "loss": 3.9188,
+      "step": 4877
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0003265355228171594,
+      "loss": 3.9332,
+      "step": 4878
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00032634110241760116,
+      "loss": 3.9029,
+      "step": 4879
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0003261467118720207,
+      "loss": 4.035,
+      "step": 4880
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.000325952351213836,
+      "loss": 4.0252,
+      "step": 4881
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00032575802047645977,
+      "loss": 3.9081,
+      "step": 4882
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0003255637196932999,
+      "loss": 3.9926,
+      "step": 4883
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00032536944889775864,
+      "loss": 3.826,
+      "step": 4884
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0003251752081232338,
+      "loss": 3.9523,
+      "step": 4885
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0003249809974031175,
+      "loss": 3.8889,
+      "step": 4886
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00032478681677079654,
+      "loss": 3.8558,
+      "step": 4887
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0003245926662596531,
+      "loss": 3.866,
+      "step": 4888
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00032439854590306396,
+      "loss": 3.8616,
+      "step": 4889
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0003242044557344003,
+      "loss": 3.8898,
+      "step": 4890
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00032401039578702887,
+      "loss": 3.8861,
+      "step": 4891
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0003238163660943106,
+      "loss": 3.7443,
+      "step": 4892
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00032362236668960145,
+      "loss": 3.82,
+      "step": 4893
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0003234283976062521,
+      "loss": 3.9357,
+      "step": 4894
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0003232344588776083,
+      "loss": 3.8647,
+      "step": 4895
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00032304055053701037,
+      "loss": 3.854,
+      "step": 4896
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0003228466726177932,
+      "loss": 3.8476,
+      "step": 4897
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0003226528251532868,
+      "loss": 3.8235,
+      "step": 4898
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0003224590081768157,
+      "loss": 3.7898,
+      "step": 4899
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00032226522172169923,
+      "loss": 3.9681,
+      "step": 4900
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0003220714658212518,
+      "loss": 3.9568,
+      "step": 4901
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0003218777405087821,
+      "loss": 3.8693,
+      "step": 4902
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00032168404581759357,
+      "loss": 3.8571,
+      "step": 4903
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.000321490381780985,
+      "loss": 3.7763,
+      "step": 4904
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0003212967484322491,
+      "loss": 3.7336,
+      "step": 4905
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00032110314580467404,
+      "loss": 4.0765,
+      "step": 4906
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0003209095739315421,
+      "loss": 3.8337,
+      "step": 4907
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0003207160328461305,
+      "loss": 3.9499,
+      "step": 4908
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00032052252258171136,
+      "loss": 3.8401,
+      "step": 4909
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0003203290431715513,
+      "loss": 3.979,
+      "step": 4910
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00032013559464891164,
+      "loss": 3.8422,
+      "step": 4911
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0003199421770470482,
+      "loss": 3.9357,
+      "step": 4912
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00031974879039921213,
+      "loss": 3.8093,
+      "step": 4913
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0003195554347386487,
+      "loss": 3.8119,
+      "step": 4914
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00031936211009859784,
+      "loss": 3.7638,
+      "step": 4915
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00031916881651229445,
+      "loss": 3.7904,
+      "step": 4916
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0003189755540129679,
+      "loss": 3.7219,
+      "step": 4917
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0003187823226338421,
+      "loss": 3.7629,
+      "step": 4918
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.000318589122408136,
+      "loss": 3.9087,
+      "step": 4919
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0003183959533690629,
+      "loss": 3.8336,
+      "step": 4920
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00031820281554983054,
+      "loss": 3.8531,
+      "step": 4921
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0003180097089836419,
+      "loss": 3.7856,
+      "step": 4922
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0003178166337036941,
+      "loss": 3.8692,
+      "step": 4923
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00031762358974317905,
+      "loss": 3.7932,
+      "step": 4924
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00031743057713528325,
+      "loss": 3.9931,
+      "step": 4925
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00031723759591318755,
+      "loss": 3.8974,
+      "step": 4926
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0003170446461100681,
+      "loss": 3.9861,
+      "step": 4927
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.000316851727759095,
+      "loss": 3.8592,
+      "step": 4928
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00031665884089343295,
+      "loss": 3.8965,
+      "step": 4929
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0003164659855462417,
+      "loss": 3.8054,
+      "step": 4930
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00031627316175067514,
+      "loss": 4.0816,
+      "step": 4931
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0003160803695398822,
+      "loss": 3.8261,
+      "step": 4932
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00031588760894700577,
+      "loss": 3.8633,
+      "step": 4933
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0003156948800051839,
+      "loss": 3.9654,
+      "step": 4934
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0003155021827475488,
+      "loss": 3.8372,
+      "step": 4935
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0003153095172072273,
+      "loss": 3.9137,
+      "step": 4936
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00031511688341734096,
+      "loss": 3.7374,
+      "step": 4937
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0003149242814110057,
+      "loss": 3.915,
+      "step": 4938
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00031473171122133205,
+      "loss": 3.8403,
+      "step": 4939
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00031453917288142496,
+      "loss": 3.8193,
+      "step": 4940
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0003143466664243841,
+      "loss": 3.8783,
+      "step": 4941
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0003141541918833037,
+      "loss": 3.9362,
+      "step": 4942
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0003139617492912722,
+      "loss": 3.8176,
+      "step": 4943
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00031376933868137264,
+      "loss": 3.8299,
+      "step": 4944
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0003135769600866828,
+      "loss": 3.7949,
+      "step": 4945
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0003133846135402748,
+      "loss": 3.6545,
+      "step": 4946
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.000313192299075215,
+      "loss": 3.8029,
+      "step": 4947
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0003130000167245648,
+      "loss": 3.7918,
+      "step": 4948
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00031280776652137937,
+      "loss": 3.8978,
+      "step": 4949
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0003126155484987092,
+      "loss": 3.9215,
+      "step": 4950
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0003124233626895985,
+      "loss": 3.8402,
+      "step": 4951
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00031223120912708647,
+      "loss": 3.7774,
+      "step": 4952
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0003120390878442063,
+      "loss": 3.663,
+      "step": 4953
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0003118469988739859,
+      "loss": 3.82,
+      "step": 4954
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0003116549422494478,
+      "loss": 3.8446,
+      "step": 4955
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0003114629180036086,
+      "loss": 3.7469,
+      "step": 4956
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0003112709261694793,
+      "loss": 3.8174,
+      "step": 4957
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00031107896678006573,
+      "loss": 3.8499,
+      "step": 4958
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0003108870398683681,
+      "loss": 3.8824,
+      "step": 4959
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00031069514546738076,
+      "loss": 3.8272,
+      "step": 4960
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00031050328361009245,
+      "loss": 3.9179,
+      "step": 4961
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00031031145432948643,
+      "loss": 3.8687,
+      "step": 4962
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00031011965765854056,
+      "loss": 3.771,
+      "step": 4963
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0003099278936302269,
+      "loss": 3.6784,
+      "step": 4964
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00030973616227751165,
+      "loss": 3.9303,
+      "step": 4965
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0003095444636333559,
+      "loss": 3.8796,
+      "step": 4966
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0003093527977307149,
+      "loss": 3.7627,
+      "step": 4967
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00030916116460253783,
+      "loss": 3.9536,
+      "step": 4968
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0003089695642817691,
+      "loss": 3.9414,
+      "step": 4969
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0003087779968013471,
+      "loss": 3.79,
+      "step": 4970
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00030858646219420415,
+      "loss": 3.7697,
+      "step": 4971
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0003083949604932674,
+      "loss": 3.8637,
+      "step": 4972
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00030820349173145824,
+      "loss": 3.7656,
+      "step": 4973
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0003080120559416924,
+      "loss": 3.8416,
+      "step": 4974
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00030782065315687975,
+      "loss": 3.7147,
+      "step": 4975
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0003076292834099248,
+      "loss": 3.9421,
+      "step": 4976
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00030743794673372606,
+      "loss": 3.8926,
+      "step": 4977
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0003072466431611768,
+      "loss": 3.8012,
+      "step": 4978
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.000307055372725164,
+      "loss": 3.7705,
+      "step": 4979
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0003068641354585697,
+      "loss": 3.9861,
+      "step": 4980
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00030667293139426946,
+      "loss": 3.7844,
+      "step": 4981
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00030648176056513355,
+      "loss": 3.9286,
+      "step": 4982
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0003062906230040263,
+      "loss": 3.8559,
+      "step": 4983
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0003060995187438068,
+      "loss": 3.8109,
+      "step": 4984
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00030590844781732795,
+      "loss": 3.8058,
+      "step": 4985
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0003057174102574367,
+      "loss": 3.8455,
+      "step": 4986
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00030552640609697514,
+      "loss": 3.7729,
+      "step": 4987
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00030533543536877895,
+      "loss": 3.8465,
+      "step": 4988
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00030514449810567826,
+      "loss": 3.7578,
+      "step": 4989
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00030495359434049707,
+      "loss": 3.8306,
+      "step": 4990
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0003047627241060544,
+      "loss": 3.8486,
+      "step": 4991
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00030457188743516274,
+      "loss": 3.853,
+      "step": 4992
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00030438108436062916,
+      "loss": 3.8267,
+      "step": 4993
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0003041903149152552,
+      "loss": 3.7515,
+      "step": 4994
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0003039995791318358,
+      "loss": 3.7349,
+      "step": 4995
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00030380887704316116,
+      "loss": 3.8206,
+      "step": 4996
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00030361820868201494,
+      "loss": 3.8573,
+      "step": 4997
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00030342757408117544,
+      "loss": 3.8118,
+      "step": 4998
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0003032369732734148,
+      "loss": 3.7482,
+      "step": 4999
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0003030464062914995,
+      "loss": 3.7985,
+      "step": 5000
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0003028558731681902,
+      "loss": 3.753,
+      "step": 5001
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0003026653739362418,
+      "loss": 3.8288,
+      "step": 5002
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00030247490862840337,
+      "loss": 3.8886,
+      "step": 5003
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.000302284477277418,
+      "loss": 3.7812,
+      "step": 5004
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0003020940799160229,
+      "loss": 3.8125,
+      "step": 5005
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00030190371657695,
+      "loss": 3.8509,
+      "step": 5006
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00030171338729292484,
+      "loss": 3.8287,
+      "step": 5007
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.000301523092096667,
+      "loss": 3.8674,
+      "step": 5008
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00030133283102089063,
+      "loss": 3.893,
+      "step": 5009
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00030114260409830384,
+      "loss": 3.9038,
+      "step": 5010
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00030095241136160857,
+      "loss": 3.7274,
+      "step": 5011
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00030076225284350144,
+      "loss": 3.8447,
+      "step": 5012
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0003005721285766729,
+      "loss": 3.8636,
+      "step": 5013
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00030038203859380716,
+      "loss": 3.8386,
+      "step": 5014
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00030019198292758324,
+      "loss": 3.9215,
+      "step": 5015
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00030000196161067394,
+      "loss": 3.9243,
+      "step": 5016
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00029981197467574607,
+      "loss": 3.7169,
+      "step": 5017
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0002996220221554604,
+      "loss": 3.659,
+      "step": 5018
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00029943210408247216,
+      "loss": 3.8499,
+      "step": 5019
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00029924222048943044,
+      "loss": 3.8456,
+      "step": 5020
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00029905237140897845,
+      "loss": 3.8434,
+      "step": 5021
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00029886255687375316,
+      "loss": 3.8245,
+      "step": 5022
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0002986727769163862,
+      "loss": 3.8951,
+      "step": 5023
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.000298483031569503,
+      "loss": 3.7696,
+      "step": 5024
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0002982933208657228,
+      "loss": 3.8616,
+      "step": 5025
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0002981036448376591,
+      "loss": 3.8021,
+      "step": 5026
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00029791400351791955,
+      "loss": 3.8168,
+      "step": 5027
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0002977243969391055,
+      "loss": 3.7705,
+      "step": 5028
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0002975348251338126,
+      "loss": 3.8959,
+      "step": 5029
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0002973452881346305,
+      "loss": 3.8875,
+      "step": 5030
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00029715578597414266,
+      "loss": 3.8025,
+      "step": 5031
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0002969663186849266,
+      "loss": 3.5915,
+      "step": 5032
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00029677688629955437,
+      "loss": 4.0933,
+      "step": 5033
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00029658748885059126,
+      "loss": 3.9596,
+      "step": 5034
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00029639812637059716,
+      "loss": 3.9851,
+      "step": 5035
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00029620879889212526,
+      "loss": 3.7815,
+      "step": 5036
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0002960195064477235,
+      "loss": 3.8162,
+      "step": 5037
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00029583024906993337,
+      "loss": 3.8888,
+      "step": 5038
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00029564102679129024,
+      "loss": 3.9295,
+      "step": 5039
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00029545183964432386,
+      "loss": 3.8792,
+      "step": 5040
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0002952626876615575,
+      "loss": 3.9424,
+      "step": 5041
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00029507357087550857,
+      "loss": 3.897,
+      "step": 5042
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0002948844893186886,
+      "loss": 3.966,
+      "step": 5043
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0002946954430236028,
+      "loss": 3.8068,
+      "step": 5044
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0002945064320227505,
+      "loss": 3.8906,
+      "step": 5045
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00029431745634862484,
+      "loss": 3.8559,
+      "step": 5046
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0002941285160337128,
+      "loss": 3.891,
+      "step": 5047
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0002939396111104956,
+      "loss": 3.9459,
+      "step": 5048
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0002937507416114481,
+      "loss": 3.7012,
+      "step": 5049
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0002935619075690391,
+      "loss": 3.9478,
+      "step": 5050
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0002933731090157313,
+      "loss": 3.8073,
+      "step": 5051
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0002931843459839817,
+      "loss": 3.8689,
+      "step": 5052
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0002929956185062407,
+      "loss": 3.9909,
+      "step": 5053
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00029280692661495247,
+      "loss": 3.8746,
+      "step": 5054
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0002926182703425557,
+      "loss": 3.7698,
+      "step": 5055
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00029242964972148243,
+      "loss": 3.7207,
+      "step": 5056
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00029224106478415856,
+      "loss": 3.8328,
+      "step": 5057
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0002920525155630044,
+      "loss": 3.7077,
+      "step": 5058
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0002918640020904334,
+      "loss": 3.7663,
+      "step": 5059
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0002916755243988533,
+      "loss": 3.9146,
+      "step": 5060
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00029148708252066586,
+      "loss": 3.8392,
+      "step": 5061
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0002912986764882662,
+      "loss": 3.7715,
+      "step": 5062
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0002911103063340435,
+      "loss": 3.7908,
+      "step": 5063
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0002909219720903808,
+      "loss": 3.8701,
+      "step": 5064
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.000290733673789655,
+      "loss": 3.7469,
+      "step": 5065
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00029054541146423645,
+      "loss": 3.8327,
+      "step": 5066
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00029035718514649014,
+      "loss": 3.7016,
+      "step": 5067
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.000290168994868774,
+      "loss": 3.7478,
+      "step": 5068
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00028998084066344,
+      "loss": 3.7185,
+      "step": 5069
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0002897927225628345,
+      "loss": 3.8216,
+      "step": 5070
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00028960464059929697,
+      "loss": 3.8454,
+      "step": 5071
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0002894165948051608,
+      "loss": 3.8109,
+      "step": 5072
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0002892285852127532,
+      "loss": 3.8036,
+      "step": 5073
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0002890406118543951,
+      "loss": 3.7522,
+      "step": 5074
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00028885267476240175,
+      "loss": 3.8843,
+      "step": 5075
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0002886647739690814,
+      "loss": 3.9255,
+      "step": 5076
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0002884769095067364,
+      "loss": 3.8899,
+      "step": 5077
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00028828908140766284,
+      "loss": 3.7124,
+      "step": 5078
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00028810128970415034,
+      "loss": 3.6806,
+      "step": 5079
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0002879135344284829,
+      "loss": 3.9443,
+      "step": 5080
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00028772581561293745,
+      "loss": 3.8597,
+      "step": 5081
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0002875381332897853,
+      "loss": 3.8707,
+      "step": 5082
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0002873504874912911,
+      "loss": 3.7614,
+      "step": 5083
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00028716287824971345,
+      "loss": 3.9168,
+      "step": 5084
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0002869753055973044,
+      "loss": 3.8489,
+      "step": 5085
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0002867877695663099,
+      "loss": 3.9169,
+      "step": 5086
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00028660027018896937,
+      "loss": 3.8896,
+      "step": 5087
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00028641280749751655,
+      "loss": 3.7452,
+      "step": 5088
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00028622538152417806,
+      "loss": 3.894,
+      "step": 5089
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00028603799230117493,
+      "loss": 3.8316,
+      "step": 5090
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0002858506398607214,
+      "loss": 3.8756,
+      "step": 5091
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0002856633242350255,
+      "loss": 3.8477,
+      "step": 5092
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.000285476045456289,
+      "loss": 3.5724,
+      "step": 5093
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00028528880355670725,
+      "loss": 3.8173,
+      "step": 5094
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0002851015985684691,
+      "loss": 3.9147,
+      "step": 5095
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00028491443052375764,
+      "loss": 3.8814,
+      "step": 5096
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00028472729945474895,
+      "loss": 3.903,
+      "step": 5097
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00028454020539361326,
+      "loss": 3.8068,
+      "step": 5098
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00028435314837251414,
+      "loss": 3.7633,
+      "step": 5099
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0002841661284236088,
+      "loss": 3.9234,
+      "step": 5100
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0002839791455790481,
+      "loss": 3.8321,
+      "step": 5101
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0002837921998709765,
+      "loss": 3.8941,
+      "step": 5102
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00028360529133153234,
+      "loss": 3.8534,
+      "step": 5103
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0002834184199928472,
+      "loss": 3.9569,
+      "step": 5104
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00028323158588704667,
+      "loss": 3.8753,
+      "step": 5105
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0002830447890462492,
+      "loss": 3.7782,
+      "step": 5106
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0002828580295025679,
+      "loss": 3.8435,
+      "step": 5107
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00028267130728810866,
+      "loss": 3.8618,
+      "step": 5108
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0002824846224349712,
+      "loss": 3.8868,
+      "step": 5109
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0002822979749752486,
+      "loss": 3.8619,
+      "step": 5110
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00028211136494102806,
+      "loss": 3.8184,
+      "step": 5111
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00028192479236438983,
+      "loss": 3.9186,
+      "step": 5112
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00028173825727740806,
+      "loss": 3.7107,
+      "step": 5113
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00028155175971215007,
+      "loss": 3.911,
+      "step": 5114
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00028136529970067714,
+      "loss": 3.9079,
+      "step": 5115
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00028117887727504355,
+      "loss": 3.7138,
+      "step": 5116
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0002809924924672978,
+      "loss": 3.8807,
+      "step": 5117
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0002808061453094817,
+      "loss": 3.7814,
+      "step": 5118
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00028061983583363035,
+      "loss": 3.8285,
+      "step": 5119
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00028043356407177254,
+      "loss": 3.8268,
+      "step": 5120
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0002802473300559305,
+      "loss": 3.7442,
+      "step": 5121
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00028006113381812,
+      "loss": 3.8034,
+      "step": 5122
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00027987497539035014,
+      "loss": 4.0038,
+      "step": 5123
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00027968885480462405,
+      "loss": 3.8217,
+      "step": 5124
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00027950277209293776,
+      "loss": 3.9698,
+      "step": 5125
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00027931672728728134,
+      "loss": 3.7409,
+      "step": 5126
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00027913072041963795,
+      "loss": 3.7224,
+      "step": 5127
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0002789447515219842,
+      "loss": 4.0254,
+      "step": 5128
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0002787588206262903,
+      "loss": 3.8645,
+      "step": 5129
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00027857292776452004,
+      "loss": 3.8933,
+      "step": 5130
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00027838707296863033,
+      "loss": 3.8439,
+      "step": 5131
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0002782012562705721,
+      "loss": 3.9163,
+      "step": 5132
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0002780154777022892,
+      "loss": 3.8897,
+      "step": 5133
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.000277829737295719,
+      "loss": 3.863,
+      "step": 5134
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00027764403508279275,
+      "loss": 3.8737,
+      "step": 5135
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0002774583710954346,
+      "loss": 3.7071,
+      "step": 5136
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00027727274536556247,
+      "loss": 3.8254,
+      "step": 5137
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0002770871579250872,
+      "loss": 3.8192,
+      "step": 5138
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00027690160880591396,
+      "loss": 3.8641,
+      "step": 5139
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00027671609803994057,
+      "loss": 4.065,
+      "step": 5140
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0002765306256590584,
+      "loss": 3.8084,
+      "step": 5141
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00027634519169515237,
+      "loss": 3.889,
+      "step": 5142
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0002761597961801006,
+      "loss": 3.9354,
+      "step": 5143
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00027597443914577493,
+      "loss": 3.9488,
+      "step": 5144
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0002757891206240404,
+      "loss": 3.9538,
+      "step": 5145
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00027560384064675503,
+      "loss": 3.8435,
+      "step": 5146
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00027541859924577104,
+      "loss": 3.9337,
+      "step": 5147
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0002752333964529335,
+      "loss": 3.6268,
+      "step": 5148
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0002750482323000808,
+      "loss": 3.7648,
+      "step": 5149
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00027486310681904485,
+      "loss": 3.7414,
+      "step": 5150
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0002746780200416509,
+      "loss": 3.8445,
+      "step": 5151
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00027449297199971714,
+      "loss": 3.8205,
+      "step": 5152
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00027430796272505587,
+      "loss": 3.8887,
+      "step": 5153
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0002741229922494724,
+      "loss": 4.0119,
+      "step": 5154
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00027393806060476523,
+      "loss": 3.6935,
+      "step": 5155
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0002737531678227261,
+      "loss": 3.9362,
+      "step": 5156
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00027356831393514033,
+      "loss": 3.8475,
+      "step": 5157
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00027338349897378637,
+      "loss": 3.7948,
+      "step": 5158
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.000273198722970436,
+      "loss": 3.908,
+      "step": 5159
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0002730139859568546,
+      "loss": 3.8395,
+      "step": 5160
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00027282928796480043,
+      "loss": 3.8573,
+      "step": 5161
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0002726446290260251,
+      "loss": 3.9126,
+      "step": 5162
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.000272460009172274,
+      "loss": 3.9908,
+      "step": 5163
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00027227542843528507,
+      "loss": 3.9228,
+      "step": 5164
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00027209088684679013,
+      "loss": 3.8118,
+      "step": 5165
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0002719063844385138,
+      "loss": 3.7469,
+      "step": 5166
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0002717219212421742,
+      "loss": 3.8546,
+      "step": 5167
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00027153749728948284,
+      "loss": 3.8645,
+      "step": 5168
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0002713531126121442,
+      "loss": 3.8719,
+      "step": 5169
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00027116876724185633,
+      "loss": 3.8713,
+      "step": 5170
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00027098446121031006,
+      "loss": 3.7709,
+      "step": 5171
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00027080019454919,
+      "loss": 3.7967,
+      "step": 5172
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00027061596729017356,
+      "loss": 4.0243,
+      "step": 5173
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00027043177946493147,
+      "loss": 3.7215,
+      "step": 5174
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00027024763110512805,
+      "loss": 3.7221,
+      "step": 5175
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0002700635222424204,
+      "loss": 3.7197,
+      "step": 5176
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00026987945290845883,
+      "loss": 3.8205,
+      "step": 5177
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0002696954231348871,
+      "loss": 3.9324,
+      "step": 5178
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0002695114329533422,
+      "loss": 3.8398,
+      "step": 5179
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0002693274823954537,
+      "loss": 3.8673,
+      "step": 5180
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0002691435714928452,
+      "loss": 3.8374,
+      "step": 5181
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0002689597002771333,
+      "loss": 3.886,
+      "step": 5182
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0002687758687799273,
+      "loss": 3.7921,
+      "step": 5183
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0002685920770328301,
+      "loss": 3.8778,
+      "step": 5184
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0002684083250674376,
+      "loss": 3.6109,
+      "step": 5185
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0002682246129153387,
+      "loss": 3.7473,
+      "step": 5186
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.000268040940608116,
+      "loss": 3.8329,
+      "step": 5187
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00026785730817734446,
+      "loss": 3.8244,
+      "step": 5188
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.000267673715654593,
+      "loss": 3.9379,
+      "step": 5189
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00026749016307142307,
+      "loss": 3.7115,
+      "step": 5190
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0002673066504593897,
+      "loss": 3.9966,
+      "step": 5191
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00026712317785004076,
+      "loss": 3.893,
+      "step": 5192
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00026693974527491725,
+      "loss": 3.7767,
+      "step": 5193
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00026675635276555347,
+      "loss": 3.7934,
+      "step": 5194
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0002665730003534764,
+      "loss": 3.7599,
+      "step": 5195
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0002663896880702069,
+      "loss": 3.8245,
+      "step": 5196
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0002662064159472583,
+      "loss": 3.9077,
+      "step": 5197
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0002660231840161371,
+      "loss": 3.8237,
+      "step": 5198
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00026583999230834294,
+      "loss": 4.0015,
+      "step": 5199
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0002656568408553689,
+      "loss": 3.8553,
+      "step": 5200
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0002654737296887006,
+      "loss": 3.8621,
+      "step": 5201
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00026529065883981707,
+      "loss": 3.9147,
+      "step": 5202
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00026510762834019004,
+      "loss": 3.754,
+      "step": 5203
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.000264924638221285,
+      "loss": 3.6899,
+      "step": 5204
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0002647416885145599,
+      "loss": 3.9218,
+      "step": 5205
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0002645587792514659,
+      "loss": 3.8409,
+      "step": 5206
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0002643759104634472,
+      "loss": 3.8382,
+      "step": 5207
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00026419308218194085,
+      "loss": 3.8651,
+      "step": 5208
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00026401029443837767,
+      "loss": 3.6969,
+      "step": 5209
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00026382754726418043,
+      "loss": 3.8598,
+      "step": 5210
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.000263644840690766,
+      "loss": 3.7847,
+      "step": 5211
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00026346217474954346,
+      "loss": 3.8162,
+      "step": 5212
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00026327954947191534,
+      "loss": 3.9361,
+      "step": 5213
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0002630969648892769,
+      "loss": 3.928,
+      "step": 5214
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00026291442103301666,
+      "loss": 3.8645,
+      "step": 5215
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0002627319179345157,
+      "loss": 3.873,
+      "step": 5216
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00026254945562514877,
+      "loss": 3.7995,
+      "step": 5217
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0002623670341362834,
+      "loss": 3.8516,
+      "step": 5218
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00026218465349927964,
+      "loss": 3.8763,
+      "step": 5219
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.000262002313745491,
+      "loss": 3.8453,
+      "step": 5220
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00026182001490626373,
+      "loss": 3.8233,
+      "step": 5221
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00026163775701293715,
+      "loss": 3.8355,
+      "step": 5222
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0002614555400968434,
+      "loss": 3.7149,
+      "step": 5223
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0002612733641893076,
+      "loss": 3.9964,
+      "step": 5224
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00026109122932164823,
+      "loss": 3.8378,
+      "step": 5225
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00026090913552517624,
+      "loss": 4.0417,
+      "step": 5226
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0002607270828311954,
+      "loss": 3.8171,
+      "step": 5227
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00026054507127100314,
+      "loss": 3.8916,
+      "step": 5228
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0002603631008758891,
+      "loss": 3.7142,
+      "step": 5229
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00026018117167713617,
+      "loss": 3.8161,
+      "step": 5230
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0002599992837060198,
+      "loss": 3.915,
+      "step": 5231
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0002598174369938091,
+      "loss": 3.7815,
+      "step": 5232
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0002596356315717654,
+      "loss": 3.8241,
+      "step": 5233
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0002594538674711432,
+      "loss": 3.7765,
+      "step": 5234
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0002592721447231898,
+      "loss": 3.8714,
+      "step": 5235
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00025909046335914534,
+      "loss": 3.9093,
+      "step": 5236
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0002589088234102432,
+      "loss": 3.9677,
+      "step": 5237
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00025872722490770915,
+      "loss": 3.7791,
+      "step": 5238
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00025854566788276244,
+      "loss": 3.8809,
+      "step": 5239
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0002583641523666146,
+      "loss": 3.6986,
+      "step": 5240
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0002581826783904702,
+      "loss": 3.81,
+      "step": 5241
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00025800124598552676,
+      "loss": 3.9332,
+      "step": 5242
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0002578198551829747,
+      "loss": 3.9414,
+      "step": 5243
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00025763850601399706,
+      "loss": 3.8435,
+      "step": 5244
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00025745719850976976,
+      "loss": 3.8845,
+      "step": 5245
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0002572759327014618,
+      "loss": 3.7628,
+      "step": 5246
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00025709470862023516,
+      "loss": 3.7444,
+      "step": 5247
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00025691352629724406,
+      "loss": 3.8668,
+      "step": 5248
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0002567323857636359,
+      "loss": 3.7722,
+      "step": 5249
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00025655128705055085,
+      "loss": 3.9495,
+      "step": 5250
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00025637023018912185,
+      "loss": 3.8366,
+      "step": 5251
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0002561892152104744,
+      "loss": 3.8018,
+      "step": 5252
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0002560082421457277,
+      "loss": 3.8978,
+      "step": 5253
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0002558273110259925,
+      "loss": 3.7947,
+      "step": 5254
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00025564642188237344,
+      "loss": 3.8422,
+      "step": 5255
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0002554655747459672,
+      "loss": 3.837,
+      "step": 5256
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0002552847696478636,
+      "loss": 3.8867,
+      "step": 5257
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.000255104006619145,
+      "loss": 3.8032,
+      "step": 5258
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00025492328569088664,
+      "loss": 3.9164,
+      "step": 5259
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00025474260689415686,
+      "loss": 3.9025,
+      "step": 5260
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0002545619702600161,
+      "loss": 3.7724,
+      "step": 5261
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0002543813758195181,
+      "loss": 3.8887,
+      "step": 5262
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.000254200823603709,
+      "loss": 3.7737,
+      "step": 5263
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0002540203136436277,
+      "loss": 3.8258,
+      "step": 5264
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00025383984597030635,
+      "loss": 3.8285,
+      "step": 5265
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0002536594206147692,
+      "loss": 3.8961,
+      "step": 5266
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00025347903760803325,
+      "loss": 3.9293,
+      "step": 5267
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00025329869698110884,
+      "loss": 3.9618,
+      "step": 5268
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00025311839876499844,
+      "loss": 3.7031,
+      "step": 5269
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0002529381429906974,
+      "loss": 3.8508,
+      "step": 5270
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00025275792968919373,
+      "loss": 3.9376,
+      "step": 5271
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00025257775889146825,
+      "loss": 3.8228,
+      "step": 5272
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00025239763062849427,
+      "loss": 3.8007,
+      "step": 5273
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00025221754493123794,
+      "loss": 3.8581,
+      "step": 5274
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0002520375018306584,
+      "loss": 3.8316,
+      "step": 5275
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0002518575013577069,
+      "loss": 3.7657,
+      "step": 5276
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0002516775435433276,
+      "loss": 3.83,
+      "step": 5277
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00025149762841845746,
+      "loss": 3.7654,
+      "step": 5278
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00025131775601402583,
+      "loss": 3.8138,
+      "step": 5279
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00025113792636095465,
+      "loss": 3.8709,
+      "step": 5280
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00025095813949015935,
+      "loss": 3.8994,
+      "step": 5281
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0002507783954325467,
+      "loss": 3.8073,
+      "step": 5282
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0002505986942190173,
+      "loss": 3.8847,
+      "step": 5283
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0002504190358804637,
+      "loss": 3.9261,
+      "step": 5284
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00025023942044777114,
+      "loss": 3.8897,
+      "step": 5285
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0002500598479518178,
+      "loss": 4.0131,
+      "step": 5286
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00024988031842347407,
+      "loss": 3.8478,
+      "step": 5287
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0002497008318936031,
+      "loss": 3.7985,
+      "step": 5288
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.000249521388393061,
+      "loss": 3.8942,
+      "step": 5289
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00024934198795269604,
+      "loss": 3.8718,
+      "step": 5290
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00024916263060334904,
+      "loss": 3.913,
+      "step": 5291
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.000248983316375854,
+      "loss": 3.8555,
+      "step": 5292
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0002488040453010369,
+      "loss": 3.9195,
+      "step": 5293
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0002486248174097165,
+      "loss": 3.8387,
+      "step": 5294
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00024844563273270404,
+      "loss": 3.8481,
+      "step": 5295
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0002482664913008037,
+      "loss": 3.8754,
+      "step": 5296
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0002480873931448119,
+      "loss": 4.0118,
+      "step": 5297
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0002479083382955176,
+      "loss": 3.8399,
+      "step": 5298
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0002477293267837025,
+      "loss": 3.9775,
+      "step": 5299
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00024755035864014067,
+      "loss": 3.8398,
+      "step": 5300
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0002473714338955987,
+      "loss": 3.8648,
+      "step": 5301
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0002471925525808361,
+      "loss": 3.7204,
+      "step": 5302
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0002470137147266043,
+      "loss": 3.8578,
+      "step": 5303
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.000246834920363648,
+      "loss": 3.7652,
+      "step": 5304
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00024665616952270386,
+      "loss": 3.8157,
+      "step": 5305
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00024647746223450126,
+      "loss": 3.9081,
+      "step": 5306
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0002462987985297619,
+      "loss": 3.9101,
+      "step": 5307
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0002461201784392002,
+      "loss": 3.8647,
+      "step": 5308
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0002459416019935229,
+      "loss": 3.9363,
+      "step": 5309
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0002457630692234296,
+      "loss": 3.8056,
+      "step": 5310
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0002455845801596121,
+      "loss": 3.9403,
+      "step": 5311
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0002454061348327548,
+      "loss": 3.8552,
+      "step": 5312
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00024522773327353444,
+      "loss": 3.8173,
+      "step": 5313
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00024504937551262024,
+      "loss": 3.8987,
+      "step": 5314
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00024487106158067407,
+      "loss": 3.8543,
+      "step": 5315
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00024469279150834993,
+      "loss": 3.9596,
+      "step": 5316
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00024451456532629493,
+      "loss": 3.8157,
+      "step": 5317
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00024433638306514796,
+      "loss": 3.7195,
+      "step": 5318
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00024415824475554045,
+      "loss": 3.9278,
+      "step": 5319
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0002439801504280968,
+      "loss": 3.7265,
+      "step": 5320
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0002438021001134334,
+      "loss": 3.8627,
+      "step": 5321
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00024362409384215904,
+      "loss": 3.9375,
+      "step": 5322
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00024344613164487512,
+      "loss": 3.9536,
+      "step": 5323
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0002432682135521752,
+      "loss": 3.8317,
+      "step": 5324
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0002430903395946459,
+      "loss": 3.6785,
+      "step": 5325
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0002429125098028655,
+      "loss": 3.9296,
+      "step": 5326
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0002427347242074051,
+      "loss": 3.7665,
+      "step": 5327
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00024255698283882793,
+      "loss": 3.7719,
+      "step": 5328
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00024237928572769013,
+      "loss": 3.8501,
+      "step": 5329
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00024220163290453972,
+      "loss": 3.8646,
+      "step": 5330
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00024202402439991706,
+      "loss": 3.7344,
+      "step": 5331
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00024184646024435547,
+      "loss": 3.9063,
+      "step": 5332
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00024166894046838018,
+      "loss": 3.8871,
+      "step": 5333
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00024149146510250884,
+      "loss": 3.6869,
+      "step": 5334
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00024131403417725156,
+      "loss": 3.8502,
+      "step": 5335
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00024113664772311072,
+      "loss": 3.8186,
+      "step": 5336
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.000240959305770581,
+      "loss": 3.7087,
+      "step": 5337
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0002407820083501498,
+      "loss": 3.9179,
+      "step": 5338
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00024060475549229633,
+      "loss": 3.9963,
+      "step": 5339
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0002404275472274927,
+      "loss": 3.7616,
+      "step": 5340
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.000240250383586203,
+      "loss": 3.8402,
+      "step": 5341
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00024007326459888357,
+      "loss": 3.8614,
+      "step": 5342
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0002398961902959833,
+      "loss": 3.8762,
+      "step": 5343
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00023971916070794337,
+      "loss": 3.7563,
+      "step": 5344
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0002395421758651969,
+      "loss": 3.8336,
+      "step": 5345
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00023936523579817003,
+      "loss": 3.8111,
+      "step": 5346
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0002391883405372805,
+      "loss": 3.9194,
+      "step": 5347
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00023901149011293905,
+      "loss": 3.9362,
+      "step": 5348
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00023883468455554807,
+      "loss": 3.8835,
+      "step": 5349
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0002386579238955025,
+      "loss": 3.837,
+      "step": 5350
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00023848120816318954,
+      "loss": 3.972,
+      "step": 5351
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0002383045373889885,
+      "loss": 3.831,
+      "step": 5352
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00023812791160327153,
+      "loss": 3.8218,
+      "step": 5353
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0002379513308364024,
+      "loss": 3.9088,
+      "step": 5354
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00023777479511873746,
+      "loss": 3.9742,
+      "step": 5355
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00023759830448062498,
+      "loss": 3.9354,
+      "step": 5356
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0002374218589524062,
+      "loss": 3.8227,
+      "step": 5357
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00023724545856441392,
+      "loss": 3.8632,
+      "step": 5358
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00023706910334697347,
+      "loss": 3.8382,
+      "step": 5359
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0002368927933304021,
+      "loss": 3.8944,
+      "step": 5360
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0002367165285450099,
+      "loss": 3.9469,
+      "step": 5361
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00023654030902109873,
+      "loss": 3.7866,
+      "step": 5362
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00023636413478896274,
+      "loss": 3.859,
+      "step": 5363
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00023618800587888827,
+      "loss": 3.7537,
+      "step": 5364
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00023601192232115375,
+      "loss": 3.7222,
+      "step": 5365
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00023583588414603047,
+      "loss": 3.8439,
+      "step": 5366
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0002356598913837809,
+      "loss": 3.8209,
+      "step": 5367
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00023548394406466074,
+      "loss": 3.884,
+      "step": 5368
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00023530804221891705,
+      "loss": 3.7547,
+      "step": 5369
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00023513218587678953,
+      "loss": 3.8126,
+      "step": 5370
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00023495637506850987,
+      "loss": 3.8005,
+      "step": 5371
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00023478060982430195,
+      "loss": 3.7805,
+      "step": 5372
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00023460489017438174,
+      "loss": 3.8308,
+      "step": 5373
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00023442921614895778,
+      "loss": 3.7913,
+      "step": 5374
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0002342535877782302,
+      "loss": 3.752,
+      "step": 5375
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0002340780050923918,
+      "loss": 3.8835,
+      "step": 5376
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00023390246812162718,
+      "loss": 3.9483,
+      "step": 5377
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00023372697689611317,
+      "loss": 3.884,
+      "step": 5378
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0002335515314460187,
+      "loss": 3.966,
+      "step": 5379
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00023337613180150497,
+      "loss": 3.8296,
+      "step": 5380
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00023320077799272493,
+      "loss": 3.8674,
+      "step": 5381
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00023302547004982433,
+      "loss": 3.8593,
+      "step": 5382
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0002328502080029405,
+      "loss": 3.8382,
+      "step": 5383
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00023267499188220277,
+      "loss": 3.765,
+      "step": 5384
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00023249982171773322,
+      "loss": 3.8382,
+      "step": 5385
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0002323246975396454,
+      "loss": 3.8254,
+      "step": 5386
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0002321496193780453,
+      "loss": 3.8292,
+      "step": 5387
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0002319745872630306,
+      "loss": 3.7801,
+      "step": 5388
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00023179960122469173,
+      "loss": 3.7255,
+      "step": 5389
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00023162466129311073,
+      "loss": 3.8349,
+      "step": 5390
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00023144976749836166,
+      "loss": 3.9615,
+      "step": 5391
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00023127491987051085,
+      "loss": 3.8757,
+      "step": 5392
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0002311001184396165,
+      "loss": 3.8807,
+      "step": 5393
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00023092536323572933,
+      "loss": 3.7463,
+      "step": 5394
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0002307506542888916,
+      "loss": 3.9264,
+      "step": 5395
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00023057599162913756,
+      "loss": 3.7773,
+      "step": 5396
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00023040137528649424,
+      "loss": 3.852,
+      "step": 5397
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00023022680529097994,
+      "loss": 3.9372,
+      "step": 5398
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00023005228167260528,
+      "loss": 3.9581,
+      "step": 5399
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00022987780446137292,
+      "loss": 3.903,
+      "step": 5400
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0002297033736872775,
+      "loss": 3.8228,
+      "step": 5401
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00022952898938030552,
+      "loss": 3.9248,
+      "step": 5402
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0002293546515704358,
+      "loss": 3.8434,
+      "step": 5403
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00022918036028763927,
+      "loss": 3.9224,
+      "step": 5404
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0002290061155618784,
+      "loss": 3.8552,
+      "step": 5405
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00022883191742310795,
+      "loss": 3.8613,
+      "step": 5406
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00022865776590127445,
+      "loss": 3.83,
+      "step": 5407
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00022848366102631675,
+      "loss": 3.7435,
+      "step": 5408
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00022830960282816515,
+      "loss": 3.7955,
+      "step": 5409
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00022813559133674278,
+      "loss": 3.844,
+      "step": 5410
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00022796162658196396,
+      "loss": 3.8442,
+      "step": 5411
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00022778770859373503,
+      "loss": 3.6564,
+      "step": 5412
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.000227613837401955,
+      "loss": 3.8128,
+      "step": 5413
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00022744001303651407,
+      "loss": 3.8796,
+      "step": 5414
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0002272662355272947,
+      "loss": 3.8768,
+      "step": 5415
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0002270925049041712,
+      "loss": 3.8395,
+      "step": 5416
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0002269188211970098,
+      "loss": 3.8293,
+      "step": 5417
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00022674518443566906,
+      "loss": 3.8357,
+      "step": 5418
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00022657159464999888,
+      "loss": 3.8604,
+      "step": 5419
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00022639805186984148,
+      "loss": 3.8757,
+      "step": 5420
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00022622455612503062,
+      "loss": 3.7503,
+      "step": 5421
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00022605110744539265,
+      "loss": 3.7755,
+      "step": 5422
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00022587770586074519,
+      "loss": 3.932,
+      "step": 5423
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0002257043514008978,
+      "loss": 3.7407,
+      "step": 5424
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0002255310440956525,
+      "loss": 3.9679,
+      "step": 5425
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00022535778397480262,
+      "loss": 3.8411,
+      "step": 5426
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0002251845710681336,
+      "loss": 3.7258,
+      "step": 5427
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0002250114054054228,
+      "loss": 3.6282,
+      "step": 5428
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00022483828701643933,
+      "loss": 3.9051,
+      "step": 5429
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00022466521593094403,
+      "loss": 3.9003,
+      "step": 5430
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0002244921921786901,
+      "loss": 3.8377,
+      "step": 5431
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0002243192157894225,
+      "loss": 3.7787,
+      "step": 5432
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00022414628679287769,
+      "loss": 3.995,
+      "step": 5433
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00022397340521878412,
+      "loss": 3.6881,
+      "step": 5434
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00022380057109686215,
+      "loss": 3.7673,
+      "step": 5435
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00022362778445682402,
+      "loss": 3.8121,
+      "step": 5436
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00022345504532837368,
+      "loss": 3.852,
+      "step": 5437
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00022328235374120692,
+      "loss": 3.8576,
+      "step": 5438
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0002231097097250115,
+      "loss": 3.7752,
+      "step": 5439
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0002229371133094671,
+      "loss": 3.9559,
+      "step": 5440
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00022276456452424497,
+      "loss": 3.8793,
+      "step": 5441
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00022259206339900818,
+      "loss": 3.9431,
+      "step": 5442
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00022241960996341166,
+      "loss": 3.9397,
+      "step": 5443
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0002222472042471022,
+      "loss": 3.9271,
+      "step": 5444
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00022207484627971813,
+      "loss": 3.9718,
+      "step": 5445
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0002219025360908901,
+      "loss": 3.7444,
+      "step": 5446
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0002217302737102402,
+      "loss": 3.9524,
+      "step": 5447
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00022155805916738213,
+      "loss": 3.8271,
+      "step": 5448
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00022138589249192154,
+      "loss": 3.7038,
+      "step": 5449
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0002212137737134562,
+      "loss": 3.8588,
+      "step": 5450
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00022104170286157504,
+      "loss": 3.8428,
+      "step": 5451
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00022086967996585898,
+      "loss": 3.7439,
+      "step": 5452
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00022069770505588111,
+      "loss": 3.7804,
+      "step": 5453
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00022052577816120567,
+      "loss": 3.815,
+      "step": 5454
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00022035389931138893,
+      "loss": 3.8488,
+      "step": 5455
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00022018206853597879,
+      "loss": 3.8462,
+      "step": 5456
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.000220010285864515,
+      "loss": 3.7734,
+      "step": 5457
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00021983855132652887,
+      "loss": 3.8395,
+      "step": 5458
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00021966686495154375,
+      "loss": 3.8309,
+      "step": 5459
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00021949522676907436,
+      "loss": 3.7454,
+      "step": 5460
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0002193236368086275,
+      "loss": 3.9038,
+      "step": 5461
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0002191520950997014,
+      "loss": 3.8742,
+      "step": 5462
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.000218980601671786,
+      "loss": 3.8273,
+      "step": 5463
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00021880915655436302,
+      "loss": 3.806,
+      "step": 5464
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00021863775977690587,
+      "loss": 3.9005,
+      "step": 5465
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00021846641136887945,
+      "loss": 3.905,
+      "step": 5466
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0002182951113597408,
+      "loss": 3.9092,
+      "step": 5467
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00021812385977893845,
+      "loss": 3.794,
+      "step": 5468
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00021795265665591236,
+      "loss": 3.7762,
+      "step": 5469
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0002177815020200944,
+      "loss": 3.7835,
+      "step": 5470
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00021761039590090798,
+      "loss": 3.8091,
+      "step": 5471
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0002174393383277683,
+      "loss": 3.7797,
+      "step": 5472
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00021726832933008185,
+      "loss": 3.8386,
+      "step": 5473
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00021709736893724745,
+      "loss": 3.9598,
+      "step": 5474
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0002169264571786551,
+      "loss": 3.8215,
+      "step": 5475
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00021675559408368622,
+      "loss": 3.8954,
+      "step": 5476
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00021658477968171453,
+      "loss": 3.8434,
+      "step": 5477
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00021641401400210487,
+      "loss": 3.8318,
+      "step": 5478
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00021624329707421375,
+      "loss": 3.7561,
+      "step": 5479
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00021607262892738954,
+      "loss": 3.8181,
+      "step": 5480
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00021590200959097173,
+      "loss": 3.8051,
+      "step": 5481
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00021573143909429227,
+      "loss": 3.8366,
+      "step": 5482
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0002155609174666739,
+      "loss": 3.8995,
+      "step": 5483
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00021539044473743135,
+      "loss": 3.7512,
+      "step": 5484
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00021522002093587078,
+      "loss": 3.8508,
+      "step": 5485
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0002150496460912899,
+      "loss": 3.7847,
+      "step": 5486
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00021487932023297852,
+      "loss": 3.8512,
+      "step": 5487
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0002147090433902172,
+      "loss": 3.8536,
+      "step": 5488
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00021453881559227884,
+      "loss": 3.6886,
+      "step": 5489
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0002143686368684274,
+      "loss": 3.8763,
+      "step": 5490
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00021419850724791866,
+      "loss": 3.7767,
+      "step": 5491
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0002140284267599998,
+      "loss": 3.8362,
+      "step": 5492
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00021385839543390967,
+      "loss": 3.8099,
+      "step": 5493
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0002136884132988784,
+      "loss": 3.8733,
+      "step": 5494
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0002135184803841283,
+      "loss": 3.8667,
+      "step": 5495
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00021334859671887234,
+      "loss": 4.0149,
+      "step": 5496
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00021317876233231605,
+      "loss": 3.9466,
+      "step": 5497
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00021300897725365554,
+      "loss": 3.884,
+      "step": 5498
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00021283924151207896,
+      "loss": 3.7432,
+      "step": 5499
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00021266955513676584,
+      "loss": 3.6874,
+      "step": 5500
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0002124999181568872,
+      "loss": 3.865,
+      "step": 5501
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00021233033060160539,
+      "loss": 3.7964,
+      "step": 5502
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00021216079250007492,
+      "loss": 3.7791,
+      "step": 5503
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00021199130388144095,
+      "loss": 3.9302,
+      "step": 5504
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00021182186477484093,
+      "loss": 3.7966,
+      "step": 5505
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00021165247520940319,
+      "loss": 3.8904,
+      "step": 5506
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00021148313521424778,
+      "loss": 3.8004,
+      "step": 5507
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00021131384481848614,
+      "loss": 3.9091,
+      "step": 5508
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0002111446040512212,
+      "loss": 3.8738,
+      "step": 5509
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0002109754129415477,
+      "loss": 3.8035,
+      "step": 5510
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00021080627151855136,
+      "loss": 3.8026,
+      "step": 5511
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00021063717981130954,
+      "loss": 3.7637,
+      "step": 5512
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00021046813784889097,
+      "loss": 3.929,
+      "step": 5513
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00021029914566035617,
+      "loss": 3.6744,
+      "step": 5514
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0002101302032747568,
+      "loss": 3.8434,
+      "step": 5515
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00020996131072113588,
+      "loss": 3.915,
+      "step": 5516
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00020979246802852792,
+      "loss": 3.8398,
+      "step": 5517
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00020962367522595922,
+      "loss": 3.866,
+      "step": 5518
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00020945493234244712,
+      "loss": 3.7535,
+      "step": 5519
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00020928623940700042,
+      "loss": 3.5612,
+      "step": 5520
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00020911759644861937,
+      "loss": 3.7627,
+      "step": 5521
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00020894900349629575,
+      "loss": 3.8547,
+      "step": 5522
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0002087804605790124,
+      "loss": 3.8187,
+      "step": 5523
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00020861196772574398,
+      "loss": 3.7135,
+      "step": 5524
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0002084435249654565,
+      "loss": 3.9714,
+      "step": 5525
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00020827513232710715,
+      "loss": 3.8468,
+      "step": 5526
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00020810678983964448,
+      "loss": 3.715,
+      "step": 5527
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00020793849753200855,
+      "loss": 3.938,
+      "step": 5528
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00020777025543313078,
+      "loss": 3.7788,
+      "step": 5529
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00020760206357193372,
+      "loss": 3.9151,
+      "step": 5530
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00020743392197733191,
+      "loss": 3.8979,
+      "step": 5531
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00020726583067823045,
+      "loss": 3.8183,
+      "step": 5532
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00020709778970352656,
+      "loss": 3.828,
+      "step": 5533
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0002069297990821082,
+      "loss": 3.8299,
+      "step": 5534
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00020676185884285493,
+      "loss": 4.02,
+      "step": 5535
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0002065939690146376,
+      "loss": 3.7987,
+      "step": 5536
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00020642612962631853,
+      "loss": 3.7657,
+      "step": 5537
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00020625834070675091,
+      "loss": 3.8947,
+      "step": 5538
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00020609060228478015,
+      "loss": 3.7851,
+      "step": 5539
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0002059229143892421,
+      "loss": 3.7679,
+      "step": 5540
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00020575527704896413,
+      "loss": 3.8704,
+      "step": 5541
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00020558769029276535,
+      "loss": 3.7648,
+      "step": 5542
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00020542015414945575,
+      "loss": 3.7993,
+      "step": 5543
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00020525266864783676,
+      "loss": 3.7493,
+      "step": 5544
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0002050852338167008,
+      "loss": 3.8426,
+      "step": 5545
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0002049178496848323,
+      "loss": 3.8731,
+      "step": 5546
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00020475051628100633,
+      "loss": 3.8244,
+      "step": 5547
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0002045832336339894,
+      "loss": 3.7257,
+      "step": 5548
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00020441600177253938,
+      "loss": 3.6803,
+      "step": 5549
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00020424882072540542,
+      "loss": 3.8748,
+      "step": 5550
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0002040816905213276,
+      "loss": 3.8049,
+      "step": 5551
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00020391461118903786,
+      "loss": 3.7428,
+      "step": 5552
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0002037475827572589,
+      "loss": 3.8392,
+      "step": 5553
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00020358060525470508,
+      "loss": 3.9099,
+      "step": 5554
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00020341367871008154,
+      "loss": 3.8202,
+      "step": 5555
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00020324680315208504,
+      "loss": 3.777,
+      "step": 5556
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0002030799786094033,
+      "loss": 3.7376,
+      "step": 5557
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00020291320511071543,
+      "loss": 3.8777,
+      "step": 5558
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00020274648268469155,
+      "loss": 3.9364,
+      "step": 5559
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0002025798113599933,
+      "loss": 3.7935,
+      "step": 5560
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00020241319116527374,
+      "loss": 3.8692,
+      "step": 5561
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0002022466221291765,
+      "loss": 3.8504,
+      "step": 5562
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00020208010428033674,
+      "loss": 3.829,
+      "step": 5563
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00020191363764738085,
+      "loss": 3.9817,
+      "step": 5564
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00020174722225892638,
+      "loss": 3.8463,
+      "step": 5565
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00020158085814358183,
+      "loss": 3.8024,
+      "step": 5566
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0002014145453299476,
+      "loss": 3.7655,
+      "step": 5567
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0002012482838466145,
+      "loss": 3.8109,
+      "step": 5568
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00020108207372216457,
+      "loss": 3.9256,
+      "step": 5569
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0002009159149851718,
+      "loss": 3.7887,
+      "step": 5570
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00020074980766420048,
+      "loss": 3.9797,
+      "step": 5571
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0002005837517878064,
+      "loss": 3.9168,
+      "step": 5572
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0002004177473845366,
+      "loss": 3.8411,
+      "step": 5573
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00020025179448292886,
+      "loss": 3.9485,
+      "step": 5574
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00020008589311151277,
+      "loss": 3.8908,
+      "step": 5575
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00019992004329880857,
+      "loss": 3.9837,
+      "step": 5576
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0001997542450733278,
+      "loss": 3.8375,
+      "step": 5577
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00019958849846357285,
+      "loss": 3.855,
+      "step": 5578
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00019942280349803781,
+      "loss": 3.7723,
+      "step": 5579
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00019925716020520752,
+      "loss": 3.686,
+      "step": 5580
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00019909156861355764,
+      "loss": 3.7764,
+      "step": 5581
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0001989260287515558,
+      "loss": 3.9626,
+      "step": 5582
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00019876054064765992,
+      "loss": 3.9025,
+      "step": 5583
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0001985951043303194,
+      "loss": 3.8779,
+      "step": 5584
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0001984297198279746,
+      "loss": 3.8973,
+      "step": 5585
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0001982643871690571,
+      "loss": 3.895,
+      "step": 5586
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00019809910638198924,
+      "loss": 3.565,
+      "step": 5587
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00019793387749518515,
+      "loss": 3.8477,
+      "step": 5588
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00019776870053704915,
+      "loss": 3.8279,
+      "step": 5589
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00019760357553597752,
+      "loss": 3.8897,
+      "step": 5590
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0001974385025203569,
+      "loss": 3.8083,
+      "step": 5591
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00019727348151856534,
+      "loss": 3.7266,
+      "step": 5592
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00019710851255897173,
+      "loss": 3.9486,
+      "step": 5593
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0001969435956699363,
+      "loss": 3.8065,
+      "step": 5594
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0001967787308798099,
+      "loss": 3.9659,
+      "step": 5595
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00019661391821693507,
+      "loss": 3.8214,
+      "step": 5596
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00019644915770964473,
+      "loss": 3.8604,
+      "step": 5597
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00019628444938626333,
+      "loss": 3.7668,
+      "step": 5598
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00019611979327510616,
+      "loss": 3.8697,
+      "step": 5599
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00019595518940447931,
+      "loss": 3.8524,
+      "step": 5600
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00019579063780268024,
+      "loss": 3.8096,
+      "step": 5601
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00019562613849799705,
+      "loss": 3.8694,
+      "step": 5602
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00019546169151870942,
+      "loss": 3.7601,
+      "step": 5603
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00019529729689308757,
+      "loss": 3.8821,
+      "step": 5604
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00019513295464939273,
+      "loss": 3.6976,
+      "step": 5605
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00019496866481587716,
+      "loss": 3.9876,
+      "step": 5606
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00019480442742078453,
+      "loss": 3.8518,
+      "step": 5607
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00019464024249234897,
+      "loss": 3.8401,
+      "step": 5608
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00019447611005879572,
+      "loss": 3.9539,
+      "step": 5609
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00019431203014834092,
+      "loss": 3.8242,
+      "step": 5610
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0001941480027891922,
+      "loss": 3.7265,
+      "step": 5611
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00019398402800954745,
+      "loss": 3.7114,
+      "step": 5612
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00019382010583759602,
+      "loss": 3.7943,
+      "step": 5613
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0001936562363015179,
+      "loss": 3.8157,
+      "step": 5614
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00019349241942948403,
+      "loss": 3.6241,
+      "step": 5615
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00019332865524965677,
+      "loss": 3.8118,
+      "step": 5616
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00019316494379018872,
+      "loss": 3.7558,
+      "step": 5617
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00019300128507922416,
+      "loss": 3.9393,
+      "step": 5618
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00019283767914489775,
+      "loss": 3.8734,
+      "step": 5619
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00019267412601533518,
+      "loss": 3.9134,
+      "step": 5620
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0001925106257186532,
+      "loss": 3.9377,
+      "step": 5621
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00019234717828295938,
+      "loss": 3.7489,
+      "step": 5622
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0001921837837363521,
+      "loss": 3.8273,
+      "step": 5623
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.000192020442106921,
+      "loss": 3.9291,
+      "step": 5624
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0001918571534227462,
+      "loss": 3.9015,
+      "step": 5625
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00019169391771189914,
+      "loss": 3.8336,
+      "step": 5626
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0001915307350024419,
+      "loss": 3.702,
+      "step": 5627
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00019136760532242731,
+      "loss": 3.7843,
+      "step": 5628
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0001912045286998994,
+      "loss": 3.8739,
+      "step": 5629
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0001910415051628928,
+      "loss": 3.9156,
+      "step": 5630
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0001908785347394331,
+      "loss": 3.8251,
+      "step": 5631
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00019071561745753713,
+      "loss": 3.9364,
+      "step": 5632
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.000190552753345212,
+      "loss": 3.7782,
+      "step": 5633
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00019038994243045581,
+      "loss": 3.9136,
+      "step": 5634
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.000190227184741258,
+      "loss": 3.9272,
+      "step": 5635
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00019006448030559831,
+      "loss": 3.7983,
+      "step": 5636
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0001899018291514476,
+      "loss": 3.7721,
+      "step": 5637
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0001897392313067672,
+      "loss": 3.851,
+      "step": 5638
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00018957668679950995,
+      "loss": 3.8452,
+      "step": 5639
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00018941419565761904,
+      "loss": 3.8319,
+      "step": 5640
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0001892517579090285,
+      "loss": 3.8586,
+      "step": 5641
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00018908937358166322,
+      "loss": 3.8628,
+      "step": 5642
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00018892704270343885,
+      "loss": 3.6341,
+      "step": 5643
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00018876476530226233,
+      "loss": 3.7849,
+      "step": 5644
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0001886025414060306,
+      "loss": 3.9403,
+      "step": 5645
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00018844037104263223,
+      "loss": 3.9234,
+      "step": 5646
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00018827825423994594,
+      "loss": 3.9589,
+      "step": 5647
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00018811619102584155,
+      "loss": 3.6831,
+      "step": 5648
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0001879541814281796,
+      "loss": 3.9389,
+      "step": 5649
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0001877922254748114,
+      "loss": 3.8745,
+      "step": 5650
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0001876303231935791,
+      "loss": 3.7889,
+      "step": 5651
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0001874684746123153,
+      "loss": 3.9192,
+      "step": 5652
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00018730667975884398,
+      "loss": 3.7223,
+      "step": 5653
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00018714493866097955,
+      "loss": 3.9821,
+      "step": 5654
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0001869832513465271,
+      "loss": 3.7866,
+      "step": 5655
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0001868216178432826,
+      "loss": 3.9031,
+      "step": 5656
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00018666003817903267,
+      "loss": 3.853,
+      "step": 5657
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00018649851238155464,
+      "loss": 4.1056,
+      "step": 5658
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00018633704047861665,
+      "loss": 3.6367,
+      "step": 5659
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00018617562249797787,
+      "loss": 3.7609,
+      "step": 5660
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00018601425846738774,
+      "loss": 3.8113,
+      "step": 5661
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0001858529484145864,
+      "loss": 3.6511,
+      "step": 5662
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00018569169236730532,
+      "loss": 3.9756,
+      "step": 5663
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00018553049035326612,
+      "loss": 3.7703,
+      "step": 5664
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00018536934240018127,
+      "loss": 3.8957,
+      "step": 5665
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0001852082485357538,
+      "loss": 3.8478,
+      "step": 5666
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00018504720878767793,
+      "loss": 3.8145,
+      "step": 5667
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00018488622318363812,
+      "loss": 3.7335,
+      "step": 5668
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0001847252917513097,
+      "loss": 3.9549,
+      "step": 5669
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0001845644145183586,
+      "loss": 3.7618,
+      "step": 5670
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0001844035915124413,
+      "loss": 3.9983,
+      "step": 5671
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00018424282276120545,
+      "loss": 3.9981,
+      "step": 5672
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.000184082108292289,
+      "loss": 3.9219,
+      "step": 5673
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0001839214481333204,
+      "loss": 3.7891,
+      "step": 5674
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0001837608423119193,
+      "loss": 3.8497,
+      "step": 5675
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00018360029085569558,
+      "loss": 3.879,
+      "step": 5676
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0001834397937922499,
+      "loss": 3.8242,
+      "step": 5677
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0001832793511491735,
+      "loss": 3.6714,
+      "step": 5678
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00018311896295404839,
+      "loss": 3.7524,
+      "step": 5679
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.000182958629234447,
+      "loss": 3.9432,
+      "step": 5680
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0001827983500179327,
+      "loss": 3.8013,
+      "step": 5681
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00018263812533205953,
+      "loss": 3.8673,
+      "step": 5682
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00018247795520437177,
+      "loss": 3.8942,
+      "step": 5683
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00018231783966240456,
+      "loss": 3.8258,
+      "step": 5684
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00018215777873368361,
+      "loss": 3.8406,
+      "step": 5685
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00018199777244572523,
+      "loss": 3.8116,
+      "step": 5686
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00018183782082603617,
+      "loss": 3.8231,
+      "step": 5687
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00018167792390211436,
+      "loss": 3.7515,
+      "step": 5688
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0001815180817014475,
+      "loss": 3.7779,
+      "step": 5689
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0001813582942515148,
+      "loss": 3.8212,
+      "step": 5690
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00018119856157978531,
+      "loss": 3.9889,
+      "step": 5691
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00018103888371371897,
+      "loss": 3.7532,
+      "step": 5692
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0001808792606807662,
+      "loss": 3.7558,
+      "step": 5693
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00018071969250836812,
+      "loss": 3.8852,
+      "step": 5694
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00018056017922395606,
+      "loss": 3.8839,
+      "step": 5695
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00018040072085495274,
+      "loss": 3.8285,
+      "step": 5696
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00018024131742877054,
+      "loss": 3.9383,
+      "step": 5697
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00018008196897281286,
+      "loss": 3.7197,
+      "step": 5698
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00017992267551447333,
+      "loss": 3.9447,
+      "step": 5699
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00017976343708113674,
+      "loss": 3.798,
+      "step": 5700
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0001796042537001778,
+      "loss": 3.7976,
+      "step": 5701
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0001794451253989618,
+      "loss": 3.8678,
+      "step": 5702
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00017928605220484512,
+      "loss": 3.794,
+      "step": 5703
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0001791270341451741,
+      "loss": 3.7317,
+      "step": 5704
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00017896807124728582,
+      "loss": 3.8902,
+      "step": 5705
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00017880916353850778,
+      "loss": 3.9311,
+      "step": 5706
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00017865031104615807,
+      "loss": 3.8332,
+      "step": 5707
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0001784915137975452,
+      "loss": 3.7487,
+      "step": 5708
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0001783327718199685,
+      "loss": 3.9544,
+      "step": 5709
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0001781740851407172,
+      "loss": 3.8807,
+      "step": 5710
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00017801545378707178,
+      "loss": 3.8218,
+      "step": 5711
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00017785687778630266,
+      "loss": 3.872,
+      "step": 5712
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0001776983571656708,
+      "loss": 3.9426,
+      "step": 5713
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00017753989195242782,
+      "loss": 3.8496,
+      "step": 5714
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00017738148217381566,
+      "loss": 3.8035,
+      "step": 5715
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00017722312785706668,
+      "loss": 3.871,
+      "step": 5716
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00017706482902940396,
+      "loss": 3.7569,
+      "step": 5717
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00017690658571804107,
+      "loss": 3.7073,
+      "step": 5718
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00017674839795018171,
+      "loss": 3.8153,
+      "step": 5719
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00017659026575302022,
+      "loss": 3.8325,
+      "step": 5720
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00017643218915374138,
+      "loss": 3.8344,
+      "step": 5721
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0001762741681795203,
+      "loss": 4.1084,
+      "step": 5722
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00017611620285752245,
+      "loss": 3.6923,
+      "step": 5723
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00017595829321490435,
+      "loss": 3.7691,
+      "step": 5724
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00017580043927881224,
+      "loss": 3.9269,
+      "step": 5725
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0001756426410763829,
+      "loss": 3.9376,
+      "step": 5726
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00017548489863474393,
+      "loss": 3.7479,
+      "step": 5727
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0001753272119810131,
+      "loss": 4.0143,
+      "step": 5728
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00017516958114229835,
+      "loss": 3.7596,
+      "step": 5729
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00017501200614569845,
+      "loss": 3.9097,
+      "step": 5730
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00017485448701830205,
+      "loss": 3.9232,
+      "step": 5731
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00017469702378718893,
+      "loss": 3.9126,
+      "step": 5732
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00017453961647942868,
+      "loss": 3.8693,
+      "step": 5733
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00017438226512208138,
+      "loss": 3.859,
+      "step": 5734
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0001742249697421976,
+      "loss": 3.8361,
+      "step": 5735
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00017406773036681805,
+      "loss": 3.92,
+      "step": 5736
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00017391054702297438,
+      "loss": 3.9246,
+      "step": 5737
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00017375341973768787,
+      "loss": 3.8501,
+      "step": 5738
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0001735963485379708,
+      "loss": 3.9095,
+      "step": 5739
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00017343933345082547,
+      "loss": 3.9182,
+      "step": 5740
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00017328237450324452,
+      "loss": 3.9058,
+      "step": 5741
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00017312547172221098,
+      "loss": 4.0271,
+      "step": 5742
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00017296862513469836,
+      "loss": 3.7784,
+      "step": 5743
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00017281183476767015,
+      "loss": 3.8832,
+      "step": 5744
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00017265510064808082,
+      "loss": 3.8996,
+      "step": 5745
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00017249842280287443,
+      "loss": 3.6952,
+      "step": 5746
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00017234180125898606,
+      "loss": 3.8772,
+      "step": 5747
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00017218523604334057,
+      "loss": 3.8827,
+      "step": 5748
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0001720287271828534,
+      "loss": 3.8325,
+      "step": 5749
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00017187227470443013,
+      "loss": 3.8782,
+      "step": 5750
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00017171587863496686,
+      "loss": 3.8059,
+      "step": 5751
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00017155953900134969,
+      "loss": 3.8614,
+      "step": 5752
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00017140325583045552,
+      "loss": 3.9458,
+      "step": 5753
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00017124702914915097,
+      "loss": 3.8663,
+      "step": 5754
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00017109085898429343,
+      "loss": 3.8292,
+      "step": 5755
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00017093474536273036,
+      "loss": 3.9115,
+      "step": 5756
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00017077868831129933,
+      "loss": 3.8973,
+      "step": 5757
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00017062268785682848,
+      "loss": 4.0014,
+      "step": 5758
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00017046674402613593,
+      "loss": 3.8506,
+      "step": 5759
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00017031085684603048,
+      "loss": 3.7083,
+      "step": 5760
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00017015502634331082,
+      "loss": 3.9479,
+      "step": 5761
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00016999925254476605,
+      "loss": 3.8093,
+      "step": 5762
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0001698435354771753,
+      "loss": 3.8147,
+      "step": 5763
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00016968787516730843,
+      "loss": 3.8564,
+      "step": 5764
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00016953227164192514,
+      "loss": 3.8615,
+      "step": 5765
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00016937672492777544,
+      "loss": 3.918,
+      "step": 5766
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00016922123505159952,
+      "loss": 3.8614,
+      "step": 5767
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00016906580204012817,
+      "loss": 3.7585,
+      "step": 5768
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00016891042592008198,
+      "loss": 3.7884,
+      "step": 5769
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00016875510671817185,
+      "loss": 3.7576,
+      "step": 5770
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00016859984446109905,
+      "loss": 3.8792,
+      "step": 5771
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00016844463917555487,
+      "loss": 3.892,
+      "step": 5772
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00016828949088822077,
+      "loss": 3.8966,
+      "step": 5773
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00016813439962576872,
+      "loss": 3.8956,
+      "step": 5774
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00016797936541486082,
+      "loss": 3.8655,
+      "step": 5775
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00016782438828214912,
+      "loss": 3.7674,
+      "step": 5776
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00016766946825427605,
+      "loss": 3.8908,
+      "step": 5777
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00016751460535787405,
+      "loss": 3.6682,
+      "step": 5778
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00016735979961956588,
+      "loss": 3.847,
+      "step": 5779
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00016720505106596428,
+      "loss": 3.7462,
+      "step": 5780
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00016705035972367272,
+      "loss": 3.848,
+      "step": 5781
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00016689572561928395,
+      "loss": 3.8149,
+      "step": 5782
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00016674114877938183,
+      "loss": 3.8463,
+      "step": 5783
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00016658662923053975,
+      "loss": 3.94,
+      "step": 5784
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0001664321669993213,
+      "loss": 3.8087,
+      "step": 5785
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0001662777621122804,
+      "loss": 3.7924,
+      "step": 5786
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0001661234145959611,
+      "loss": 3.8291,
+      "step": 5787
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0001659691244768972,
+      "loss": 3.7517,
+      "step": 5788
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00016581489178161347,
+      "loss": 3.8541,
+      "step": 5789
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00016566071653662402,
+      "loss": 3.7991,
+      "step": 5790
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00016550659876843316,
+      "loss": 3.9061,
+      "step": 5791
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.000165352538503536,
+      "loss": 3.7663,
+      "step": 5792
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.000165198535768417,
+      "loss": 3.7512,
+      "step": 5793
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00016504459058955107,
+      "loss": 3.6666,
+      "step": 5794
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00016489070299340297,
+      "loss": 3.7238,
+      "step": 5795
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00016473687300642815,
+      "loss": 3.8122,
+      "step": 5796
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00016458310065507154,
+      "loss": 3.7878,
+      "step": 5797
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00016442938596576839,
+      "loss": 3.9438,
+      "step": 5798
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0001642757289649441,
+      "loss": 3.8967,
+      "step": 5799
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00016412212967901386,
+      "loss": 3.916,
+      "step": 5800
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00016396858813438355,
+      "loss": 3.8278,
+      "step": 5801
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0001638151043574485,
+      "loss": 3.9885,
+      "step": 5802
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00016366167837459427,
+      "loss": 3.9455,
+      "step": 5803
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00016350831021219687,
+      "loss": 3.7844,
+      "step": 5804
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00016335499989662185,
+      "loss": 3.8156,
+      "step": 5805
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0001632017474542251,
+      "loss": 3.8509,
+      "step": 5806
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00016304855291135245,
+      "loss": 3.7955,
+      "step": 5807
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00016289541629433985,
+      "loss": 3.6956,
+      "step": 5808
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00016274233762951306,
+      "loss": 3.893,
+      "step": 5809
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00016258931694318829,
+      "loss": 3.8643,
+      "step": 5810
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0001624363542616717,
+      "loss": 3.9213,
+      "step": 5811
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00016228344961125913,
+      "loss": 3.8849,
+      "step": 5812
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00016213060301823674,
+      "loss": 3.7976,
+      "step": 5813
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00016197781450888067,
+      "loss": 3.7799,
+      "step": 5814
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0001618250841094569,
+      "loss": 3.7216,
+      "step": 5815
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0001616724118462215,
+      "loss": 3.8802,
+      "step": 5816
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00016151979774542085,
+      "loss": 3.8573,
+      "step": 5817
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00016136724183329104,
+      "loss": 3.7897,
+      "step": 5818
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0001612147441360579,
+      "loss": 3.7668,
+      "step": 5819
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.000161062304679938,
+      "loss": 3.9242,
+      "step": 5820
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0001609099234911372,
+      "loss": 3.7887,
+      "step": 5821
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00016075760059585165,
+      "loss": 3.8832,
+      "step": 5822
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00016060533602026733,
+      "loss": 3.8999,
+      "step": 5823
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00016045312979056026,
+      "loss": 3.8551,
+      "step": 5824
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00016030098193289667,
+      "loss": 3.9357,
+      "step": 5825
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00016014889247343238,
+      "loss": 3.8649,
+      "step": 5826
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00015999686143831343,
+      "loss": 3.7184,
+      "step": 5827
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00015984488885367542,
+      "loss": 3.7911,
+      "step": 5828
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00015969297474564465,
+      "loss": 3.8287,
+      "step": 5829
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0001595411191403367,
+      "loss": 3.8041,
+      "step": 5830
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00015938932206385715,
+      "loss": 3.7383,
+      "step": 5831
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00015923758354230194,
+      "loss": 3.819,
+      "step": 5832
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00015908590360175662,
+      "loss": 3.8114,
+      "step": 5833
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00015893428226829671,
+      "loss": 3.8628,
+      "step": 5834
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00015878271956798763,
+      "loss": 3.9582,
+      "step": 5835
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00015863121552688476,
+      "loss": 3.8973,
+      "step": 5836
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00015847977017103326,
+      "loss": 3.9666,
+      "step": 5837
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00015832838352646856,
+      "loss": 3.8456,
+      "step": 5838
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00015817705561921587,
+      "loss": 3.8823,
+      "step": 5839
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00015802578647529004,
+      "loss": 3.7975,
+      "step": 5840
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00015787457612069606,
+      "loss": 3.6642,
+      "step": 5841
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00015772342458142868,
+      "loss": 3.8265,
+      "step": 5842
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00015757233188347264,
+      "loss": 3.8175,
+      "step": 5843
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00015742129805280247,
+      "loss": 3.7901,
+      "step": 5844
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00015727032311538258,
+      "loss": 3.8016,
+      "step": 5845
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00015711940709716755,
+      "loss": 3.8246,
+      "step": 5846
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00015696855002410128,
+      "loss": 3.7615,
+      "step": 5847
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00015681775192211817,
+      "loss": 3.7433,
+      "step": 5848
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00015666701281714202,
+      "loss": 3.876,
+      "step": 5849
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00015651633273508665,
+      "loss": 3.9719,
+      "step": 5850
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00015636571170185565,
+      "loss": 3.7724,
+      "step": 5851
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00015621514974334245,
+      "loss": 3.8185,
+      "step": 5852
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00015606464688543064,
+      "loss": 3.8353,
+      "step": 5853
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00015591420315399325,
+      "loss": 3.7561,
+      "step": 5854
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00015576381857489337,
+      "loss": 3.9464,
+      "step": 5855
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0001556134931739836,
+      "loss": 3.975,
+      "step": 5856
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00015546322697710698,
+      "loss": 3.9145,
+      "step": 5857
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0001553130200100959,
+      "loss": 3.7244,
+      "step": 5858
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00015516287229877242,
+      "loss": 3.9217,
+      "step": 5859
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00015501278386894908,
+      "loss": 3.9835,
+      "step": 5860
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00015486275474642762,
+      "loss": 3.6933,
+      "step": 5861
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0001547127849569998,
+      "loss": 3.8533,
+      "step": 5862
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00015456287452644718,
+      "loss": 3.8583,
+      "step": 5863
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00015441302348054103,
+      "loss": 3.7761,
+      "step": 5864
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00015426323184504243,
+      "loss": 3.7287,
+      "step": 5865
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00015411349964570254,
+      "loss": 3.7662,
+      "step": 5866
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00015396382690826172,
+      "loss": 3.75,
+      "step": 5867
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00015381421365845082,
+      "loss": 3.9311,
+      "step": 5868
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00015366465992198996,
+      "loss": 3.7347,
+      "step": 5869
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00015351516572458912,
+      "loss": 3.7609,
+      "step": 5870
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00015336573109194808,
+      "loss": 3.8098,
+      "step": 5871
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00015321635604975637,
+      "loss": 3.8479,
+      "step": 5872
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0001530670406236932,
+      "loss": 3.9118,
+      "step": 5873
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0001529177848394277,
+      "loss": 3.7738,
+      "step": 5874
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0001527685887226189,
+      "loss": 3.8303,
+      "step": 5875
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00015261945229891511,
+      "loss": 3.8866,
+      "step": 5876
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00015247037559395466,
+      "loss": 3.9707,
+      "step": 5877
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00015232135863336554,
+      "loss": 3.7557,
+      "step": 5878
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00015217240144276556,
+      "loss": 3.8611,
+      "step": 5879
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00015202350404776193,
+      "loss": 3.897,
+      "step": 5880
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00015187466647395226,
+      "loss": 3.8111,
+      "step": 5881
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00015172588874692316,
+      "loss": 3.7883,
+      "step": 5882
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00015157717089225143,
+      "loss": 3.6859,
+      "step": 5883
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00015142851293550302,
+      "loss": 3.8528,
+      "step": 5884
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00015127991490223447,
+      "loss": 3.889,
+      "step": 5885
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0001511313768179912,
+      "loss": 3.7892,
+      "step": 5886
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00015098289870830867,
+      "loss": 3.8653,
+      "step": 5887
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00015083448059871192,
+      "loss": 3.7846,
+      "step": 5888
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00015068612251471593,
+      "loss": 3.7218,
+      "step": 5889
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00015053782448182508,
+      "loss": 3.8267,
+      "step": 5890
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00015038958652553353,
+      "loss": 3.9475,
+      "step": 5891
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00015024140867132507,
+      "loss": 3.8661,
+      "step": 5892
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0001500932909446731,
+      "loss": 3.7545,
+      "step": 5893
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00014994523337104099,
+      "loss": 3.7668,
+      "step": 5894
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0001497972359758813,
+      "loss": 3.8185,
+      "step": 5895
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00014964929878463683,
+      "loss": 3.8792,
+      "step": 5896
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00014950142182273946,
+      "loss": 3.8247,
+      "step": 5897
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0001493536051156111,
+      "loss": 3.6246,
+      "step": 5898
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00014920584868866293,
+      "loss": 3.7307,
+      "step": 5899
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0001490581525672962,
+      "loss": 3.8841,
+      "step": 5900
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00014891051677690153,
+      "loss": 3.8214,
+      "step": 5901
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.000148762941342859,
+      "loss": 3.694,
+      "step": 5902
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0001486154262905388,
+      "loss": 3.956,
+      "step": 5903
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00014846797164530051,
+      "loss": 3.8698,
+      "step": 5904
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00014832057743249328,
+      "loss": 3.7505,
+      "step": 5905
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00014817324367745572,
+      "loss": 3.8586,
+      "step": 5906
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00014802597040551634,
+      "loss": 3.8284,
+      "step": 5907
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0001478787576419931,
+      "loss": 3.9437,
+      "step": 5908
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00014773160541219337,
+      "loss": 3.7692,
+      "step": 5909
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00014758451374141467,
+      "loss": 3.8905,
+      "step": 5910
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0001474374826549435,
+      "loss": 3.8728,
+      "step": 5911
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00014729051217805646,
+      "loss": 3.7611,
+      "step": 5912
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0001471436023360193,
+      "loss": 3.8513,
+      "step": 5913
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00014699675315408756,
+      "loss": 3.9291,
+      "step": 5914
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00014684996465750638,
+      "loss": 3.9336,
+      "step": 5915
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0001467032368715101,
+      "loss": 3.7791,
+      "step": 5916
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00014655656982132337,
+      "loss": 3.8846,
+      "step": 5917
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0001464099635321598,
+      "loss": 3.7435,
+      "step": 5918
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0001462634180292226,
+      "loss": 3.7412,
+      "step": 5919
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0001461169333377047,
+      "loss": 3.9309,
+      "step": 5920
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00014597050948278838,
+      "loss": 3.8688,
+      "step": 5921
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00014582414648964593,
+      "loss": 3.889,
+      "step": 5922
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00014567784438343867,
+      "loss": 3.7694,
+      "step": 5923
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0001455316031893174,
+      "loss": 3.7561,
+      "step": 5924
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00014538542293242307,
+      "loss": 3.8529,
+      "step": 5925
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0001452393036378856,
+      "loss": 3.878,
+      "step": 5926
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00014509324533082459,
+      "loss": 3.8386,
+      "step": 5927
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00014494724803634912,
+      "loss": 3.8908,
+      "step": 5928
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0001448013117795578,
+      "loss": 3.8804,
+      "step": 5929
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0001446554365855387,
+      "loss": 3.8082,
+      "step": 5930
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0001445096224793695,
+      "loss": 3.8823,
+      "step": 5931
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00014436386948611763,
+      "loss": 3.881,
+      "step": 5932
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0001442181776308394,
+      "loss": 3.8348,
+      "step": 5933
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00014407254693858107,
+      "loss": 3.8729,
+      "step": 5934
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00014392697743437815,
+      "loss": 3.8747,
+      "step": 5935
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0001437814691432558,
+      "loss": 3.7202,
+      "step": 5936
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00014363602209022835,
+      "loss": 3.9427,
+      "step": 5937
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00014349063630030018,
+      "loss": 3.8104,
+      "step": 5938
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00014334531179846456,
+      "loss": 3.7735,
+      "step": 5939
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0001432000486097046,
+      "loss": 3.8289,
+      "step": 5940
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00014305484675899272,
+      "loss": 3.7584,
+      "step": 5941
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00014290970627129073,
+      "loss": 3.8202,
+      "step": 5942
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00014276462717154997,
+      "loss": 3.751,
+      "step": 5943
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0001426196094847112,
+      "loss": 3.8006,
+      "step": 5944
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0001424746532357046,
+      "loss": 3.8656,
+      "step": 5945
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00014232975844944996,
+      "loss": 3.8076,
+      "step": 5946
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00014218492515085635,
+      "loss": 3.9079,
+      "step": 5947
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00014204015336482213,
+      "loss": 3.8906,
+      "step": 5948
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00014189544311623547,
+      "loss": 3.8792,
+      "step": 5949
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00014175079442997369,
+      "loss": 3.7537,
+      "step": 5950
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00014160620733090352,
+      "loss": 3.8206,
+      "step": 5951
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00014146168184388098,
+      "loss": 3.9028,
+      "step": 5952
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00014131721799375208,
+      "loss": 3.9565,
+      "step": 5953
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0001411728158053516,
+      "loss": 3.7933,
+      "step": 5954
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00014102847530350388,
+      "loss": 3.7828,
+      "step": 5955
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00014088419651302287,
+      "loss": 3.896,
+      "step": 5956
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0001407399794587117,
+      "loss": 3.8954,
+      "step": 5957
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0001405958241653628,
+      "loss": 3.9834,
+      "step": 5958
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0001404517306577585,
+      "loss": 4.0012,
+      "step": 5959
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00014030769896066975,
+      "loss": 3.7044,
+      "step": 5960
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0001401637290988576,
+      "loss": 3.8922,
+      "step": 5961
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.000140019821097072,
+      "loss": 3.6333,
+      "step": 5962
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00013987597498005245,
+      "loss": 3.9663,
+      "step": 5963
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00013973219077252768,
+      "loss": 3.7855,
+      "step": 5964
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0001395884684992159,
+      "loss": 3.7973,
+      "step": 5965
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00013944480818482447,
+      "loss": 3.8358,
+      "step": 5966
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0001393012098540505,
+      "loss": 3.8729,
+      "step": 5967
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0001391576735315802,
+      "loss": 3.8438,
+      "step": 5968
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00013901419924208908,
+      "loss": 3.8626,
+      "step": 5969
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00013887078701024203,
+      "loss": 3.8223,
+      "step": 5970
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00013872743686069327,
+      "loss": 3.8694,
+      "step": 5971
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00013858414881808635,
+      "loss": 3.8482,
+      "step": 5972
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00013844092290705396,
+      "loss": 3.775,
+      "step": 5973
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0001382977591522186,
+      "loss": 3.851,
+      "step": 5974
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00013815465757819175,
+      "loss": 3.9652,
+      "step": 5975
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00013801161820957387,
+      "loss": 3.7999,
+      "step": 5976
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0001378686410709556,
+      "loss": 3.7256,
+      "step": 5977
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00013772572618691602,
+      "loss": 3.7867,
+      "step": 5978
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00013758287358202405,
+      "loss": 3.9078,
+      "step": 5979
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00013744008328083756,
+      "loss": 3.8239,
+      "step": 5980
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00013729735530790377,
+      "loss": 3.8743,
+      "step": 5981
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00013715468968775952,
+      "loss": 3.831,
+      "step": 5982
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00013701208644493064,
+      "loss": 3.7782,
+      "step": 5983
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00013686954560393217,
+      "loss": 3.7729,
+      "step": 5984
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00013672706718926848,
+      "loss": 3.935,
+      "step": 5985
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00013658465122543345,
+      "loss": 3.8556,
+      "step": 5986
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00013644229773690996,
+      "loss": 3.6722,
+      "step": 5987
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00013630000674817012,
+      "loss": 3.8808,
+      "step": 5988
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0001361577782836756,
+      "loss": 3.999,
+      "step": 5989
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00013601561236787702,
+      "loss": 3.8352,
+      "step": 5990
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00013587350902521435,
+      "loss": 3.853,
+      "step": 5991
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0001357314682801168,
+      "loss": 3.9244,
+      "step": 5992
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00013558949015700277,
+      "loss": 3.8639,
+      "step": 5993
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00013544757468027984,
+      "loss": 3.7625,
+      "step": 5994
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0001353057218743453,
+      "loss": 3.935,
+      "step": 5995
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00013516393176358487,
+      "loss": 3.8033,
+      "step": 5996
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00013502220437237427,
+      "loss": 3.7208,
+      "step": 5997
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0001348805397250779,
+      "loss": 3.7653,
+      "step": 5998
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00013473893784604962,
+      "loss": 3.7891,
+      "step": 5999
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00013459739875963238,
+      "loss": 3.9052,
+      "step": 6000
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0001344559224901584,
+      "loss": 3.741,
+      "step": 6001
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00013431450906194891,
+      "loss": 3.8791,
+      "step": 6002
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00013417315849931493,
+      "loss": 3.8071,
+      "step": 6003
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00013403187082655584,
+      "loss": 3.8437,
+      "step": 6004
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00013389064606796096,
+      "loss": 3.9436,
+      "step": 6005
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00013374948424780837,
+      "loss": 3.8689,
+      "step": 6006
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0001336083853903653,
+      "loss": 3.9415,
+      "step": 6007
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00013346734951988842,
+      "loss": 3.837,
+      "step": 6008
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0001333263766606232,
+      "loss": 3.9403,
+      "step": 6009
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00013318546683680483,
+      "loss": 3.824,
+      "step": 6010
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00013304462007265717,
+      "loss": 3.9641,
+      "step": 6011
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00013290383639239344,
+      "loss": 3.7974,
+      "step": 6012
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00013276311582021582,
+      "loss": 3.7669,
+      "step": 6013
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00013262245838031617,
+      "loss": 3.7395,
+      "step": 6014
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0001324818640968749,
+      "loss": 3.7608,
+      "step": 6015
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0001323413329940618,
+      "loss": 3.8691,
+      "step": 6016
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00013220086509603567,
+      "loss": 3.8805,
+      "step": 6017
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00013206046042694492,
+      "loss": 3.7473,
+      "step": 6018
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00013192011901092655,
+      "loss": 3.8496,
+      "step": 6019
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0001317798408721068,
+      "loss": 3.7617,
+      "step": 6020
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0001316396260346012,
+      "loss": 3.7914,
+      "step": 6021
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0001314994745225142,
+      "loss": 3.7575,
+      "step": 6022
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00013135938635993965,
+      "loss": 3.7157,
+      "step": 6023
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0001312193615709601,
+      "loss": 3.9184,
+      "step": 6024
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0001310794001796477,
+      "loss": 3.7667,
+      "step": 6025
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00013093950221006328,
+      "loss": 3.6996,
+      "step": 6026
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.000130799667686257,
+      "loss": 3.8488,
+      "step": 6027
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00013065989663226795,
+      "loss": 3.9081,
+      "step": 6028
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00013052018907212448,
+      "loss": 3.76,
+      "step": 6029
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0001303805450298437,
+      "loss": 3.8729,
+      "step": 6030
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00013024096452943234,
+      "loss": 3.8188,
+      "step": 6031
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00013010144759488567,
+      "loss": 3.8121,
+      "step": 6032
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00012996199425018855,
+      "loss": 3.9123,
+      "step": 6033
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00012982260451931444,
+      "loss": 3.8415,
+      "step": 6034
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00012968327842622612,
+      "loss": 3.719,
+      "step": 6035
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0001295440159948753,
+      "loss": 3.8316,
+      "step": 6036
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00012940481724920284,
+      "loss": 3.7635,
+      "step": 6037
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0001292656822131384,
+      "loss": 3.7871,
+      "step": 6038
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0001291266109106013,
+      "loss": 3.938,
+      "step": 6039
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0001289876033654993,
+      "loss": 3.7565,
+      "step": 6040
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0001288486596017293,
+      "loss": 3.8855,
+      "step": 6041
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0001287097796431776,
+      "loss": 3.7318,
+      "step": 6042
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00012857096351371917,
+      "loss": 3.8946,
+      "step": 6043
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00012843221123721803,
+      "loss": 3.8078,
+      "step": 6044
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00012829352283752726,
+      "loss": 3.8371,
+      "step": 6045
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00012815489833848927,
+      "loss": 3.8667,
+      "step": 6046
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00012801633776393501,
+      "loss": 3.7342,
+      "step": 6047
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0001278778411376847,
+      "loss": 3.7791,
+      "step": 6048
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00012773940848354753,
+      "loss": 3.8582,
+      "step": 6049
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0001276010398253215,
+      "loss": 3.8991,
+      "step": 6050
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00012746273518679407,
+      "loss": 3.9111,
+      "step": 6051
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00012732449459174134,
+      "loss": 3.6223,
+      "step": 6052
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00012718631806392822,
+      "loss": 3.8091,
+      "step": 6053
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00012704820562710922,
+      "loss": 3.8127,
+      "step": 6054
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0001269101573050273,
+      "loss": 3.7417,
+      "step": 6055
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00012677217312141454,
+      "loss": 3.7454,
+      "step": 6056
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00012663425309999204,
+      "loss": 3.7311,
+      "step": 6057
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00012649639726446994,
+      "loss": 3.815,
+      "step": 6058
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00012635860563854696,
+      "loss": 3.9674,
+      "step": 6059
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00012622087824591127,
+      "loss": 3.7913,
+      "step": 6060
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00012608321511024006,
+      "loss": 3.8998,
+      "step": 6061
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0001259456162551989,
+      "loss": 3.7961,
+      "step": 6062
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0001258080817044428,
+      "loss": 3.8194,
+      "step": 6063
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0001256706114816154,
+      "loss": 3.8517,
+      "step": 6064
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00012553320561034954,
+      "loss": 3.7183,
+      "step": 6065
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00012539586411426662,
+      "loss": 3.7968,
+      "step": 6066
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00012525858701697761,
+      "loss": 3.7946,
+      "step": 6067
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00012512137434208187,
+      "loss": 3.8751,
+      "step": 6068
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00012498422611316768,
+      "loss": 3.6197,
+      "step": 6069
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00012484714235381277,
+      "loss": 3.7408,
+      "step": 6070
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00012471012308758322,
+      "loss": 3.7742,
+      "step": 6071
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00012457316833803421,
+      "loss": 3.7633,
+      "step": 6072
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00012443627812870988,
+      "loss": 3.7422,
+      "step": 6073
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00012429945248314318,
+      "loss": 3.9036,
+      "step": 6074
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00012416269142485615,
+      "loss": 3.6479,
+      "step": 6075
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0001240259949773596,
+      "loss": 3.7225,
+      "step": 6076
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00012388936316415317,
+      "loss": 3.8312,
+      "step": 6077
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0001237527960087253,
+      "loss": 3.8367,
+      "step": 6078
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0001236162935345538,
+      "loss": 3.7716,
+      "step": 6079
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0001234798557651048,
+      "loss": 3.9124,
+      "step": 6080
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0001233434827238335,
+      "loss": 3.8477,
+      "step": 6081
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00012320717443418423,
+      "loss": 3.7373,
+      "step": 6082
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0001230709309195898,
+      "loss": 3.8235,
+      "step": 6083
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00012293475220347212,
+      "loss": 3.7966,
+      "step": 6084
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0001227986383092418,
+      "loss": 3.7325,
+      "step": 6085
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0001226625892602985,
+      "loss": 3.8909,
+      "step": 6086
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00012252660508003043,
+      "loss": 3.8116,
+      "step": 6087
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00012239068579181494,
+      "loss": 3.6475,
+      "step": 6088
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00012225483141901837,
+      "loss": 3.8187,
+      "step": 6089
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00012211904198499552,
+      "loss": 3.7938,
+      "step": 6090
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00012198331751309005,
+      "loss": 3.8772,
+      "step": 6091
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00012184765802663466,
+      "loss": 3.7213,
+      "step": 6092
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00012171206354895081,
+      "loss": 3.8662,
+      "step": 6093
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00012157653410334874,
+      "loss": 3.7843,
+      "step": 6094
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0001214410697131273,
+      "loss": 3.855,
+      "step": 6095
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00012130567040157465,
+      "loss": 3.8113,
+      "step": 6096
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0001211703361919676,
+      "loss": 3.912,
+      "step": 6097
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00012103506710757156,
+      "loss": 3.8356,
+      "step": 6098
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00012089986317164075,
+      "loss": 3.7728,
+      "step": 6099
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00012076472440741842,
+      "loss": 3.8554,
+      "step": 6100
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00012062965083813642,
+      "loss": 3.9632,
+      "step": 6101
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00012049464248701536,
+      "loss": 3.7775,
+      "step": 6102
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00012035969937726498,
+      "loss": 3.8348,
+      "step": 6103
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00012022482153208347,
+      "loss": 3.762,
+      "step": 6104
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00012009000897465783,
+      "loss": 3.8236,
+      "step": 6105
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00011995526172816384,
+      "loss": 3.7395,
+      "step": 6106
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00011982057981576632,
+      "loss": 3.8891,
+      "step": 6107
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00011968596326061858,
+      "loss": 3.786,
+      "step": 6108
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00011955141208586257,
+      "loss": 3.9105,
+      "step": 6109
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00011941692631462953,
+      "loss": 3.948,
+      "step": 6110
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00011928250597003892,
+      "loss": 3.8166,
+      "step": 6111
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00011914815107519922,
+      "loss": 3.7442,
+      "step": 6112
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00011901386165320755,
+      "loss": 3.8901,
+      "step": 6113
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00011887963772714983,
+      "loss": 3.6409,
+      "step": 6114
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00011874547932010055,
+      "loss": 3.8052,
+      "step": 6115
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00011861138645512343,
+      "loss": 3.6848,
+      "step": 6116
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00011847735915527025,
+      "loss": 3.8099,
+      "step": 6117
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00011834339744358218,
+      "loss": 3.878,
+      "step": 6118
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0001182095013430886,
+      "loss": 3.8425,
+      "step": 6119
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00011807567087680787,
+      "loss": 3.7569,
+      "step": 6120
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00011794190606774697,
+      "loss": 3.7262,
+      "step": 6121
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0001178082069389016,
+      "loss": 3.787,
+      "step": 6122
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00011767457351325605,
+      "loss": 3.72,
+      "step": 6123
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00011754100581378364,
+      "loss": 3.7522,
+      "step": 6124
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00011740750386344628,
+      "loss": 3.8568,
+      "step": 6125
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00011727406768519445,
+      "loss": 3.8773,
+      "step": 6126
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0001171406973019673,
+      "loss": 3.6756,
+      "step": 6127
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00011700739273669275,
+      "loss": 3.7326,
+      "step": 6128
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0001168741540122874,
+      "loss": 3.7517,
+      "step": 6129
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00011674098115165643,
+      "loss": 3.8205,
+      "step": 6130
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00011660787417769409,
+      "loss": 3.7142,
+      "step": 6131
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00011647483311328283,
+      "loss": 3.9357,
+      "step": 6132
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00011634185798129382,
+      "loss": 3.9159,
+      "step": 6133
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00011620894880458732,
+      "loss": 3.8015,
+      "step": 6134
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00011607610560601179,
+      "loss": 3.8023,
+      "step": 6135
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00011594332840840455,
+      "loss": 3.8072,
+      "step": 6136
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00011581061723459153,
+      "loss": 3.8216,
+      "step": 6137
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00011567797210738712,
+      "loss": 3.7814,
+      "step": 6138
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00011554539304959494,
+      "loss": 3.8124,
+      "step": 6139
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00011541288008400663,
+      "loss": 3.7181,
+      "step": 6140
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00011528043323340281,
+      "loss": 3.6641,
+      "step": 6141
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0001151480525205525,
+      "loss": 3.8813,
+      "step": 6142
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00011501573796821346,
+      "loss": 3.705,
+      "step": 6143
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0001148834895991323,
+      "loss": 3.9279,
+      "step": 6144
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00011475130743604379,
+      "loss": 3.7438,
+      "step": 6145
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00011461919150167188,
+      "loss": 3.8413,
+      "step": 6146
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00011448714181872865,
+      "loss": 3.6654,
+      "step": 6147
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.000114355158409915,
+      "loss": 3.745,
+      "step": 6148
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00011422324129792039,
+      "loss": 3.7964,
+      "step": 6149
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00011409139050542295,
+      "loss": 3.717,
+      "step": 6150
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00011395960605508915,
+      "loss": 3.8164,
+      "step": 6151
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00011382788796957454,
+      "loss": 3.7541,
+      "step": 6152
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00011369623627152276,
+      "loss": 3.9293,
+      "step": 6153
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00011356465098356656,
+      "loss": 3.9107,
+      "step": 6154
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00011343313212832674,
+      "loss": 3.779,
+      "step": 6155
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00011330167972841299,
+      "loss": 3.9014,
+      "step": 6156
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00011317029380642352,
+      "loss": 3.8575,
+      "step": 6157
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00011303897438494504,
+      "loss": 3.6042,
+      "step": 6158
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00011290772148655276,
+      "loss": 3.8541,
+      "step": 6159
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0001127765351338108,
+      "loss": 3.7741,
+      "step": 6160
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00011264541534927147,
+      "loss": 3.9057,
+      "step": 6161
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00011251436215547595,
+      "loss": 3.7836,
+      "step": 6162
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0001123833755749537,
+      "loss": 3.8208,
+      "step": 6163
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00011225245563022285,
+      "loss": 3.8534,
+      "step": 6164
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00011212160234378999,
+      "loss": 3.8073,
+      "step": 6165
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00011199081573815024,
+      "loss": 3.949,
+      "step": 6166
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0001118600958357876,
+      "loss": 3.8381,
+      "step": 6167
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00011172944265917417,
+      "loss": 3.7358,
+      "step": 6168
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00011159885623077076,
+      "loss": 3.85,
+      "step": 6169
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00011146833657302658,
+      "loss": 3.9465,
+      "step": 6170
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00011133788370837972,
+      "loss": 3.8784,
+      "step": 6171
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00011120749765925642,
+      "loss": 3.8505,
+      "step": 6172
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00011107717844807153,
+      "loss": 3.8071,
+      "step": 6173
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00011094692609722829,
+      "loss": 3.901,
+      "step": 6174
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0001108167406291189,
+      "loss": 3.8827,
+      "step": 6175
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00011068662206612362,
+      "loss": 3.9482,
+      "step": 6176
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00011055657043061124,
+      "loss": 3.8279,
+      "step": 6177
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0001104265857449393,
+      "loss": 3.7283,
+      "step": 6178
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00011029666803145355,
+      "loss": 3.7424,
+      "step": 6179
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00011016681731248829,
+      "loss": 3.7469,
+      "step": 6180
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00011003703361036643,
+      "loss": 3.8771,
+      "step": 6181
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00010990731694739947,
+      "loss": 4.0069,
+      "step": 6182
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00010977766734588706,
+      "loss": 3.9257,
+      "step": 6183
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0001096480848281175,
+      "loss": 3.8299,
+      "step": 6184
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00010951856941636751,
+      "loss": 3.8265,
+      "step": 6185
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00010938912113290229,
+      "loss": 3.8399,
+      "step": 6186
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00010925973999997535,
+      "loss": 3.8009,
+      "step": 6187
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00010913042603982909,
+      "loss": 3.8297,
+      "step": 6188
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00010900117927469383,
+      "loss": 4.0049,
+      "step": 6189
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0001088719997267888,
+      "loss": 3.989,
+      "step": 6190
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00010874288741832139,
+      "loss": 3.8768,
+      "step": 6191
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00010861384237148748,
+      "loss": 3.7972,
+      "step": 6192
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00010848486460847134,
+      "loss": 3.7925,
+      "step": 6193
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00010835595415144588,
+      "loss": 3.7274,
+      "step": 6194
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00010822711102257204,
+      "loss": 3.8321,
+      "step": 6195
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00010809833524399982,
+      "loss": 3.8252,
+      "step": 6196
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00010796962683786709,
+      "loss": 3.9545,
+      "step": 6197
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00010784098582630015,
+      "loss": 3.7278,
+      "step": 6198
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00010771241223141415,
+      "loss": 3.6559,
+      "step": 6199
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00010758390607531232,
+      "loss": 3.7372,
+      "step": 6200
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00010745546738008627,
+      "loss": 3.6907,
+      "step": 6201
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00010732709616781606,
+      "loss": 3.844,
+      "step": 6202
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0001071987924605704,
+      "loss": 3.7689,
+      "step": 6203
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00010707055628040602,
+      "loss": 3.8423,
+      "step": 6204
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00010694238764936826,
+      "loss": 3.7854,
+      "step": 6205
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00010681428658949082,
+      "loss": 3.6677,
+      "step": 6206
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00010668625312279551,
+      "loss": 3.7594,
+      "step": 6207
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00010655828727129318,
+      "loss": 3.9397,
+      "step": 6208
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00010643038905698237,
+      "loss": 3.763,
+      "step": 6209
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00010630255850185022,
+      "loss": 3.7824,
+      "step": 6210
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00010617479562787258,
+      "loss": 3.7436,
+      "step": 6211
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0001060471004570131,
+      "loss": 3.8298,
+      "step": 6212
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00010591947301122424,
+      "loss": 3.9193,
+      "step": 6213
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00010579191331244648,
+      "loss": 3.9425,
+      "step": 6214
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00010566442138260884,
+      "loss": 3.8596,
+      "step": 6215
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00010553699724362865,
+      "loss": 3.8746,
+      "step": 6216
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00010540964091741157,
+      "loss": 3.7383,
+      "step": 6217
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00010528235242585187,
+      "loss": 3.7039,
+      "step": 6218
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00010515513179083175,
+      "loss": 3.7711,
+      "step": 6219
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0001050279790342219,
+      "loss": 3.9544,
+      "step": 6220
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00010490089417788134,
+      "loss": 3.7551,
+      "step": 6221
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0001047738772436575,
+      "loss": 3.6855,
+      "step": 6222
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00010464692825338585,
+      "loss": 3.9116,
+      "step": 6223
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00010452004722889069,
+      "loss": 3.9814,
+      "step": 6224
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00010439323419198421,
+      "loss": 3.7759,
+      "step": 6225
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0001042664891644669,
+      "loss": 3.8304,
+      "step": 6226
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00010413981216812801,
+      "loss": 3.8581,
+      "step": 6227
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00010401320322474456,
+      "loss": 3.7358,
+      "step": 6228
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00010388666235608219,
+      "loss": 3.8196,
+      "step": 6229
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00010376018958389472,
+      "loss": 3.8961,
+      "step": 6230
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00010363378492992409,
+      "loss": 3.8186,
+      "step": 6231
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00010350744841590103,
+      "loss": 3.872,
+      "step": 6232
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00010338118006354413,
+      "loss": 3.775,
+      "step": 6233
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0001032549798945604,
+      "loss": 3.9123,
+      "step": 6234
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00010312884793064492,
+      "loss": 3.7948,
+      "step": 6235
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00010300278419348158,
+      "loss": 3.9231,
+      "step": 6236
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00010287678870474199,
+      "loss": 3.9149,
+      "step": 6237
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00010275086148608614,
+      "loss": 4.0144,
+      "step": 6238
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00010262500255916268,
+      "loss": 3.8324,
+      "step": 6239
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00010249921194560802,
+      "loss": 3.8818,
+      "step": 6240
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00010237348966704707,
+      "loss": 3.9045,
+      "step": 6241
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0001022478357450929,
+      "loss": 3.8772,
+      "step": 6242
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00010212225020134691,
+      "loss": 3.691,
+      "step": 6243
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00010199673305739853,
+      "loss": 3.8489,
+      "step": 6244
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.000101871284334826,
+      "loss": 4.0923,
+      "step": 6245
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00010174590405519501,
+      "loss": 3.9493,
+      "step": 6246
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00010162059224006026,
+      "loss": 3.7984,
+      "step": 6247
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00010149534891096408,
+      "loss": 3.9177,
+      "step": 6248
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00010137017408943727,
+      "loss": 3.8604,
+      "step": 6249
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00010124506779699882,
+      "loss": 3.7349,
+      "step": 6250
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00010112003005515602,
+      "loss": 3.8543,
+      "step": 6251
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00010099506088540417,
+      "loss": 3.7709,
+      "step": 6252
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00010087016030922708,
+      "loss": 3.8471,
+      "step": 6253
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00010074532834809646,
+      "loss": 3.7382,
+      "step": 6254
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00010062056502347256,
+      "loss": 3.6302,
+      "step": 6255
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00010049587035680352,
+      "loss": 3.7873,
+      "step": 6256
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00010037124436952588,
+      "loss": 3.837,
+      "step": 6257
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00010024668708306417,
+      "loss": 3.8266,
+      "step": 6258
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0001001221985188312,
+      "loss": 3.9363,
+      "step": 6259
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.999777869822818e-05,
+      "loss": 3.8142,
+      "step": 6260
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.987342764264423e-05,
+      "loss": 3.8467,
+      "step": 6261
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.974914537345675e-05,
+      "loss": 3.8989,
+      "step": 6262
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.96249319120311e-05,
+      "loss": 3.7782,
+      "step": 6263
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.950078727972139e-05,
+      "loss": 3.8284,
+      "step": 6264
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.937671149786931e-05,
+      "loss": 3.8396,
+      "step": 6265
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.925270458780495e-05,
+      "loss": 3.9022,
+      "step": 6266
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.912876657084641e-05,
+      "loss": 3.8539,
+      "step": 6267
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.900489746830032e-05,
+      "loss": 3.7141,
+      "step": 6268
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.888109730146111e-05,
+      "loss": 3.8565,
+      "step": 6269
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.87573660916114e-05,
+      "loss": 3.8422,
+      "step": 6270
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.863370386002212e-05,
+      "loss": 3.7894,
+      "step": 6271
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.851011062795201e-05,
+      "loss": 3.8527,
+      "step": 6272
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.838658641664849e-05,
+      "loss": 3.8741,
+      "step": 6273
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.826313124734653e-05,
+      "loss": 3.9249,
+      "step": 6274
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.813974514126978e-05,
+      "loss": 3.943,
+      "step": 6275
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.801642811962964e-05,
+      "loss": 3.8178,
+      "step": 6276
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.789318020362564e-05,
+      "loss": 3.661,
+      "step": 6277
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.777000141444564e-05,
+      "loss": 3.908,
+      "step": 6278
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.764689177326541e-05,
+      "loss": 3.8172,
+      "step": 6279
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.75238513012488e-05,
+      "loss": 3.8006,
+      "step": 6280
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.740088001954811e-05,
+      "loss": 3.6883,
+      "step": 6281
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.72779779493036e-05,
+      "loss": 3.6521,
+      "step": 6282
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.715514511164342e-05,
+      "loss": 3.6242,
+      "step": 6283
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.703238152768401e-05,
+      "loss": 3.8945,
+      "step": 6284
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.690968721852978e-05,
+      "loss": 3.8272,
+      "step": 6285
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.678706220527333e-05,
+      "loss": 3.8817,
+      "step": 6286
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.666450650899533e-05,
+      "loss": 3.7564,
+      "step": 6287
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.654202015076441e-05,
+      "loss": 3.6886,
+      "step": 6288
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.641960315163767e-05,
+      "loss": 3.6744,
+      "step": 6289
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.629725553265983e-05,
+      "loss": 3.9837,
+      "step": 6290
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.617497731486374e-05,
+      "loss": 3.8438,
+      "step": 6291
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.605276851927075e-05,
+      "loss": 3.7282,
+      "step": 6292
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.593062916688982e-05,
+      "loss": 3.8099,
+      "step": 6293
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.580855927871807e-05,
+      "loss": 3.784,
+      "step": 6294
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.568655887574068e-05,
+      "loss": 3.8211,
+      "step": 6295
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.556462797893111e-05,
+      "loss": 3.8739,
+      "step": 6296
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.544276660925066e-05,
+      "loss": 3.6493,
+      "step": 6297
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.532097478764862e-05,
+      "loss": 3.8423,
+      "step": 6298
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.519925253506245e-05,
+      "loss": 3.7885,
+      "step": 6299
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.507759987241754e-05,
+      "loss": 3.8954,
+      "step": 6300
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.495601682062755e-05,
+      "loss": 3.7451,
+      "step": 6301
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.483450340059385e-05,
+      "loss": 3.8214,
+      "step": 6302
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.47130596332062e-05,
+      "loss": 3.8522,
+      "step": 6303
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.459168553934211e-05,
+      "loss": 3.8809,
+      "step": 6304
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.447038113986717e-05,
+      "loss": 3.8629,
+      "step": 6305
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.434914645563497e-05,
+      "loss": 3.8153,
+      "step": 6306
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.422798150748724e-05,
+      "loss": 3.7226,
+      "step": 6307
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.410688631625364e-05,
+      "loss": 3.7491,
+      "step": 6308
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.398586090275164e-05,
+      "loss": 3.8802,
+      "step": 6309
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.386490528778702e-05,
+      "loss": 3.9208,
+      "step": 6310
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.374401949215366e-05,
+      "loss": 3.9582,
+      "step": 6311
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.362320353663311e-05,
+      "loss": 3.8927,
+      "step": 6312
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.350245744199498e-05,
+      "loss": 3.7227,
+      "step": 6313
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.338178122899693e-05,
+      "loss": 3.668,
+      "step": 6314
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.32611749183846e-05,
+      "loss": 3.9554,
+      "step": 6315
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.31406385308915e-05,
+      "loss": 3.8812,
+      "step": 6316
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.30201720872395e-05,
+      "loss": 3.9025,
+      "step": 6317
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.289977560813789e-05,
+      "loss": 3.9206,
+      "step": 6318
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.277944911428449e-05,
+      "loss": 3.8334,
+      "step": 6319
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.265919262636468e-05,
+      "loss": 3.7356,
+      "step": 6320
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.2539006165052e-05,
+      "loss": 3.7648,
+      "step": 6321
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.241888975100782e-05,
+      "loss": 3.8878,
+      "step": 6322
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.229884340488149e-05,
+      "loss": 3.7249,
+      "step": 6323
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.217886714731055e-05,
+      "loss": 3.7996,
+      "step": 6324
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.205896099892019e-05,
+      "loss": 3.8595,
+      "step": 6325
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.193912498032375e-05,
+      "loss": 3.7379,
+      "step": 6326
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.18193591121223e-05,
+      "loss": 3.6783,
+      "step": 6327
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.169966341490499e-05,
+      "loss": 3.6736,
+      "step": 6328
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.158003790924906e-05,
+      "loss": 3.7526,
+      "step": 6329
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.146048261571944e-05,
+      "loss": 3.8,
+      "step": 6330
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.134099755486891e-05,
+      "loss": 3.7404,
+      "step": 6331
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.122158274723863e-05,
+      "loss": 3.8538,
+      "step": 6332
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.110223821335723e-05,
+      "loss": 3.9872,
+      "step": 6333
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.098296397374145e-05,
+      "loss": 3.6362,
+      "step": 6334
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.08637600488959e-05,
+      "loss": 3.8014,
+      "step": 6335
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.074462645931308e-05,
+      "loss": 3.8441,
+      "step": 6336
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.062556322547333e-05,
+      "loss": 3.7748,
+      "step": 6337
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.050657036784515e-05,
+      "loss": 3.8938,
+      "step": 6338
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.038764790688492e-05,
+      "loss": 3.8473,
+      "step": 6339
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.026879586303666e-05,
+      "loss": 3.8939,
+      "step": 6340
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.015001425673241e-05,
+      "loss": 3.7301,
+      "step": 6341
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.003130310839203e-05,
+      "loss": 3.8662,
+      "step": 6342
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.99126624384235e-05,
+      "loss": 3.7654,
+      "step": 6343
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.979409226722224e-05,
+      "loss": 3.9325,
+      "step": 6344
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.967559261517217e-05,
+      "loss": 3.899,
+      "step": 6345
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.955716350264454e-05,
+      "loss": 3.6333,
+      "step": 6346
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.943880494999885e-05,
+      "loss": 3.8019,
+      "step": 6347
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.932051697758225e-05,
+      "loss": 3.835,
+      "step": 6348
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.920229960572973e-05,
+      "loss": 3.974,
+      "step": 6349
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.908415285476434e-05,
+      "loss": 3.7904,
+      "step": 6350
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.896607674499679e-05,
+      "loss": 3.9161,
+      "step": 6351
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.884807129672567e-05,
+      "loss": 3.8073,
+      "step": 6352
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.873013653023765e-05,
+      "loss": 3.8738,
+      "step": 6353
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.861227246580705e-05,
+      "loss": 3.8992,
+      "step": 6354
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.84944791236959e-05,
+      "loss": 3.7841,
+      "step": 6355
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.83767565241545e-05,
+      "loss": 3.8718,
+      "step": 6356
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.82591046874206e-05,
+      "loss": 3.9118,
+      "step": 6357
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.814152363371992e-05,
+      "loss": 3.7597,
+      "step": 6358
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.802401338326582e-05,
+      "loss": 3.8633,
+      "step": 6359
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.790657395626001e-05,
+      "loss": 3.7583,
+      "step": 6360
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.778920537289154e-05,
+      "loss": 3.7987,
+      "step": 6361
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.767190765333744e-05,
+      "loss": 3.7589,
+      "step": 6362
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.755468081776253e-05,
+      "loss": 3.8669,
+      "step": 6363
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.743752488631945e-05,
+      "loss": 3.9051,
+      "step": 6364
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.732043987914857e-05,
+      "loss": 3.8629,
+      "step": 6365
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.720342581637836e-05,
+      "loss": 3.8896,
+      "step": 6366
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.70864827181247e-05,
+      "loss": 3.7774,
+      "step": 6367
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.696961060449165e-05,
+      "loss": 3.7691,
+      "step": 6368
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.685280949557084e-05,
+      "loss": 3.8547,
+      "step": 6369
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.673607941144169e-05,
+      "loss": 3.7746,
+      "step": 6370
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.66194203721714e-05,
+      "loss": 3.8086,
+      "step": 6371
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.650283239781515e-05,
+      "loss": 3.8663,
+      "step": 6372
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.638631550841552e-05,
+      "loss": 3.8748,
+      "step": 6373
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.626986972400325e-05,
+      "loss": 3.7201,
+      "step": 6374
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.615349506459692e-05,
+      "loss": 4.0123,
+      "step": 6375
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.603719155020246e-05,
+      "loss": 3.6415,
+      "step": 6376
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.592095920081383e-05,
+      "loss": 3.8093,
+      "step": 6377
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.580479803641278e-05,
+      "loss": 3.8554,
+      "step": 6378
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.568870807696872e-05,
+      "loss": 3.858,
+      "step": 6379
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.557268934243867e-05,
+      "loss": 3.7612,
+      "step": 6380
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.545674185276791e-05,
+      "loss": 3.8266,
+      "step": 6381
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.534086562788906e-05,
+      "loss": 3.8269,
+      "step": 6382
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.52250606877224e-05,
+      "loss": 3.8639,
+      "step": 6383
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.510932705217644e-05,
+      "loss": 3.8489,
+      "step": 6384
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.499366474114695e-05,
+      "loss": 3.8198,
+      "step": 6385
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.487807377451767e-05,
+      "loss": 3.7663,
+      "step": 6386
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.476255417216005e-05,
+      "loss": 3.7779,
+      "step": 6387
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.464710595393305e-05,
+      "loss": 3.834,
+      "step": 6388
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.453172913968382e-05,
+      "loss": 3.7928,
+      "step": 6389
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.44164237492469e-05,
+      "loss": 3.7701,
+      "step": 6390
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.430118980244461e-05,
+      "loss": 3.7989,
+      "step": 6391
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.418602731908686e-05,
+      "loss": 3.7614,
+      "step": 6392
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.407093631897167e-05,
+      "loss": 3.8864,
+      "step": 6393
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.395591682188442e-05,
+      "loss": 3.7824,
+      "step": 6394
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.384096884759807e-05,
+      "loss": 3.86,
+      "step": 6395
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.372609241587387e-05,
+      "loss": 3.7489,
+      "step": 6396
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.361128754646024e-05,
+      "loss": 3.7292,
+      "step": 6397
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.349655425909347e-05,
+      "loss": 3.8746,
+      "step": 6398
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.338189257349754e-05,
+      "loss": 3.7733,
+      "step": 6399
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.326730250938413e-05,
+      "loss": 3.6614,
+      "step": 6400
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.31527840864525e-05,
+      "loss": 3.793,
+      "step": 6401
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.303833732438987e-05,
+      "loss": 3.8544,
+      "step": 6402
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.29239622428708e-05,
+      "loss": 3.8838,
+      "step": 6403
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.28096588615579e-05,
+      "loss": 3.8975,
+      "step": 6404
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.269542720010109e-05,
+      "loss": 3.7152,
+      "step": 6405
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.25812672781382e-05,
+      "loss": 3.686,
+      "step": 6406
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.246717911529456e-05,
+      "loss": 3.8791,
+      "step": 6407
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.235316273118332e-05,
+      "loss": 3.8619,
+      "step": 6408
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.223921814540503e-05,
+      "loss": 3.7311,
+      "step": 6409
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.21253453775484e-05,
+      "loss": 3.7083,
+      "step": 6410
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.201154444718917e-05,
+      "loss": 3.7978,
+      "step": 6411
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.189781537389135e-05,
+      "loss": 3.8776,
+      "step": 6412
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.178415817720608e-05,
+      "loss": 3.8739,
+      "step": 6413
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.16705728766724e-05,
+      "loss": 3.8292,
+      "step": 6414
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.155705949181691e-05,
+      "loss": 3.8053,
+      "step": 6415
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.144361804215383e-05,
+      "loss": 3.8278,
+      "step": 6416
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.133024854718524e-05,
+      "loss": 3.7001,
+      "step": 6417
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.121695102640054e-05,
+      "loss": 3.8385,
+      "step": 6418
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.110372549927692e-05,
+      "loss": 3.6954,
+      "step": 6419
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.099057198527899e-05,
+      "loss": 3.8394,
+      "step": 6420
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.087749050385951e-05,
+      "loss": 3.8067,
+      "step": 6421
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.076448107445822e-05,
+      "loss": 3.8021,
+      "step": 6422
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.065154371650285e-05,
+      "loss": 3.7714,
+      "step": 6423
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.053867844940854e-05,
+      "loss": 3.8258,
+      "step": 6424
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.042588529257828e-05,
+      "loss": 3.7888,
+      "step": 6425
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.031316426540252e-05,
+      "loss": 3.8487,
+      "step": 6426
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.02005153872592e-05,
+      "loss": 3.6938,
+      "step": 6427
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.008793867751402e-05,
+      "loss": 3.7081,
+      "step": 6428
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.997543415552011e-05,
+      "loss": 3.876,
+      "step": 6429
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.986300184061856e-05,
+      "loss": 3.8533,
+      "step": 6430
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.975064175213748e-05,
+      "loss": 3.8874,
+      "step": 6431
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.963835390939316e-05,
+      "loss": 3.7503,
+      "step": 6432
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.952613833168909e-05,
+      "loss": 3.8554,
+      "step": 6433
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.941399503831637e-05,
+      "loss": 3.807,
+      "step": 6434
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.930192404855374e-05,
+      "loss": 3.6929,
+      "step": 6435
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.918992538166752e-05,
+      "loss": 3.9622,
+      "step": 6436
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.907799905691143e-05,
+      "loss": 3.7612,
+      "step": 6437
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.896614509352723e-05,
+      "loss": 3.7797,
+      "step": 6438
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.885436351074355e-05,
+      "loss": 3.8188,
+      "step": 6439
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.874265432777728e-05,
+      "loss": 3.7884,
+      "step": 6440
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.863101756383235e-05,
+      "loss": 3.8121,
+      "step": 6441
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.851945323810045e-05,
+      "loss": 3.8042,
+      "step": 6442
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.840796136976075e-05,
+      "loss": 3.7544,
+      "step": 6443
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.829654197797998e-05,
+      "loss": 3.869,
+      "step": 6444
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.818519508191235e-05,
+      "loss": 3.89,
+      "step": 6445
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.807392070069991e-05,
+      "loss": 3.8893,
+      "step": 6446
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.796271885347189e-05,
+      "loss": 3.808,
+      "step": 6447
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.785158955934507e-05,
+      "loss": 3.8443,
+      "step": 6448
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.774053283742406e-05,
+      "loss": 3.7738,
+      "step": 6449
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.762954870680067e-05,
+      "loss": 3.8838,
+      "step": 6450
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.751863718655444e-05,
+      "loss": 3.8439,
+      "step": 6451
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.740779829575217e-05,
+      "loss": 3.8794,
+      "step": 6452
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.729703205344863e-05,
+      "loss": 3.8815,
+      "step": 6453
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.718633847868567e-05,
+      "loss": 3.77,
+      "step": 6454
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.707571759049281e-05,
+      "loss": 3.7538,
+      "step": 6455
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.6965169407887e-05,
+      "loss": 3.8029,
+      "step": 6456
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.685469394987271e-05,
+      "loss": 3.8761,
+      "step": 6457
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.674429123544218e-05,
+      "loss": 3.8209,
+      "step": 6458
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.66339612835748e-05,
+      "loss": 3.8731,
+      "step": 6459
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.652370411323744e-05,
+      "loss": 3.8063,
+      "step": 6460
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.641351974338478e-05,
+      "loss": 3.6581,
+      "step": 6461
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.630340819295878e-05,
+      "loss": 3.9715,
+      "step": 6462
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.619336948088879e-05,
+      "loss": 3.8331,
+      "step": 6463
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.608340362609173e-05,
+      "loss": 3.8235,
+      "step": 6464
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.59735106474721e-05,
+      "loss": 3.8208,
+      "step": 6465
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.586369056392161e-05,
+      "loss": 3.7261,
+      "step": 6466
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.575394339431969e-05,
+      "loss": 3.9212,
+      "step": 6467
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.564426915753331e-05,
+      "loss": 3.8837,
+      "step": 6468
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.553466787241664e-05,
+      "loss": 3.7855,
+      "step": 6469
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.542513955781138e-05,
+      "loss": 3.8563,
+      "step": 6470
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.53156842325467e-05,
+      "loss": 3.7702,
+      "step": 6471
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.520630191543932e-05,
+      "loss": 3.8458,
+      "step": 6472
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.509699262529307e-05,
+      "loss": 3.856,
+      "step": 6473
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.498775638089988e-05,
+      "loss": 3.8895,
+      "step": 6474
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.487859320103846e-05,
+      "loss": 3.7707,
+      "step": 6475
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.476950310447522e-05,
+      "loss": 3.8001,
+      "step": 6476
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.466048610996423e-05,
+      "loss": 3.7763,
+      "step": 6477
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.455154223624661e-05,
+      "loss": 3.7501,
+      "step": 6478
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.444267150205108e-05,
+      "loss": 3.8614,
+      "step": 6479
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.433387392609387e-05,
+      "loss": 3.7361,
+      "step": 6480
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.422514952707832e-05,
+      "loss": 3.8173,
+      "step": 6481
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.411649832369566e-05,
+      "loss": 3.817,
+      "step": 6482
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.400792033462428e-05,
+      "loss": 3.7252,
+      "step": 6483
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.389941557852986e-05,
+      "loss": 3.8519,
+      "step": 6484
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.379098407406553e-05,
+      "loss": 3.7893,
+      "step": 6485
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.368262583987228e-05,
+      "loss": 3.7305,
+      "step": 6486
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.357434089457788e-05,
+      "loss": 3.7848,
+      "step": 6487
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.346612925679775e-05,
+      "loss": 3.734,
+      "step": 6488
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.33579909451349e-05,
+      "loss": 3.8242,
+      "step": 6489
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.324992597817947e-05,
+      "loss": 3.8984,
+      "step": 6490
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.314193437450912e-05,
+      "loss": 3.8627,
+      "step": 6491
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.303401615268878e-05,
+      "loss": 3.9573,
+      "step": 6492
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.292617133127083e-05,
+      "loss": 3.8396,
+      "step": 6493
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.281839992879502e-05,
+      "loss": 3.8322,
+      "step": 6494
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.27107019637886e-05,
+      "loss": 3.6538,
+      "step": 6495
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.260307745476619e-05,
+      "loss": 3.9288,
+      "step": 6496
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.249552642022955e-05,
+      "loss": 3.7418,
+      "step": 6497
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.238804887866796e-05,
+      "loss": 3.7818,
+      "step": 6498
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.228064484855806e-05,
+      "loss": 3.8279,
+      "step": 6499
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.217331434836394e-05,
+      "loss": 3.9035,
+      "step": 6500
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.206605739653682e-05,
+      "loss": 3.8609,
+      "step": 6501
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.195887401151535e-05,
+      "loss": 3.6489,
+      "step": 6502
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.185176421172573e-05,
+      "loss": 3.9457,
+      "step": 6503
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.174472801558146e-05,
+      "loss": 3.6243,
+      "step": 6504
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.16377654414832e-05,
+      "loss": 3.8062,
+      "step": 6505
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.153087650781909e-05,
+      "loss": 3.9306,
+      "step": 6506
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.142406123296453e-05,
+      "loss": 3.8506,
+      "step": 6507
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.131731963528232e-05,
+      "loss": 3.7288,
+      "step": 6508
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.121065173312253e-05,
+      "loss": 3.9169,
+      "step": 6509
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.110405754482268e-05,
+      "loss": 3.7452,
+      "step": 6510
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.09975370887076e-05,
+      "loss": 3.8964,
+      "step": 6511
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.089109038308927e-05,
+      "loss": 3.7389,
+      "step": 6512
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.078471744626707e-05,
+      "loss": 3.8825,
+      "step": 6513
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.067841829652794e-05,
+      "loss": 3.8208,
+      "step": 6514
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.057219295214579e-05,
+      "loss": 3.7879,
+      "step": 6515
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.046604143138197e-05,
+      "loss": 3.7748,
+      "step": 6516
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.035996375248527e-05,
+      "loss": 3.8477,
+      "step": 6517
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.025395993369165e-05,
+      "loss": 3.8842,
+      "step": 6518
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.01480299932244e-05,
+      "loss": 3.8662,
+      "step": 6519
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.004217394929402e-05,
+      "loss": 3.7778,
+      "step": 6520
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 6.993639182009842e-05,
+      "loss": 3.7956,
+      "step": 6521
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 6.983068362382273e-05,
+      "loss": 3.7892,
+      "step": 6522
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 6.972504937863955e-05,
+      "loss": 3.7978,
+      "step": 6523
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.961948910270843e-05,
+      "loss": 3.6544,
+      "step": 6524
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.951400281417669e-05,
+      "loss": 3.7647,
+      "step": 6525
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.940859053117842e-05,
+      "loss": 3.7698,
+      "step": 6526
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.930325227183538e-05,
+      "loss": 3.9012,
+      "step": 6527
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.919798805425626e-05,
+      "loss": 3.7047,
+      "step": 6528
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.909279789653733e-05,
+      "loss": 3.9118,
+      "step": 6529
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.898768181676179e-05,
+      "loss": 3.8016,
+      "step": 6530
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.888263983300047e-05,
+      "loss": 3.7686,
+      "step": 6531
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.877767196331147e-05,
+      "loss": 3.7311,
+      "step": 6532
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.867277822573975e-05,
+      "loss": 3.7387,
+      "step": 6533
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.856795863831788e-05,
+      "loss": 3.7005,
+      "step": 6534
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.846321321906552e-05,
+      "loss": 3.7693,
+      "step": 6535
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.835854198598956e-05,
+      "loss": 3.7251,
+      "step": 6536
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.825394495708415e-05,
+      "loss": 3.8438,
+      "step": 6537
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.814942215033098e-05,
+      "loss": 3.7885,
+      "step": 6538
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.804497358369854e-05,
+      "loss": 3.7198,
+      "step": 6539
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.794059927514268e-05,
+      "loss": 3.8003,
+      "step": 6540
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.783629924260681e-05,
+      "loss": 3.613,
+      "step": 6541
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.773207350402117e-05,
+      "loss": 3.854,
+      "step": 6542
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.762792207730334e-05,
+      "loss": 3.6577,
+      "step": 6543
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.752384498035824e-05,
+      "loss": 3.9545,
+      "step": 6544
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.741984223107773e-05,
+      "loss": 3.7467,
+      "step": 6545
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.731591384734137e-05,
+      "loss": 3.6758,
+      "step": 6546
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.72120598470155e-05,
+      "loss": 3.838,
+      "step": 6547
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.710828024795391e-05,
+      "loss": 3.906,
+      "step": 6548
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.70045750679974e-05,
+      "loss": 3.8626,
+      "step": 6549
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.690094432497407e-05,
+      "loss": 3.7397,
+      "step": 6550
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.679738803669943e-05,
+      "loss": 3.7608,
+      "step": 6551
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.669390622097576e-05,
+      "loss": 3.9251,
+      "step": 6552
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.65904988955931e-05,
+      "loss": 3.8796,
+      "step": 6553
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.648716607832811e-05,
+      "loss": 3.7867,
+      "step": 6554
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.638390778694503e-05,
+      "loss": 3.7386,
+      "step": 6555
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.62807240391951e-05,
+      "loss": 3.7741,
+      "step": 6556
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.617761485281687e-05,
+      "loss": 3.6431,
+      "step": 6557
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.607458024553575e-05,
+      "loss": 3.7476,
+      "step": 6558
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.597162023506492e-05,
+      "loss": 3.9206,
+      "step": 6559
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.586873483910416e-05,
+      "loss": 3.6936,
+      "step": 6560
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.576592407534088e-05,
+      "loss": 3.846,
+      "step": 6561
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.566318796144933e-05,
+      "loss": 4.0213,
+      "step": 6562
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.556052651509104e-05,
+      "loss": 3.7499,
+      "step": 6563
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.545793975391468e-05,
+      "loss": 3.8706,
+      "step": 6564
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.535542769555608e-05,
+      "loss": 3.9132,
+      "step": 6565
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.525299035763827e-05,
+      "loss": 3.7151,
+      "step": 6566
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.515062775777148e-05,
+      "loss": 3.832,
+      "step": 6567
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.504833991355292e-05,
+      "loss": 3.7997,
+      "step": 6568
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.494612684256718e-05,
+      "loss": 3.7333,
+      "step": 6569
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.484398856238583e-05,
+      "loss": 3.8587,
+      "step": 6570
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.47419250905676e-05,
+      "loss": 3.7446,
+      "step": 6571
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.463993644465843e-05,
+      "loss": 3.6138,
+      "step": 6572
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.453802264219117e-05,
+      "loss": 3.7001,
+      "step": 6573
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.443618370068622e-05,
+      "loss": 3.7853,
+      "step": 6574
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.43344196376508e-05,
+      "loss": 3.8414,
+      "step": 6575
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.423273047057932e-05,
+      "loss": 3.6463,
+      "step": 6576
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.413111621695323e-05,
+      "loss": 3.939,
+      "step": 6577
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.402957689424138e-05,
+      "loss": 3.8877,
+      "step": 6578
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.392811251989945e-05,
+      "loss": 3.7851,
+      "step": 6579
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.382672311137039e-05,
+      "loss": 3.7027,
+      "step": 6580
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.3725408686084e-05,
+      "loss": 3.7777,
+      "step": 6581
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.362416926145775e-05,
+      "loss": 3.7821,
+      "step": 6582
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.352300485489571e-05,
+      "loss": 3.8644,
+      "step": 6583
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.342191548378923e-05,
+      "loss": 3.765,
+      "step": 6584
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.33209011655167e-05,
+      "loss": 3.853,
+      "step": 6585
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.321996191744367e-05,
+      "loss": 3.8916,
+      "step": 6586
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.311909775692265e-05,
+      "loss": 3.6689,
+      "step": 6587
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.301830870129349e-05,
+      "loss": 3.8248,
+      "step": 6588
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.291759476788312e-05,
+      "loss": 3.8693,
+      "step": 6589
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.281695597400534e-05,
+      "loss": 3.6881,
+      "step": 6590
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.271639233696103e-05,
+      "loss": 3.9544,
+      "step": 6591
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.261590387403832e-05,
+      "loss": 3.8177,
+      "step": 6592
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.251549060251233e-05,
+      "loss": 3.8547,
+      "step": 6593
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.241515253964513e-05,
+      "loss": 3.7908,
+      "step": 6594
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.231488970268628e-05,
+      "loss": 3.6965,
+      "step": 6595
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.221470210887181e-05,
+      "loss": 3.9243,
+      "step": 6596
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.211458977542545e-05,
+      "loss": 3.9195,
+      "step": 6597
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.201455271955747e-05,
+      "loss": 3.9323,
+      "step": 6598
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.191459095846546e-05,
+      "loss": 3.7322,
+      "step": 6599
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.181470450933397e-05,
+      "loss": 3.8506,
+      "step": 6600
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.171489338933466e-05,
+      "loss": 3.8049,
+      "step": 6601
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.161515761562614e-05,
+      "loss": 3.8545,
+      "step": 6602
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.151549720535432e-05,
+      "loss": 3.8084,
+      "step": 6603
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.141591217565185e-05,
+      "loss": 3.8209,
+      "step": 6604
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.131640254363846e-05,
+      "loss": 3.7271,
+      "step": 6605
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.121696832642126e-05,
+      "loss": 3.8679,
+      "step": 6606
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.111760954109402e-05,
+      "loss": 3.6951,
+      "step": 6607
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.101832620473763e-05,
+      "loss": 3.8394,
+      "step": 6608
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.0919118334419985e-05,
+      "loss": 3.9717,
+      "step": 6609
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.081998594719629e-05,
+      "loss": 3.893,
+      "step": 6610
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.07209290601084e-05,
+      "loss": 3.8497,
+      "step": 6611
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.062194769018542e-05,
+      "loss": 4.0016,
+      "step": 6612
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.0523041854443295e-05,
+      "loss": 3.9007,
+      "step": 6613
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.042421156988498e-05,
+      "loss": 3.8599,
+      "step": 6614
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.032545685350088e-05,
+      "loss": 3.8077,
+      "step": 6615
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.0226777722267814e-05,
+      "loss": 3.8054,
+      "step": 6616
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.012817419314992e-05,
+      "loss": 3.7157,
+      "step": 6617
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.002964628309837e-05,
+      "loss": 3.8274,
+      "step": 6618
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.993119400905123e-05,
+      "loss": 3.8494,
+      "step": 6619
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.98328173879335e-05,
+      "loss": 3.8678,
+      "step": 6620
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.9734516436657335e-05,
+      "loss": 3.8546,
+      "step": 6621
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.9636291172121824e-05,
+      "loss": 3.8648,
+      "step": 6622
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.95381416112129e-05,
+      "loss": 3.6461,
+      "step": 6623
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.944006777080363e-05,
+      "loss": 3.8148,
+      "step": 6624
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.9342069667754295e-05,
+      "loss": 3.7107,
+      "step": 6625
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.92441473189117e-05,
+      "loss": 3.853,
+      "step": 6626
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.9146300741109885e-05,
+      "loss": 3.7369,
+      "step": 6627
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.9048529951169775e-05,
+      "loss": 3.8212,
+      "step": 6628
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.8950834965899315e-05,
+      "loss": 3.7433,
+      "step": 6629
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.8853215802093244e-05,
+      "loss": 3.9394,
+      "step": 6630
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.875567247653374e-05,
+      "loss": 3.7308,
+      "step": 6631
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.8658205005989506e-05,
+      "loss": 3.8858,
+      "step": 6632
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.85608134072162e-05,
+      "loss": 3.8994,
+      "step": 6633
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.8463497696956746e-05,
+      "loss": 3.6671,
+      "step": 6634
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.836625789194078e-05,
+      "loss": 3.7854,
+      "step": 6635
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.826909400888491e-05,
+      "loss": 3.8798,
+      "step": 6636
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.81720060644928e-05,
+      "loss": 3.8936,
+      "step": 6637
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.807499407545486e-05,
+      "loss": 3.7097,
+      "step": 6638
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.797805805844875e-05,
+      "loss": 3.9607,
+      "step": 6639
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.7881198030138806e-05,
+      "loss": 3.7198,
+      "step": 6640
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.7784414007176436e-05,
+      "loss": 3.6591,
+      "step": 6641
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.768770600619977e-05,
+      "loss": 3.8144,
+      "step": 6642
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.759107404383429e-05,
+      "loss": 3.7144,
+      "step": 6643
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.749451813669204e-05,
+      "loss": 3.8356,
+      "step": 6644
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.7398038301371916e-05,
+      "loss": 3.7936,
+      "step": 6645
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.7301634554460246e-05,
+      "loss": 3.8868,
+      "step": 6646
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.7205306912529786e-05,
+      "loss": 3.7774,
+      "step": 6647
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.71090553921404e-05,
+      "loss": 3.7301,
+      "step": 6648
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.701288000983884e-05,
+      "loss": 3.7799,
+      "step": 6649
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.691678078215873e-05,
+      "loss": 3.8604,
+      "step": 6650
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.682075772562051e-05,
+      "loss": 3.8119,
+      "step": 6651
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.672481085673198e-05,
+      "loss": 3.9804,
+      "step": 6652
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.662894019198722e-05,
+      "loss": 3.7274,
+      "step": 6653
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.653314574786778e-05,
+      "loss": 3.76,
+      "step": 6654
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.643742754084163e-05,
+      "loss": 3.8472,
+      "step": 6655
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.634178558736397e-05,
+      "loss": 3.8069,
+      "step": 6656
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.624621990387668e-05,
+      "loss": 3.7367,
+      "step": 6657
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.615073050680858e-05,
+      "loss": 3.7217,
+      "step": 6658
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.6055317412575354e-05,
+      "loss": 3.9585,
+      "step": 6659
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.5959980637579844e-05,
+      "loss": 3.6489,
+      "step": 6660
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.5864720198211316e-05,
+      "loss": 3.8663,
+      "step": 6661
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.5769536110846344e-05,
+      "loss": 3.7169,
+      "step": 6662
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.567442839184805e-05,
+      "loss": 3.8608,
+      "step": 6663
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.5579397057566646e-05,
+      "loss": 3.8081,
+      "step": 6664
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.548444212433901e-05,
+      "loss": 3.8985,
+      "step": 6665
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.53895636084889e-05,
+      "loss": 3.7948,
+      "step": 6666
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.5294761526327345e-05,
+      "loss": 3.8312,
+      "step": 6667
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.520003589415168e-05,
+      "loss": 3.6694,
+      "step": 6668
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.510538672824644e-05,
+      "loss": 3.7729,
+      "step": 6669
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.501081404488278e-05,
+      "loss": 3.6923,
+      "step": 6670
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.491631786031903e-05,
+      "loss": 3.7685,
+      "step": 6671
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.482189819080013e-05,
+      "loss": 3.6435,
+      "step": 6672
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.472755505255783e-05,
+      "loss": 3.7618,
+      "step": 6673
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.463328846181082e-05,
+      "loss": 3.7617,
+      "step": 6674
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.453909843476468e-05,
+      "loss": 3.7943,
+      "step": 6675
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.44449849876118e-05,
+      "loss": 3.7231,
+      "step": 6676
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.4350948136531296e-05,
+      "loss": 3.756,
+      "step": 6677
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.4256987897689234e-05,
+      "loss": 3.9032,
+      "step": 6678
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.416310428723836e-05,
+      "loss": 3.665,
+      "step": 6679
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.406929732131849e-05,
+      "loss": 3.8858,
+      "step": 6680
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.397556701605605e-05,
+      "loss": 3.7361,
+      "step": 6681
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.388191338756443e-05,
+      "loss": 3.8982,
+      "step": 6682
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.378833645194375e-05,
+      "loss": 3.7905,
+      "step": 6683
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.369483622528104e-05,
+      "loss": 3.7376,
+      "step": 6684
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.360141272364993e-05,
+      "loss": 3.7728,
+      "step": 6685
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.350806596311109e-05,
+      "loss": 3.7417,
+      "step": 6686
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.341479595971177e-05,
+      "loss": 3.7744,
+      "step": 6687
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.3321602729486275e-05,
+      "loss": 3.8914,
+      "step": 6688
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.322848628845578e-05,
+      "loss": 3.6826,
+      "step": 6689
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.313544665262782e-05,
+      "loss": 3.9094,
+      "step": 6690
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.3042483837997145e-05,
+      "loss": 3.7775,
+      "step": 6691
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.294959786054504e-05,
+      "loss": 3.7497,
+      "step": 6692
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.285678873623972e-05,
+      "loss": 3.7058,
+      "step": 6693
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.276405648103616e-05,
+      "loss": 3.8267,
+      "step": 6694
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.267140111087604e-05,
+      "loss": 3.8298,
+      "step": 6695
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.257882264168795e-05,
+      "loss": 3.815,
+      "step": 6696
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.248632108938728e-05,
+      "loss": 3.7653,
+      "step": 6697
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.239389646987591e-05,
+      "loss": 3.7132,
+      "step": 6698
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.230154879904292e-05,
+      "loss": 3.7619,
+      "step": 6699
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.2209278092763824e-05,
+      "loss": 3.7891,
+      "step": 6700
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.21170843669011e-05,
+      "loss": 3.841,
+      "step": 6701
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.20249676373038e-05,
+      "loss": 3.7754,
+      "step": 6702
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.1932927919808025e-05,
+      "loss": 3.8982,
+      "step": 6703
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.184096523023635e-05,
+      "loss": 3.894,
+      "step": 6704
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.1749079584398284e-05,
+      "loss": 3.8801,
+      "step": 6705
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.165727099808998e-05,
+      "loss": 3.6746,
+      "step": 6706
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.156553948709436e-05,
+      "loss": 3.7433,
+      "step": 6707
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.147388506718126e-05,
+      "loss": 3.7012,
+      "step": 6708
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.138230775410713e-05,
+      "loss": 3.7782,
+      "step": 6709
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.1290807563614984e-05,
+      "loss": 3.8505,
+      "step": 6710
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.1199384511435023e-05,
+      "loss": 3.6901,
+      "step": 6711
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.110803861328378e-05,
+      "loss": 3.8009,
+      "step": 6712
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.10167698848647e-05,
+      "loss": 3.8534,
+      "step": 6713
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.092557834186795e-05,
+      "loss": 3.7889,
+      "step": 6714
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.083446399997044e-05,
+      "loss": 3.8607,
+      "step": 6715
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.0743426874835574e-05,
+      "loss": 3.737,
+      "step": 6716
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.0652466982113896e-05,
+      "loss": 3.8669,
+      "step": 6717
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.0561584337442514e-05,
+      "loss": 3.8729,
+      "step": 6718
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.047077895644514e-05,
+      "loss": 3.9644,
+      "step": 6719
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.038005085473224e-05,
+      "loss": 3.6553,
+      "step": 6720
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.0289400047901055e-05,
+      "loss": 3.7374,
+      "step": 6721
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.019882655153546e-05,
+      "loss": 3.7681,
+      "step": 6722
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.0108330381206044e-05,
+      "loss": 3.7212,
+      "step": 6723
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.001791155247032e-05,
+      "loss": 3.6764,
+      "step": 6724
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.9927570080872176e-05,
+      "loss": 3.9732,
+      "step": 6725
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.983730598194247e-05,
+      "loss": 3.7344,
+      "step": 6726
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.974711927119863e-05,
+      "loss": 3.8307,
+      "step": 6727
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.9657009964144794e-05,
+      "loss": 3.7244,
+      "step": 6728
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.956697807627169e-05,
+      "loss": 3.8122,
+      "step": 6729
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.9477023623057e-05,
+      "loss": 3.8411,
+      "step": 6730
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.9387146619964684e-05,
+      "loss": 3.8287,
+      "step": 6731
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.929734708244599e-05,
+      "loss": 3.7816,
+      "step": 6732
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.9207625025938285e-05,
+      "loss": 3.7733,
+      "step": 6733
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.911798046586591e-05,
+      "loss": 3.8087,
+      "step": 6734
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.902841341763964e-05,
+      "loss": 3.8507,
+      "step": 6735
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.893892389665727e-05,
+      "loss": 3.731,
+      "step": 6736
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.8849511918303117e-05,
+      "loss": 3.7434,
+      "step": 6737
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.8760177497947866e-05,
+      "loss": 3.8141,
+      "step": 6738
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.867092065094947e-05,
+      "loss": 3.875,
+      "step": 6739
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.858174139265209e-05,
+      "loss": 3.9065,
+      "step": 6740
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.849263973838664e-05,
+      "loss": 4.0558,
+      "step": 6741
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.840361570347068e-05,
+      "loss": 3.8964,
+      "step": 6742
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.8314669303208605e-05,
+      "loss": 3.8441,
+      "step": 6743
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.822580055289116e-05,
+      "loss": 3.8671,
+      "step": 6744
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.8137009467795976e-05,
+      "loss": 3.8266,
+      "step": 6745
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.8048296063187446e-05,
+      "loss": 3.7395,
+      "step": 6746
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.795966035431626e-05,
+      "loss": 3.7438,
+      "step": 6747
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.787110235642001e-05,
+      "loss": 3.7422,
+      "step": 6748
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.778262208472273e-05,
+      "loss": 3.7719,
+      "step": 6749
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.7694219554435316e-05,
+      "loss": 3.7927,
+      "step": 6750
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.760589478075511e-05,
+      "loss": 3.6311,
+      "step": 6751
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.751764777886614e-05,
+      "loss": 3.6952,
+      "step": 6752
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.742947856393909e-05,
+      "loss": 3.9276,
+      "step": 6753
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.734138715113145e-05,
+      "loss": 3.6257,
+      "step": 6754
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.7253373555587006e-05,
+      "loss": 3.7406,
+      "step": 6755
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.716543779243637e-05,
+      "loss": 3.7871,
+      "step": 6756
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.707757987679667e-05,
+      "loss": 3.9658,
+      "step": 6757
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.6989799823771714e-05,
+      "loss": 3.8145,
+      "step": 6758
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.6902097648451824e-05,
+      "loss": 3.793,
+      "step": 6759
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.681447336591421e-05,
+      "loss": 3.7813,
+      "step": 6760
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.672692699122238e-05,
+      "loss": 3.7728,
+      "step": 6761
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.663945853942653e-05,
+      "loss": 3.7357,
+      "step": 6762
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.655206802556361e-05,
+      "loss": 3.7614,
+      "step": 6763
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.646475546465706e-05,
+      "loss": 3.7089,
+      "step": 6764
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.637752087171687e-05,
+      "loss": 3.9452,
+      "step": 6765
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.6290364261739545e-05,
+      "loss": 3.9694,
+      "step": 6766
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.6203285649708603e-05,
+      "loss": 3.9176,
+      "step": 6767
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.611628505059373e-05,
+      "loss": 3.7071,
+      "step": 6768
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.60293624793513e-05,
+      "loss": 3.8459,
+      "step": 6769
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.59425179509243e-05,
+      "loss": 3.8504,
+      "step": 6770
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.58557514802424e-05,
+      "loss": 3.7425,
+      "step": 6771
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.576906308222162e-05,
+      "loss": 3.7619,
+      "step": 6772
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.568245277176486e-05,
+      "loss": 3.9365,
+      "step": 6773
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.559592056376133e-05,
+      "loss": 3.8201,
+      "step": 6774
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.550946647308701e-05,
+      "loss": 3.8155,
+      "step": 6775
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.5423090514604304e-05,
+      "loss": 3.7281,
+      "step": 6776
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.533679270316227e-05,
+      "loss": 3.8866,
+      "step": 6777
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.525057305359648e-05,
+      "loss": 3.8293,
+      "step": 6778
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.5164431580729084e-05,
+      "loss": 4.0109,
+      "step": 6779
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.5078368299368675e-05,
+      "loss": 3.7449,
+      "step": 6780
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.499238322431071e-05,
+      "loss": 3.8993,
+      "step": 6781
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.490647637033701e-05,
+      "loss": 3.7961,
+      "step": 6782
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.482064775221595e-05,
+      "loss": 3.8086,
+      "step": 6783
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.473489738470243e-05,
+      "loss": 3.7437,
+      "step": 6784
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.464922528253795e-05,
+      "loss": 3.7223,
+      "step": 6785
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.4563631460450505e-05,
+      "loss": 3.8851,
+      "step": 6786
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.447811593315459e-05,
+      "loss": 3.7836,
+      "step": 6787
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.439267871535152e-05,
+      "loss": 3.8625,
+      "step": 6788
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.430731982172886e-05,
+      "loss": 3.6914,
+      "step": 6789
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.422203926696061e-05,
+      "loss": 3.8691,
+      "step": 6790
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.4136837065707756e-05,
+      "loss": 3.7503,
+      "step": 6791
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.405171323261742e-05,
+      "loss": 3.6392,
+      "step": 6792
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.396666778232344e-05,
+      "loss": 3.7815,
+      "step": 6793
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.388170072944603e-05,
+      "loss": 3.826,
+      "step": 6794
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.379681208859193e-05,
+      "loss": 3.7279,
+      "step": 6795
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.37120018743547e-05,
+      "loss": 3.9134,
+      "step": 6796
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.3627270101314064e-05,
+      "loss": 3.7354,
+      "step": 6797
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.3542616784036426e-05,
+      "loss": 3.8387,
+      "step": 6798
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.345804193707459e-05,
+      "loss": 3.8224,
+      "step": 6799
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.3373545574968144e-05,
+      "loss": 3.8108,
+      "step": 6800
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.328912771224286e-05,
+      "loss": 3.7526,
+      "step": 6801
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.320478836341113e-05,
+      "loss": 3.7861,
+      "step": 6802
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.312052754297197e-05,
+      "loss": 3.8765,
+      "step": 6803
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.303634526541067e-05,
+      "loss": 3.8132,
+      "step": 6804
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.2952241545199225e-05,
+      "loss": 3.7646,
+      "step": 6805
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.2868216396796055e-05,
+      "loss": 3.8776,
+      "step": 6806
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.278426983464595e-05,
+      "loss": 3.6195,
+      "step": 6807
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.2700401873180303e-05,
+      "loss": 3.8085,
+      "step": 6808
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.261661252681714e-05,
+      "loss": 3.7449,
+      "step": 6809
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.253290180996067e-05,
+      "loss": 3.8095,
+      "step": 6810
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.2449269737001804e-05,
+      "loss": 3.7703,
+      "step": 6811
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.2365716322317894e-05,
+      "loss": 3.8348,
+      "step": 6812
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.228224158027272e-05,
+      "loss": 3.9054,
+      "step": 6813
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.2198845525216524e-05,
+      "loss": 3.9057,
+      "step": 6814
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.211552817148606e-05,
+      "loss": 3.7049,
+      "step": 6815
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.2032289533404434e-05,
+      "loss": 3.8797,
+      "step": 6816
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.194912962528158e-05,
+      "loss": 3.7677,
+      "step": 6817
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.186604846141345e-05,
+      "loss": 3.7689,
+      "step": 6818
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.17830460560828e-05,
+      "loss": 3.7749,
+      "step": 6819
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.170012242355864e-05,
+      "loss": 3.8029,
+      "step": 6820
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.1617277578096526e-05,
+      "loss": 3.7392,
+      "step": 6821
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.153451153393839e-05,
+      "loss": 3.7522,
+      "step": 6822
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.1451824305312634e-05,
+      "loss": 3.7138,
+      "step": 6823
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.136921590643433e-05,
+      "loss": 3.9602,
+      "step": 6824
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.1286686351504735e-05,
+      "loss": 3.7957,
+      "step": 6825
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.120423565471154e-05,
+      "loss": 3.8346,
+      "step": 6826
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.1121863830229036e-05,
+      "loss": 3.8958,
+      "step": 6827
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.103957089221799e-05,
+      "loss": 3.8252,
+      "step": 6828
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.0957356854825435e-05,
+      "loss": 3.8469,
+      "step": 6829
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.0875221732184884e-05,
+      "loss": 3.8955,
+      "step": 6830
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.0793165538416275e-05,
+      "loss": 3.7812,
+      "step": 6831
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.0711188287626156e-05,
+      "loss": 3.7515,
+      "step": 6832
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.062928999390736e-05,
+      "loss": 3.7249,
+      "step": 6833
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.054747067133907e-05,
+      "loss": 3.8369,
+      "step": 6834
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.0465730333986985e-05,
+      "loss": 3.7839,
+      "step": 6835
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.038406899590313e-05,
+      "loss": 3.7689,
+      "step": 6836
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.030248667112629e-05,
+      "loss": 3.885,
+      "step": 6837
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.022098337368113e-05,
+      "loss": 3.8374,
+      "step": 6838
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.013955911757922e-05,
+      "loss": 3.7326,
+      "step": 6839
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.0058213916818295e-05,
+      "loss": 3.691,
+      "step": 6840
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.997694778538247e-05,
+      "loss": 3.7451,
+      "step": 6841
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.989576073724238e-05,
+      "loss": 3.7124,
+      "step": 6842
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.981465278635504e-05,
+      "loss": 3.8943,
+      "step": 6843
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.973362394666369e-05,
+      "loss": 3.8978,
+      "step": 6844
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.965267423209834e-05,
+      "loss": 3.9624,
+      "step": 6845
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.957180365657498e-05,
+      "loss": 3.878,
+      "step": 6846
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.949101223399643e-05,
+      "loss": 3.7558,
+      "step": 6847
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.941029997825152e-05,
+      "loss": 3.7812,
+      "step": 6848
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.932966690321571e-05,
+      "loss": 3.6936,
+      "step": 6849
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.9249113022750694e-05,
+      "loss": 3.7976,
+      "step": 6850
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.9168638350704614e-05,
+      "loss": 3.8347,
+      "step": 6851
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.908824290091195e-05,
+      "loss": 3.7582,
+      "step": 6852
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.900792668719372e-05,
+      "loss": 3.8187,
+      "step": 6853
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.892768972335725e-05,
+      "loss": 3.8373,
+      "step": 6854
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.884753202319596e-05,
+      "loss": 3.7175,
+      "step": 6855
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.8767453600490167e-05,
+      "loss": 3.7861,
+      "step": 6856
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.868745446900618e-05,
+      "loss": 3.8977,
+      "step": 6857
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.8607534642496777e-05,
+      "loss": 3.7631,
+      "step": 6858
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.852769413470098e-05,
+      "loss": 3.8454,
+      "step": 6859
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.844793295934451e-05,
+      "loss": 3.8753,
+      "step": 6860
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.8368251130139086e-05,
+      "loss": 3.724,
+      "step": 6861
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.828864866078302e-05,
+      "loss": 3.7522,
+      "step": 6862
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.820912556496087e-05,
+      "loss": 3.8462,
+      "step": 6863
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.812968185634352e-05,
+      "loss": 3.7413,
+      "step": 6864
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.805031754858834e-05,
+      "loss": 3.7479,
+      "step": 6865
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.7971032655339024e-05,
+      "loss": 3.8497,
+      "step": 6866
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.789182719022538e-05,
+      "loss": 3.7822,
+      "step": 6867
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.781270116686392e-05,
+      "loss": 3.6132,
+      "step": 6868
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.77336545988573e-05,
+      "loss": 3.8949,
+      "step": 6869
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.765468749979445e-05,
+      "loss": 3.8932,
+      "step": 6870
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.757579988325083e-05,
+      "loss": 3.7872,
+      "step": 6871
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.749699176278809e-05,
+      "loss": 3.8016,
+      "step": 6872
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.741826315195418e-05,
+      "loss": 3.7763,
+      "step": 6873
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.733961406428354e-05,
+      "loss": 3.7891,
+      "step": 6874
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.7261044513296926e-05,
+      "loss": 3.6889,
+      "step": 6875
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.7182554512501296e-05,
+      "loss": 3.6861,
+      "step": 6876
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.710414407538998e-05,
+      "loss": 3.8507,
+      "step": 6877
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.7025813215442684e-05,
+      "loss": 3.812,
+      "step": 6878
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.694756194612531e-05,
+      "loss": 3.7686,
+      "step": 6879
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.686939028089015e-05,
+      "loss": 3.6443,
+      "step": 6880
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.679129823317595e-05,
+      "loss": 3.8077,
+      "step": 6881
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.6713285816407535e-05,
+      "loss": 3.8308,
+      "step": 6882
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.663535304399607e-05,
+      "loss": 3.871,
+      "step": 6883
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.655749992933932e-05,
+      "loss": 3.8396,
+      "step": 6884
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.647972648582104e-05,
+      "loss": 3.845,
+      "step": 6885
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.640203272681131e-05,
+      "loss": 3.8794,
+      "step": 6886
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.632441866566672e-05,
+      "loss": 3.8146,
+      "step": 6887
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.624688431572981e-05,
+      "loss": 3.6848,
+      "step": 6888
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.6169429690329825e-05,
+      "loss": 3.8316,
+      "step": 6889
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.60920548027821e-05,
+      "loss": 3.7958,
+      "step": 6890
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.6014759666388284e-05,
+      "loss": 3.7423,
+      "step": 6891
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.593754429443613e-05,
+      "loss": 3.7832,
+      "step": 6892
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.586040870020008e-05,
+      "loss": 3.962,
+      "step": 6893
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.578335289694051e-05,
+      "loss": 3.9006,
+      "step": 6894
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.570637689790418e-05,
+      "loss": 3.7883,
+      "step": 6895
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.562948071632427e-05,
+      "loss": 3.828,
+      "step": 6896
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.5552664365420094e-05,
+      "loss": 3.8769,
+      "step": 6897
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.547592785839726e-05,
+      "loss": 3.8877,
+      "step": 6898
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.539927120844766e-05,
+      "loss": 3.8914,
+      "step": 6899
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.532269442874941e-05,
+      "loss": 3.8307,
+      "step": 6900
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.524619753246694e-05,
+      "loss": 3.7033,
+      "step": 6901
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.5169780532751105e-05,
+      "loss": 3.9517,
+      "step": 6902
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.509344344273863e-05,
+      "loss": 3.9869,
+      "step": 6903
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.501718627555301e-05,
+      "loss": 3.8712,
+      "step": 6904
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.4941009044303585e-05,
+      "loss": 3.9151,
+      "step": 6905
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.486491176208617e-05,
+      "loss": 3.8392,
+      "step": 6906
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.4788894441982685e-05,
+      "loss": 3.942,
+      "step": 6907
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.471295709706146e-05,
+      "loss": 3.7437,
+      "step": 6908
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.4637099740376935e-05,
+      "loss": 3.8078,
+      "step": 6909
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.456132238496984e-05,
+      "loss": 3.7022,
+      "step": 6910
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.448562504386738e-05,
+      "loss": 3.7687,
+      "step": 6911
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.4410007730082685e-05,
+      "loss": 3.6948,
+      "step": 6912
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.4334470456615254e-05,
+      "loss": 3.6613,
+      "step": 6913
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.425901323645075e-05,
+      "loss": 3.8524,
+      "step": 6914
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.418363608256125e-05,
+      "loss": 3.8701,
+      "step": 6915
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.410833900790483e-05,
+      "loss": 3.8378,
+      "step": 6916
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.403312202542608e-05,
+      "loss": 3.9004,
+      "step": 6917
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.395798514805565e-05,
+      "loss": 3.9263,
+      "step": 6918
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.3882928388710376e-05,
+      "loss": 3.8773,
+      "step": 6919
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.3807951760293275e-05,
+      "loss": 3.7997,
+      "step": 6920
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.373305527569398e-05,
+      "loss": 3.7904,
+      "step": 6921
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.3658238947787855e-05,
+      "loss": 3.7992,
+      "step": 6922
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.358350278943678e-05,
+      "loss": 3.8023,
+      "step": 6923
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.3508846813488646e-05,
+      "loss": 3.8641,
+      "step": 6924
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.343427103277785e-05,
+      "loss": 3.7631,
+      "step": 6925
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.335977546012481e-05,
+      "loss": 3.7399,
+      "step": 6926
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.328536010833605e-05,
+      "loss": 3.9006,
+      "step": 6927
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.3211024990204565e-05,
+      "loss": 3.8328,
+      "step": 6928
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.313677011850924e-05,
+      "loss": 3.7927,
+      "step": 6929
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.306259550601559e-05,
+      "loss": 3.8182,
+      "step": 6930
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.298850116547486e-05,
+      "loss": 3.7481,
+      "step": 6931
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.291448710962497e-05,
+      "loss": 3.7908,
+      "step": 6932
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.284055335118963e-05,
+      "loss": 3.8808,
+      "step": 6933
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.276669990287895e-05,
+      "loss": 3.8418,
+      "step": 6934
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.269292677738922e-05,
+      "loss": 3.8444,
+      "step": 6935
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.261923398740285e-05,
+      "loss": 3.8686,
+      "step": 6936
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.2545621545588436e-05,
+      "loss": 3.7808,
+      "step": 6937
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.247208946460084e-05,
+      "loss": 3.8508,
+      "step": 6938
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.2398637757081185e-05,
+      "loss": 3.8038,
+      "step": 6939
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.232526643565664e-05,
+      "loss": 3.8584,
+      "step": 6940
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.2251975512940514e-05,
+      "loss": 3.8037,
+      "step": 6941
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.2178765001532386e-05,
+      "loss": 3.7898,
+      "step": 6942
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.210563491401808e-05,
+      "loss": 3.7816,
+      "step": 6943
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.203258526296937e-05,
+      "loss": 3.8906,
+      "step": 6944
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.195961606094433e-05,
+      "loss": 3.62,
+      "step": 6945
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.188672732048731e-05,
+      "loss": 3.7553,
+      "step": 6946
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.181391905412867e-05,
+      "loss": 3.7777,
+      "step": 6947
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.174119127438507e-05,
+      "loss": 3.8042,
+      "step": 6948
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.166854399375929e-05,
+      "loss": 3.714,
+      "step": 6949
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.159597722474006e-05,
+      "loss": 3.6769,
+      "step": 6950
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.1523490979802614e-05,
+      "loss": 3.7914,
+      "step": 6951
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.145108527140805e-05,
+      "loss": 3.7098,
+      "step": 6952
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.1378760112003853e-05,
+      "loss": 3.7714,
+      "step": 6953
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.1306515514023527e-05,
+      "loss": 3.9572,
+      "step": 6954
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.1234351489886736e-05,
+      "loss": 3.8239,
+      "step": 6955
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.116226805199929e-05,
+      "loss": 3.8255,
+      "step": 6956
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.109026521275321e-05,
+      "loss": 3.7715,
+      "step": 6957
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.1018342984526604e-05,
+      "loss": 3.8107,
+      "step": 6958
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.0946501379683686e-05,
+      "loss": 3.8456,
+      "step": 6959
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.087474041057503e-05,
+      "loss": 3.896,
+      "step": 6960
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.080306008953704e-05,
+      "loss": 3.7103,
+      "step": 6961
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.073146042889241e-05,
+      "loss": 3.6669,
+      "step": 6962
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.065994144094997e-05,
+      "loss": 3.8413,
+      "step": 6963
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.05885031380046e-05,
+      "loss": 3.866,
+      "step": 6964
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.0517145532337476e-05,
+      "loss": 3.9093,
+      "step": 6965
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.0445868636215678e-05,
+      "loss": 3.9192,
+      "step": 6966
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.037467246189257e-05,
+      "loss": 3.8093,
+      "step": 6967
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.0303557021607754e-05,
+      "loss": 3.8139,
+      "step": 6968
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.0232522327586664e-05,
+      "loss": 3.6653,
+      "step": 6969
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.0161568392040982e-05,
+      "loss": 3.7888,
+      "step": 6970
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.0090695227168507e-05,
+      "loss": 3.8907,
+      "step": 6971
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.0019902845153214e-05,
+      "loss": 3.8344,
+      "step": 6972
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.9949191258164976e-05,
+      "loss": 3.8201,
+      "step": 6973
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.9878560478360183e-05,
+      "loss": 3.8492,
+      "step": 6974
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.9808010517880847e-05,
+      "loss": 3.9079,
+      "step": 6975
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.9737541388855536e-05,
+      "loss": 3.8074,
+      "step": 6976
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.966715310339857e-05,
+      "loss": 3.7786,
+      "step": 6977
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.9596845673610596e-05,
+      "loss": 3.8339,
+      "step": 6978
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.9526619111578222e-05,
+      "loss": 3.9025,
+      "step": 6979
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.945647342937413e-05,
+      "loss": 3.9123,
+      "step": 6980
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.9386408639057392e-05,
+      "loss": 3.7537,
+      "step": 6981
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.9316424752672767e-05,
+      "loss": 3.7725,
+      "step": 6982
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.92465217822514e-05,
+      "loss": 3.9438,
+      "step": 6983
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.9176699739810298e-05,
+      "loss": 3.9432,
+      "step": 6984
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.9106958637352797e-05,
+      "loss": 3.8651,
+      "step": 6985
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.90372984868682e-05,
+      "loss": 3.9016,
+      "step": 6986
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.8967719300331875e-05,
+      "loss": 3.855,
+      "step": 6987
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.889822108970519e-05,
+      "loss": 3.9016,
+      "step": 6988
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.8828803866935815e-05,
+      "loss": 3.8521,
+      "step": 6989
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.8759467643957372e-05,
+      "loss": 3.8662,
+      "step": 6990
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.8690212432689544e-05,
+      "loss": 3.7974,
+      "step": 6991
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.862103824503809e-05,
+      "loss": 3.9429,
+      "step": 6992
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.8551945092894825e-05,
+      "loss": 3.8761,
+      "step": 6993
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.8482932988137644e-05,
+      "loss": 3.6908,
+      "step": 6994
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.841400194263055e-05,
+      "loss": 3.798,
+      "step": 6995
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.8345151968223738e-05,
+      "loss": 3.6825,
+      "step": 6996
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.8276383076753175e-05,
+      "loss": 3.8067,
+      "step": 6997
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.8207695280041023e-05,
+      "loss": 3.8251,
+      "step": 6998
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.8139088589895557e-05,
+      "loss": 3.7108,
+      "step": 6999
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.8070563018111063e-05,
+      "loss": 3.7783,
+      "step": 7000
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.8002118576467785e-05,
+      "loss": 3.8134,
+      "step": 7001
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.7933755276732254e-05,
+      "loss": 3.9837,
+      "step": 7002
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.7865473130656793e-05,
+      "loss": 3.7795,
+      "step": 7003
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.779727214997996e-05,
+      "loss": 3.8896,
+      "step": 7004
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.7729152346426366e-05,
+      "loss": 3.7694,
+      "step": 7005
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.7661113731706488e-05,
+      "loss": 3.8344,
+      "step": 7006
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.7593156317516966e-05,
+      "loss": 3.7387,
+      "step": 7007
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.7525280115540506e-05,
+      "loss": 3.8348,
+      "step": 7008
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.7457485137445727e-05,
+      "loss": 3.6775,
+      "step": 7009
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.738977139488752e-05,
+      "loss": 3.7122,
+      "step": 7010
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.732213889950652e-05,
+      "loss": 3.772,
+      "step": 7011
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.7254587662929543e-05,
+      "loss": 3.8394,
+      "step": 7012
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.718711769676957e-05,
+      "loss": 3.861,
+      "step": 7013
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.7119729012625382e-05,
+      "loss": 3.6236,
+      "step": 7014
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.705242162208188e-05,
+      "loss": 3.8067,
+      "step": 7015
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.6985195536709918e-05,
+      "loss": 3.9207,
+      "step": 7016
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.6918050768066528e-05,
+      "loss": 3.7341,
+      "step": 7017
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.6850987327694698e-05,
+      "loss": 3.7325,
+      "step": 7018
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.678400522712332e-05,
+      "loss": 3.7557,
+      "step": 7019
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.671710447786746e-05,
+      "loss": 3.7653,
+      "step": 7020
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.665028509142803e-05,
+      "loss": 3.805,
+      "step": 7021
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.6583547079292224e-05,
+      "loss": 3.6582,
+      "step": 7022
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.6516890452932928e-05,
+      "loss": 3.6729,
+      "step": 7023
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.645031522380925e-05,
+      "loss": 3.912,
+      "step": 7024
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.6383821403366258e-05,
+      "loss": 3.8114,
+      "step": 7025
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.6317409003035032e-05,
+      "loss": 3.86,
+      "step": 7026
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.62510780342326e-05,
+      "loss": 3.8434,
+      "step": 7027
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.6184828508362013e-05,
+      "loss": 3.7776,
+      "step": 7028
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.6118660436812324e-05,
+      "loss": 3.7686,
+      "step": 7029
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.6052573830958547e-05,
+      "loss": 3.8366,
+      "step": 7030
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5986568702161818e-05,
+      "loss": 3.7322,
+      "step": 7031
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.592064506176922e-05,
+      "loss": 3.8073,
+      "step": 7032
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.585480292111375e-05,
+      "loss": 3.7921,
+      "step": 7033
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.578904229151441e-05,
+      "loss": 3.8649,
+      "step": 7034
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5723363184276206e-05,
+      "loss": 3.7012,
+      "step": 7035
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.565776561069022e-05,
+      "loss": 3.8398,
+      "step": 7036
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.559224958203321e-05,
+      "loss": 3.7084,
+      "step": 7037
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5526815109568447e-05,
+      "loss": 3.8007,
+      "step": 7038
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5461462204544704e-05,
+      "loss": 3.8486,
+      "step": 7039
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.539619087819689e-05,
+      "loss": 3.8531,
+      "step": 7040
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5331001141745968e-05,
+      "loss": 3.829,
+      "step": 7041
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.526589300639881e-05,
+      "loss": 3.7717,
+      "step": 7042
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.520086648334824e-05,
+      "loss": 3.8392,
+      "step": 7043
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5135921583773036e-05,
+      "loss": 3.8873,
+      "step": 7044
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5071058318837936e-05,
+      "loss": 3.805,
+      "step": 7045
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5006276699693852e-05,
+      "loss": 3.9205,
+      "step": 7046
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.4941576737477434e-05,
+      "loss": 3.6363,
+      "step": 7047
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.4876958443311282e-05,
+      "loss": 3.7754,
+      "step": 7048
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.481242182830401e-05,
+      "loss": 3.6932,
+      "step": 7049
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.4747966903550358e-05,
+      "loss": 3.6368,
+      "step": 7050
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.4683593680130732e-05,
+      "loss": 3.8372,
+      "step": 7051
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.461930216911168e-05,
+      "loss": 3.8614,
+      "step": 7052
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.455509238154574e-05,
+      "loss": 3.9,
+      "step": 7053
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.4490964328471254e-05,
+      "loss": 3.8306,
+      "step": 7054
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.442691802091257e-05,
+      "loss": 3.7787,
+      "step": 7055
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.4362953469879935e-05,
+      "loss": 3.8558,
+      "step": 7056
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.4299070686369717e-05,
+      "loss": 3.7247,
+      "step": 7057
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.423526968136397e-05,
+      "loss": 3.8126,
+      "step": 7058
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.417155046583097e-05,
+      "loss": 3.8714,
+      "step": 7059
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.4107913050724628e-05,
+      "loss": 3.8025,
+      "step": 7060
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.4044357446985133e-05,
+      "loss": 3.8742,
+      "step": 7061
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.3980883665538357e-05,
+      "loss": 3.8761,
+      "step": 7062
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.391749171729618e-05,
+      "loss": 3.7935,
+      "step": 7063
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.385418161315639e-05,
+      "loss": 3.7856,
+      "step": 7064
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.379095336400272e-05,
+      "loss": 3.7245,
+      "step": 7065
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.3727806980704758e-05,
+      "loss": 3.9594,
+      "step": 7066
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.3664742474118316e-05,
+      "loss": 3.8335,
+      "step": 7067
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.360175985508467e-05,
+      "loss": 3.7729,
+      "step": 7068
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.3538859134431546e-05,
+      "loss": 3.643,
+      "step": 7069
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.347604032297207e-05,
+      "loss": 3.6751,
+      "step": 7070
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.3413303431505607e-05,
+      "loss": 3.8172,
+      "step": 7071
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.3350648470817413e-05,
+      "loss": 3.7264,
+      "step": 7072
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.328807545167838e-05,
+      "loss": 3.8558,
+      "step": 7073
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.3225584384845845e-05,
+      "loss": 3.8773,
+      "step": 7074
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.3163175281062544e-05,
+      "loss": 3.8858,
+      "step": 7075
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.3100848151057395e-05,
+      "loss": 3.7801,
+      "step": 7076
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.3038603005545113e-05,
+      "loss": 3.8107,
+      "step": 7077
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.2976439855226404e-05,
+      "loss": 3.8369,
+      "step": 7078
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.291435871078784e-05,
+      "loss": 3.8547,
+      "step": 7079
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.2852359582901826e-05,
+      "loss": 3.8176,
+      "step": 7080
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.2790442482226725e-05,
+      "loss": 3.7504,
+      "step": 7081
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.2728607419406965e-05,
+      "loss": 3.8981,
+      "step": 7082
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.2666854405072547e-05,
+      "loss": 3.6287,
+      "step": 7083
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.2605183449839583e-05,
+      "loss": 3.8164,
+      "step": 7084
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.2543594564309988e-05,
+      "loss": 3.7554,
+      "step": 7085
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.2482087759071622e-05,
+      "loss": 3.7553,
+      "step": 7086
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.2420663044698253e-05,
+      "loss": 3.8419,
+      "step": 7087
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.235932043174943e-05,
+      "loss": 3.7182,
+      "step": 7088
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.229805993077083e-05,
+      "loss": 3.6902,
+      "step": 7089
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.2236881552293642e-05,
+      "loss": 3.7585,
+      "step": 7090
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.2175785306835284e-05,
+      "loss": 3.7429,
+      "step": 7091
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.21147712048988e-05,
+      "loss": 3.6429,
+      "step": 7092
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.2053839256973295e-05,
+      "loss": 3.7879,
+      "step": 7093
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.1992989473533563e-05,
+      "loss": 3.8302,
+      "step": 7094
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.193222186504057e-05,
+      "loss": 3.653,
+      "step": 7095
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.1871536441940786e-05,
+      "loss": 3.767,
+      "step": 7096
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.181093321466693e-05,
+      "loss": 3.8467,
+      "step": 7097
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.1750412193637214e-05,
+      "loss": 3.7881,
+      "step": 7098
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.1689973389256044e-05,
+      "loss": 3.8114,
+      "step": 7099
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.16296168119135e-05,
+      "loss": 3.8007,
+      "step": 7100
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.1569342471985554e-05,
+      "loss": 3.883,
+      "step": 7101
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.150915037983403e-05,
+      "loss": 3.8054,
+      "step": 7102
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.1449040545806764e-05,
+      "loss": 3.8936,
+      "step": 7103
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.1389012980237265e-05,
+      "loss": 3.7102,
+      "step": 7104
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.1329067693444893e-05,
+      "loss": 3.7225,
+      "step": 7105
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.1269204695735066e-05,
+      "loss": 3.7089,
+      "step": 7106
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.120942399739889e-05,
+      "loss": 3.717,
+      "step": 7107
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.1149725608713367e-05,
+      "loss": 3.7829,
+      "step": 7108
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.1090109539941228e-05,
+      "loss": 3.9982,
+      "step": 7109
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.1030575801331332e-05,
+      "loss": 3.7756,
+      "step": 7110
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.0971124403118157e-05,
+      "loss": 3.8154,
+      "step": 7111
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.091175535552209e-05,
+      "loss": 3.7576,
+      "step": 7112
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.0852468668749293e-05,
+      "loss": 3.8307,
+      "step": 7113
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.079326435299189e-05,
+      "loss": 3.7223,
+      "step": 7114
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.0734142418427805e-05,
+      "loss": 3.9134,
+      "step": 7115
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.067510287522073e-05,
+      "loss": 3.8638,
+      "step": 7116
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.0616145733520274e-05,
+      "loss": 3.7773,
+      "step": 7117
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.055727100346194e-05,
+      "loss": 3.8967,
+      "step": 7118
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.0498478695166855e-05,
+      "loss": 3.8216,
+      "step": 7119
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.0439768818742155e-05,
+      "loss": 3.7695,
+      "step": 7120
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.038114138428071e-05,
+      "loss": 3.8265,
+      "step": 7121
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.0322596401861294e-05,
+      "loss": 3.8772,
+      "step": 7122
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.0264133881548406e-05,
+      "loss": 3.7554,
+      "step": 7123
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.020575383339246e-05,
+      "loss": 3.7723,
+      "step": 7124
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.0147456267429755e-05,
+      "loss": 3.8539,
+      "step": 7125
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.0089241193682272e-05,
+      "loss": 3.6777,
+      "step": 7126
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.003110862215779e-05,
+      "loss": 3.8839,
+      "step": 7127
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.997305856285003e-05,
+      "loss": 3.9525,
+      "step": 7128
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9915091025738462e-05,
+      "loss": 3.7727,
+      "step": 7129
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9857206020788276e-05,
+      "loss": 3.7993,
+      "step": 7130
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.979940355795079e-05,
+      "loss": 3.7256,
+      "step": 7131
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9741683647162724e-05,
+      "loss": 3.6828,
+      "step": 7132
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9684046298346857e-05,
+      "loss": 3.8835,
+      "step": 7133
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9626491521411772e-05,
+      "loss": 3.8479,
+      "step": 7134
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.956901932625177e-05,
+      "loss": 3.8737,
+      "step": 7135
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9511629722747004e-05,
+      "loss": 3.629,
+      "step": 7136
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9454322720763362e-05,
+      "loss": 3.8479,
+      "step": 7137
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.939709833015252e-05,
+      "loss": 3.796,
+      "step": 7138
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9339956560752214e-05,
+      "loss": 3.7162,
+      "step": 7139
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9282897422385593e-05,
+      "loss": 3.7033,
+      "step": 7140
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.922592092486192e-05,
+      "loss": 3.7538,
+      "step": 7141
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9169027077975965e-05,
+      "loss": 3.8172,
+      "step": 7142
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9112215891508632e-05,
+      "loss": 3.8422,
+      "step": 7143
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9055487375226277e-05,
+      "loss": 3.8534,
+      "step": 7144
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.8998841538881152e-05,
+      "loss": 3.736,
+      "step": 7145
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.8942278392211464e-05,
+      "loss": 3.9243,
+      "step": 7146
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.8885797944941052e-05,
+      "loss": 3.7149,
+      "step": 7147
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.8829400206779536e-05,
+      "loss": 3.7337,
+      "step": 7148
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8773085187422323e-05,
+      "loss": 3.8267,
+      "step": 7149
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8716852896550618e-05,
+      "loss": 3.695,
+      "step": 7150
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.866070334383135e-05,
+      "loss": 3.8048,
+      "step": 7151
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8604636538917364e-05,
+      "loss": 3.7349,
+      "step": 7152
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8548652491447214e-05,
+      "loss": 3.764,
+      "step": 7153
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8492751211045157e-05,
+      "loss": 3.9766,
+      "step": 7154
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8436932707321273e-05,
+      "loss": 3.7587,
+      "step": 7155
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8381196989871452e-05,
+      "loss": 3.6994,
+      "step": 7156
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8325544068277245e-05,
+      "loss": 3.8896,
+      "step": 7157
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8269973952106055e-05,
+      "loss": 3.6652,
+      "step": 7158
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.821448665091102e-05,
+      "loss": 3.8339,
+      "step": 7159
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8159082174231012e-05,
+      "loss": 3.8253,
+      "step": 7160
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8103760531590852e-05,
+      "loss": 3.7553,
+      "step": 7161
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8048521732500878e-05,
+      "loss": 3.7805,
+      "step": 7162
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7993365786457217e-05,
+      "loss": 3.5768,
+      "step": 7163
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.793829270294195e-05,
+      "loss": 3.8284,
+      "step": 7164
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7883302491422672e-05,
+      "loss": 3.7233,
+      "step": 7165
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7828395161352817e-05,
+      "loss": 3.7828,
+      "step": 7166
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.777357072217173e-05,
+      "loss": 3.7461,
+      "step": 7167
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7718829183304252e-05,
+      "loss": 3.6646,
+      "step": 7168
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7664170554161142e-05,
+      "loss": 3.8278,
+      "step": 7169
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7609594844138766e-05,
+      "loss": 3.8451,
+      "step": 7170
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7555102062619455e-05,
+      "loss": 3.6435,
+      "step": 7171
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.750069221897105e-05,
+      "loss": 3.7927,
+      "step": 7172
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.744636532254723e-05,
+      "loss": 3.8769,
+      "step": 7173
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.739212138268753e-05,
+      "loss": 3.7451,
+      "step": 7174
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.733796040871699e-05,
+      "loss": 4.0483,
+      "step": 7175
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7283882409946548e-05,
+      "loss": 3.7627,
+      "step": 7176
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.722988739567288e-05,
+      "loss": 3.8554,
+      "step": 7177
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7175975375178344e-05,
+      "loss": 3.8798,
+      "step": 7178
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7122146357730907e-05,
+      "loss": 3.8484,
+      "step": 7179
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.706840035258461e-05,
+      "loss": 3.8273,
+      "step": 7180
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7014737368978793e-05,
+      "loss": 3.7267,
+      "step": 7181
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.696115741613902e-05,
+      "loss": 3.8759,
+      "step": 7182
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.690766050327608e-05,
+      "loss": 3.9346,
+      "step": 7183
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.685424663958679e-05,
+      "loss": 3.889,
+      "step": 7184
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6800915834253637e-05,
+      "loss": 3.7038,
+      "step": 7185
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.674766809644479e-05,
+      "loss": 3.8143,
+      "step": 7186
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6694503435314036e-05,
+      "loss": 3.8697,
+      "step": 7187
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.664142186000117e-05,
+      "loss": 3.8359,
+      "step": 7188
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6588423379631458e-05,
+      "loss": 3.8145,
+      "step": 7189
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6535508003315937e-05,
+      "loss": 3.7164,
+      "step": 7190
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6482675740151442e-05,
+      "loss": 3.801,
+      "step": 7191
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6429926599220378e-05,
+      "loss": 3.9192,
+      "step": 7192
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6377260589590938e-05,
+      "loss": 3.6931,
+      "step": 7193
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.632467772031704e-05,
+      "loss": 3.8325,
+      "step": 7194
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.627217800043829e-05,
+      "loss": 3.8515,
+      "step": 7195
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.621976143898002e-05,
+      "loss": 3.8185,
+      "step": 7196
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6167428044953137e-05,
+      "loss": 3.7885,
+      "step": 7197
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6115177827354554e-05,
+      "loss": 3.663,
+      "step": 7198
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6063010795166534e-05,
+      "loss": 3.7974,
+      "step": 7199
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.601092695735723e-05,
+      "loss": 3.7937,
+      "step": 7200
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.5958926322880484e-05,
+      "loss": 3.6004,
+      "step": 7201
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.5907008900675758e-05,
+      "loss": 3.8008,
+      "step": 7202
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.5855174699668296e-05,
+      "loss": 3.775,
+      "step": 7203
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.580342372876903e-05,
+      "loss": 3.7558,
+      "step": 7204
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.5751755996874452e-05,
+      "loss": 3.7339,
+      "step": 7205
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.5700171512866957e-05,
+      "loss": 3.7154,
+      "step": 7206
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.5648670285614396e-05,
+      "loss": 3.798,
+      "step": 7207
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.559725232397058e-05,
+      "loss": 3.7928,
+      "step": 7208
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.5545917636774653e-05,
+      "loss": 3.8784,
+      "step": 7209
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.5494666232851894e-05,
+      "loss": 3.8271,
+      "step": 7210
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.5443498121012812e-05,
+      "loss": 3.631,
+      "step": 7211
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.5392413310053867e-05,
+      "loss": 3.8509,
+      "step": 7212
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.5341411808757143e-05,
+      "loss": 3.7593,
+      "step": 7213
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.529049362589041e-05,
+      "loss": 3.7816,
+      "step": 7214
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.5239658770206943e-05,
+      "loss": 3.6661,
+      "step": 7215
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.5188907250446027e-05,
+      "loss": 3.6987,
+      "step": 7216
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.513823907533235e-05,
+      "loss": 3.7735,
+      "step": 7217
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.5087654253576388e-05,
+      "loss": 3.7977,
+      "step": 7218
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.5037152793874243e-05,
+      "loss": 3.6281,
+      "step": 7219
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.4986734704907745e-05,
+      "loss": 3.7823,
+      "step": 7220
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.4936399995344297e-05,
+      "loss": 3.7365,
+      "step": 7221
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.4886148673836974e-05,
+      "loss": 3.7933,
+      "step": 7222
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.4835980749024591e-05,
+      "loss": 3.7907,
+      "step": 7223
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.4785896229531693e-05,
+      "loss": 3.7264,
+      "step": 7224
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.473589512396828e-05,
+      "loss": 3.816,
+      "step": 7225
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.4685977440930199e-05,
+      "loss": 3.7787,
+      "step": 7226
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4636143188998807e-05,
+      "loss": 3.8738,
+      "step": 7227
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4586392376741254e-05,
+      "loss": 3.8539,
+      "step": 7228
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.453672501271025e-05,
+      "loss": 3.8704,
+      "step": 7229
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4487141105444134e-05,
+      "loss": 3.6897,
+      "step": 7230
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4437640663467033e-05,
+      "loss": 3.7701,
+      "step": 7231
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4388223695288694e-05,
+      "loss": 3.7106,
+      "step": 7232
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.433889020940432e-05,
+      "loss": 3.7527,
+      "step": 7233
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4289640214294964e-05,
+      "loss": 3.8508,
+      "step": 7234
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4240473718427348e-05,
+      "loss": 3.8707,
+      "step": 7235
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4191390730253716e-05,
+      "loss": 3.6981,
+      "step": 7236
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4142391258211984e-05,
+      "loss": 3.7346,
+      "step": 7237
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.409347531072569e-05,
+      "loss": 3.903,
+      "step": 7238
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4044642896204107e-05,
+      "loss": 3.7993,
+      "step": 7239
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3995894023042133e-05,
+      "loss": 3.8858,
+      "step": 7240
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.394722869962023e-05,
+      "loss": 3.7318,
+      "step": 7241
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3898646934304538e-05,
+      "loss": 3.7294,
+      "step": 7242
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3850148735446766e-05,
+      "loss": 3.7508,
+      "step": 7243
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.38017341113843e-05,
+      "loss": 3.8164,
+      "step": 7244
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3753403070440262e-05,
+      "loss": 3.6836,
+      "step": 7245
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3705155620923337e-05,
+      "loss": 3.9341,
+      "step": 7246
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.365699177112778e-05,
+      "loss": 3.7545,
+      "step": 7247
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3608911529333467e-05,
+      "loss": 3.9702,
+      "step": 7248
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3560914903806065e-05,
+      "loss": 3.7343,
+      "step": 7249
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.351300190279664e-05,
+      "loss": 3.8163,
+      "step": 7250
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3465172534541937e-05,
+      "loss": 3.8751,
+      "step": 7251
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3417426807264599e-05,
+      "loss": 3.8698,
+      "step": 7252
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3369764729172451e-05,
+      "loss": 3.7181,
+      "step": 7253
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3322186308459271e-05,
+      "loss": 3.7949,
+      "step": 7254
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3274691553304353e-05,
+      "loss": 3.903,
+      "step": 7255
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.322728047187255e-05,
+      "loss": 3.9686,
+      "step": 7256
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3179953072314398e-05,
+      "loss": 3.7838,
+      "step": 7257
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3132709362766004e-05,
+      "loss": 3.7968,
+      "step": 7258
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3085549351349091e-05,
+      "loss": 3.6666,
+      "step": 7259
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3038473046171061e-05,
+      "loss": 3.8752,
+      "step": 7260
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.299148045532489e-05,
+      "loss": 3.8742,
+      "step": 7261
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2944571586888997e-05,
+      "loss": 3.7496,
+      "step": 7262
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2897746448927827e-05,
+      "loss": 3.7841,
+      "step": 7263
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2851005049490939e-05,
+      "loss": 3.6499,
+      "step": 7264
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2804347396613847e-05,
+      "loss": 3.8972,
+      "step": 7265
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2757773498317416e-05,
+      "loss": 3.737,
+      "step": 7266
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.271128336260835e-05,
+      "loss": 3.9043,
+      "step": 7267
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.266487699747887e-05,
+      "loss": 3.8228,
+      "step": 7268
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2618554410906646e-05,
+      "loss": 3.7682,
+      "step": 7269
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.25723156108552e-05,
+      "loss": 3.8609,
+      "step": 7270
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.252616060527334e-05,
+      "loss": 3.7509,
+      "step": 7271
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.248008940209583e-05,
+      "loss": 3.8726,
+      "step": 7272
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.243410200924272e-05,
+      "loss": 3.8899,
+      "step": 7273
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2388198434619802e-05,
+      "loss": 3.7748,
+      "step": 7274
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2342378686118538e-05,
+      "loss": 3.8552,
+      "step": 7275
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.229664277161574e-05,
+      "loss": 3.8794,
+      "step": 7276
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2250990698974007e-05,
+      "loss": 3.8496,
+      "step": 7277
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.220542247604145e-05,
+      "loss": 3.598,
+      "step": 7278
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2159938110651803e-05,
+      "loss": 3.7053,
+      "step": 7279
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2114537610624254e-05,
+      "loss": 3.6224,
+      "step": 7280
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2069220983763729e-05,
+      "loss": 3.9039,
+      "step": 7281
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2023988237860717e-05,
+      "loss": 3.8237,
+      "step": 7282
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1978839380691275e-05,
+      "loss": 3.7797,
+      "step": 7283
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1933774420016974e-05,
+      "loss": 3.7701,
+      "step": 7284
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1888793363584948e-05,
+      "loss": 3.8655,
+      "step": 7285
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.184389621912807e-05,
+      "loss": 3.8698,
+      "step": 7286
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1799082994364551e-05,
+      "loss": 3.7368,
+      "step": 7287
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.17543536969984e-05,
+      "loss": 3.8372,
+      "step": 7288
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1709708334719127e-05,
+      "loss": 3.8014,
+      "step": 7289
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1665146915201653e-05,
+      "loss": 3.8023,
+      "step": 7290
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1620669446106735e-05,
+      "loss": 3.8892,
+      "step": 7291
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1576275935080538e-05,
+      "loss": 3.7322,
+      "step": 7292
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1531966389754788e-05,
+      "loss": 3.7837,
+      "step": 7293
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1487740817746839e-05,
+      "loss": 3.7549,
+      "step": 7294
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1443599226659495e-05,
+      "loss": 3.6735,
+      "step": 7295
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1399541624081355e-05,
+      "loss": 3.7705,
+      "step": 7296
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1355568017586304e-05,
+      "loss": 3.8317,
+      "step": 7297
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1311678414734016e-05,
+      "loss": 3.7673,
+      "step": 7298
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1267872823069458e-05,
+      "loss": 3.7504,
+      "step": 7299
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1224151250123548e-05,
+      "loss": 3.6638,
+      "step": 7300
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1180513703412387e-05,
+      "loss": 3.8792,
+      "step": 7301
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1136960190437751e-05,
+      "loss": 3.7737,
+      "step": 7302
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1093490718687094e-05,
+      "loss": 3.7238,
+      "step": 7303
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1050105295633272e-05,
+      "loss": 3.9852,
+      "step": 7304
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.100680392873471e-05,
+      "loss": 3.998,
+      "step": 7305
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0963586625435506e-05,
+      "loss": 3.829,
+      "step": 7306
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.092045339316511e-05,
+      "loss": 3.8182,
+      "step": 7307
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.087740423933864e-05,
+      "loss": 3.7914,
+      "step": 7308
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0834439171356846e-05,
+      "loss": 3.8831,
+      "step": 7309
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0791558196605822e-05,
+      "loss": 3.8871,
+      "step": 7310
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0748761322457335e-05,
+      "loss": 3.7263,
+      "step": 7311
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0706048556268666e-05,
+      "loss": 3.8484,
+      "step": 7312
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0663419905382666e-05,
+      "loss": 3.7053,
+      "step": 7313
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0620875377127637e-05,
+      "loss": 4.0379,
+      "step": 7314
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0578414978817508e-05,
+      "loss": 3.7817,
+      "step": 7315
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0536038717751606e-05,
+      "loss": 3.8642,
+      "step": 7316
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0493746601215104e-05,
+      "loss": 3.8244,
+      "step": 7317
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0451538636478353e-05,
+      "loss": 3.8031,
+      "step": 7318
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0409414830797493e-05,
+      "loss": 3.7289,
+      "step": 7319
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0367375191414064e-05,
+      "loss": 3.7549,
+      "step": 7320
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0325419725555119e-05,
+      "loss": 3.731,
+      "step": 7321
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0283548440433333e-05,
+      "loss": 3.8508,
+      "step": 7322
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.024176134324678e-05,
+      "loss": 3.7441,
+      "step": 7323
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0200058441179271e-05,
+      "loss": 3.9404,
+      "step": 7324
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.015843974139996e-05,
+      "loss": 3.8838,
+      "step": 7325
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0116905251063623e-05,
+      "loss": 3.8335,
+      "step": 7326
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0075454977310383e-05,
+      "loss": 3.9023,
+      "step": 7327
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0034088927266205e-05,
+      "loss": 3.8361,
+      "step": 7328
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.99280710804229e-06,
+      "loss": 3.7389,
+      "step": 7329
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.951609526735517e-06,
+      "loss": 3.9196,
+      "step": 7330
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.910496190428163e-06,
+      "loss": 3.7724,
+      "step": 7331
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.869467106188135e-06,
+      "loss": 3.895,
+      "step": 7332
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.828522281068786e-06,
+      "loss": 3.8121,
+      "step": 7333
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.787661722108987e-06,
+      "loss": 3.8829,
+      "step": 7334
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.746885436333231e-06,
+      "loss": 3.7296,
+      "step": 7335
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.706193430751298e-06,
+      "loss": 3.8254,
+      "step": 7336
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.665585712358705e-06,
+      "loss": 3.9892,
+      "step": 7337
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.625062288136366e-06,
+      "loss": 3.8141,
+      "step": 7338
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.584623165050766e-06,
+      "loss": 3.7971,
+      "step": 7339
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.544268350053842e-06,
+      "loss": 3.8789,
+      "step": 7340
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.50399785008299e-06,
+      "loss": 3.7591,
+      "step": 7341
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.463811672061284e-06,
+      "loss": 3.7637,
+      "step": 7342
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.423709822897086e-06,
+      "loss": 3.8335,
+      "step": 7343
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.383692309484381e-06,
+      "loss": 3.8673,
+      "step": 7344
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.343759138702724e-06,
+      "loss": 3.5563,
+      "step": 7345
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.303910317417064e-06,
+      "loss": 3.7439,
+      "step": 7346
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.264145852477868e-06,
+      "loss": 3.847,
+      "step": 7347
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.224465750721056e-06,
+      "loss": 3.7257,
+      "step": 7348
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.184870018968172e-06,
+      "loss": 3.9923,
+      "step": 7349
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.14535866402616e-06,
+      "loss": 3.9748,
+      "step": 7350
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.105931692687419e-06,
+      "loss": 3.8604,
+      "step": 7351
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.066589111729972e-06,
+      "loss": 3.9302,
+      "step": 7352
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.027330927917243e-06,
+      "loss": 3.6989,
+      "step": 7353
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.988157147998222e-06,
+      "loss": 3.8057,
+      "step": 7354
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.949067778707187e-06,
+      "loss": 3.7271,
+      "step": 7355
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.910062826764208e-06,
+      "loss": 3.8246,
+      "step": 7356
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.871142298874646e-06,
+      "loss": 3.788,
+      "step": 7357
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.832306201729368e-06,
+      "loss": 3.8303,
+      "step": 7358
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.793554542004756e-06,
+      "loss": 3.7367,
+      "step": 7359
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.754887326362648e-06,
+      "loss": 3.9724,
+      "step": 7360
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.71630456145045e-06,
+      "loss": 3.8364,
+      "step": 7361
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.677806253900967e-06,
+      "loss": 3.745,
+      "step": 7362
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.639392410332403e-06,
+      "loss": 3.6978,
+      "step": 7363
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.601063037348644e-06,
+      "loss": 3.8222,
+      "step": 7364
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.562818141538975e-06,
+      "loss": 3.8391,
+      "step": 7365
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.52465772947808e-06,
+      "loss": 3.8409,
+      "step": 7366
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.486581807726157e-06,
+      "loss": 3.7704,
+      "step": 7367
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.448590382829024e-06,
+      "loss": 3.7558,
+      "step": 7368
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.410683461317681e-06,
+      "loss": 3.7473,
+      "step": 7369
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.372861049708858e-06,
+      "loss": 3.8308,
+      "step": 7370
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.335123154504687e-06,
+      "loss": 3.7776,
+      "step": 7371
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.297469782192702e-06,
+      "loss": 3.8127,
+      "step": 7372
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.25990093924589e-06,
+      "loss": 3.9758,
+      "step": 7373
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.222416632122865e-06,
+      "loss": 3.7299,
+      "step": 7374
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.185016867267691e-06,
+      "loss": 3.7018,
+      "step": 7375
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.147701651109674e-06,
+      "loss": 3.859,
+      "step": 7376
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.110470990063789e-06,
+      "loss": 3.6717,
+      "step": 7377
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.07332489053042e-06,
+      "loss": 3.8828,
+      "step": 7378
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.036263358895402e-06,
+      "loss": 3.7459,
+      "step": 7379
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 7.99928640152997e-06,
+      "loss": 3.9514,
+      "step": 7380
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 7.962394024791042e-06,
+      "loss": 3.8069,
+      "step": 7381
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 7.925586235020766e-06,
+      "loss": 3.7751,
+      "step": 7382
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.888863038546801e-06,
+      "loss": 3.8605,
+      "step": 7383
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.852224441682376e-06,
+      "loss": 3.8498,
+      "step": 7384
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.81567045072601e-06,
+      "loss": 3.7279,
+      "step": 7385
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.779201071961784e-06,
+      "loss": 3.9288,
+      "step": 7386
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.742816311659185e-06,
+      "loss": 3.771,
+      "step": 7387
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.706516176073208e-06,
+      "loss": 3.8154,
+      "step": 7388
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.670300671444308e-06,
+      "loss": 3.7109,
+      "step": 7389
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.634169803998226e-06,
+      "loss": 3.8252,
+      "step": 7390
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.598123579946381e-06,
+      "loss": 3.7613,
+      "step": 7391
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.562162005485485e-06,
+      "loss": 3.9197,
+      "step": 7392
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.5262850867978124e-06,
+      "loss": 3.855,
+      "step": 7393
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.490492830050932e-06,
+      "loss": 3.9321,
+      "step": 7394
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.454785241398032e-06,
+      "loss": 3.7379,
+      "step": 7395
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.419162326977591e-06,
+      "loss": 3.6822,
+      "step": 7396
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.383624092913654e-06,
+      "loss": 3.7078,
+      "step": 7397
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.348170545315613e-06,
+      "loss": 3.7863,
+      "step": 7398
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.312801690278425e-06,
+      "loss": 3.7272,
+      "step": 7399
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.277517533882283e-06,
+      "loss": 3.7928,
+      "step": 7400
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.242318082193e-06,
+      "loss": 3.8874,
+      "step": 7401
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.207203341261792e-06,
+      "loss": 3.9057,
+      "step": 7402
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.172173317125275e-06,
+      "loss": 3.8317,
+      "step": 7403
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.1372280158055196e-06,
+      "loss": 3.7079,
+      "step": 7404
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.102367443310054e-06,
+      "loss": 3.7499,
+      "step": 7405
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.067591605631751e-06,
+      "loss": 3.6975,
+      "step": 7406
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.032900508749051e-06,
+      "loss": 3.7171,
+      "step": 7407
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.998294158625684e-06,
+      "loss": 3.8113,
+      "step": 7408
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.96377256121089e-06,
+      "loss": 3.8072,
+      "step": 7409
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.9293357224393675e-06,
+      "loss": 3.8345,
+      "step": 7410
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.894983648231212e-06,
+      "loss": 3.8716,
+      "step": 7411
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.86071634449198e-06,
+      "loss": 3.856,
+      "step": 7412
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.826533817112512e-06,
+      "loss": 3.9106,
+      "step": 7413
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.792436071969277e-06,
+      "loss": 3.8332,
+      "step": 7414
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.758423114924028e-06,
+      "loss": 3.8482,
+      "step": 7415
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.724494951823978e-06,
+      "loss": 3.7252,
+      "step": 7416
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.6906515885017945e-06,
+      "loss": 3.8092,
+      "step": 7417
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.656893030775546e-06,
+      "loss": 3.7595,
+      "step": 7418
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.623219284448645e-06,
+      "loss": 3.6624,
+      "step": 7419
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.589630355310128e-06,
+      "loss": 3.7314,
+      "step": 7420
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.556126249134209e-06,
+      "loss": 3.7689,
+      "step": 7421
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.522706971680725e-06,
+      "loss": 3.8921,
+      "step": 7422
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.489372528694748e-06,
+      "loss": 3.8321,
+      "step": 7423
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.456122925906971e-06,
+      "loss": 3.9569,
+      "step": 7424
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.422958169033266e-06,
+      "loss": 3.9405,
+      "step": 7425
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.389878263775128e-06,
+      "loss": 3.6667,
+      "step": 7426
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.3568832158192865e-06,
+      "loss": 3.8271,
+      "step": 7427
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.323973030838037e-06,
+      "loss": 3.8337,
+      "step": 7428
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.291147714488965e-06,
+      "loss": 3.9602,
+      "step": 7429
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.258407272415223e-06,
+      "loss": 3.7543,
+      "step": 7430
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.2257517102451975e-06,
+      "loss": 3.7151,
+      "step": 7431
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.193181033592787e-06,
+      "loss": 3.7938,
+      "step": 7432
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.1606952480572356e-06,
+      "loss": 3.702,
+      "step": 7433
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.128294359223296e-06,
+      "loss": 3.7267,
+      "step": 7434
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.09597837266096e-06,
+      "loss": 3.8024,
+      "step": 7435
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.063747293925781e-06,
+      "loss": 3.7049,
+      "step": 7436
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.031601128558606e-06,
+      "loss": 3.7536,
+      "step": 7437
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.999539882085791e-06,
+      "loss": 3.6845,
+      "step": 7438
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.967563560019041e-06,
+      "loss": 3.9868,
+      "step": 7439
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.935672167855455e-06,
+      "loss": 3.7129,
+      "step": 7440
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.903865711077483e-06,
+      "loss": 3.9657,
+      "step": 7441
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.872144195153029e-06,
+      "loss": 3.8718,
+      "step": 7442
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.840507625535396e-06,
+      "loss": 3.8515,
+      "step": 7443
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.808956007663291e-06,
+      "loss": 3.8235,
+      "step": 7444
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.777489346960874e-06,
+      "loss": 3.757,
+      "step": 7445
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.746107648837484e-06,
+      "loss": 3.7154,
+      "step": 7446
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.714810918688085e-06,
+      "loss": 3.7274,
+      "step": 7447
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.683599161892928e-06,
+      "loss": 3.7219,
+      "step": 7448
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.65247238381772e-06,
+      "loss": 3.7309,
+      "step": 7449
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.621430589813459e-06,
+      "loss": 3.9509,
+      "step": 7450
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.590473785216654e-06,
+      "loss": 3.6585,
+      "step": 7451
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.559601975348994e-06,
+      "loss": 3.7677,
+      "step": 7452
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.528815165517842e-06,
+      "loss": 3.7651,
+      "step": 7453
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.498113361015799e-06,
+      "loss": 3.7969,
+      "step": 7454
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.467496567120811e-06,
+      "loss": 3.786,
+      "step": 7455
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.436964789096221e-06,
+      "loss": 3.8254,
+      "step": 7456
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.406518032190943e-06,
+      "loss": 3.8139,
+      "step": 7457
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.376156301638957e-06,
+      "loss": 3.7996,
+      "step": 7458
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.34587960265992e-06,
+      "loss": 3.8023,
+      "step": 7459
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.315687940458669e-06,
+      "loss": 3.7379,
+      "step": 7460
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.285581320225552e-06,
+      "loss": 3.7327,
+      "step": 7461
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.255559747136263e-06,
+      "loss": 3.8424,
+      "step": 7462
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.2256232263517835e-06,
+      "loss": 3.8129,
+      "step": 7463
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.195771763018609e-06,
+      "loss": 3.8573,
+      "step": 7464
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.166005362268467e-06,
+      "loss": 3.7636,
+      "step": 7465
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.136324029218653e-06,
+      "loss": 3.9288,
+      "step": 7466
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.106727768971697e-06,
+      "loss": 3.72,
+      "step": 7467
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.07721658661553e-06,
+      "loss": 3.8254,
+      "step": 7468
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.04779048722348e-06,
+      "loss": 3.9064,
+      "step": 7469
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.018449475854225e-06,
+      "loss": 3.864,
+      "step": 7470
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.989193557551786e-06,
+      "loss": 3.8502,
+      "step": 7471
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.960022737345693e-06,
+      "loss": 3.9497,
+      "step": 7472
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.930937020250603e-06,
+      "loss": 3.8229,
+      "step": 7473
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.90193641126685e-06,
+      "loss": 3.8329,
+      "step": 7474
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.873020915379834e-06,
+      "loss": 3.7584,
+      "step": 7475
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.8441905375605774e-06,
+      "loss": 3.977,
+      "step": 7476
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.815445282765285e-06,
+      "loss": 3.7937,
+      "step": 7477
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.786785155935669e-06,
+      "loss": 3.7791,
+      "step": 7478
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.758210161998622e-06,
+      "loss": 3.8569,
+      "step": 7479
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.729720305866603e-06,
+      "loss": 3.7275,
+      "step": 7480
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.70131559243736e-06,
+      "loss": 3.8203,
+      "step": 7481
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.672996026593934e-06,
+      "loss": 3.7333,
+      "step": 7482
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.644761613204817e-06,
+      "loss": 3.7004,
+      "step": 7483
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.616612357123795e-06,
+      "loss": 3.8567,
+      "step": 7484
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.588548263190107e-06,
+      "loss": 3.8021,
+      "step": 7485
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.560569336228337e-06,
+      "loss": 3.6639,
+      "step": 7486
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.532675581048251e-06,
+      "loss": 3.8548,
+      "step": 7487
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.504867002445179e-06,
+      "loss": 3.7422,
+      "step": 7488
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.477143605199796e-06,
+      "loss": 3.8335,
+      "step": 7489
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.449505394078013e-06,
+      "loss": 3.7994,
+      "step": 7490
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.421952373831139e-06,
+      "loss": 3.7116,
+      "step": 7491
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.394484549195887e-06,
+      "loss": 3.8948,
+      "step": 7492
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.3671019248943125e-06,
+      "loss": 3.9758,
+      "step": 7493
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.3398045056337594e-06,
+      "loss": 3.7589,
+      "step": 7494
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.312592296106976e-06,
+      "loss": 3.9307,
+      "step": 7495
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.285465300992164e-06,
+      "loss": 3.7721,
+      "step": 7496
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.258423524952649e-06,
+      "loss": 3.7641,
+      "step": 7497
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.23146697263721e-06,
+      "loss": 3.74,
+      "step": 7498
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.204595648680088e-06,
+      "loss": 3.8135,
+      "step": 7499
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.177809557700751e-06,
+      "loss": 3.8243,
+      "step": 7500
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.151108704303963e-06,
+      "loss": 3.8067,
+      "step": 7501
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.12449309307994e-06,
+      "loss": 3.8561,
+      "step": 7502
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.097962728604299e-06,
+      "loss": 3.7785,
+      "step": 7503
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.071517615437836e-06,
+      "loss": 3.8636,
+      "step": 7504
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.045157758126749e-06,
+      "loss": 3.7444,
+      "step": 7505
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.018883161202691e-06,
+      "loss": 4.0491,
+      "step": 7506
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.992693829182548e-06,
+      "loss": 3.7428,
+      "step": 7507
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.966589766568496e-06,
+      "loss": 3.8467,
+      "step": 7508
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.940570977848168e-06,
+      "loss": 3.954,
+      "step": 7509
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.914637467494542e-06,
+      "loss": 3.8961,
+      "step": 7510
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.888789239965884e-06,
+      "loss": 3.716,
+      "step": 7511
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.863026299705696e-06,
+      "loss": 3.6933,
+      "step": 7512
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.837348651142991e-06,
+      "loss": 3.7382,
+      "step": 7513
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.811756298692126e-06,
+      "loss": 3.7737,
+      "step": 7514
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.786249246752638e-06,
+      "loss": 3.8659,
+      "step": 7515
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.7608274997095184e-06,
+      "loss": 3.6999,
+      "step": 7516
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.7354910619329942e-06,
+      "loss": 3.7371,
+      "step": 7517
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.710239937778803e-06,
+      "loss": 3.7817,
+      "step": 7518
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.685074131587862e-06,
+      "loss": 3.836,
+      "step": 7519
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.6599936476864326e-06,
+      "loss": 3.7683,
+      "step": 7520
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.634998490386121e-06,
+      "loss": 3.8609,
+      "step": 7521
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.61008866398399e-06,
+      "loss": 3.8622,
+      "step": 7522
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.5852641727622257e-06,
+      "loss": 3.7673,
+      "step": 7523
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.5605250209885253e-06,
+      "loss": 3.7684,
+      "step": 7524
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.5358712129157642e-06,
+      "loss": 3.8733,
+      "step": 7525
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.5113027527822192e-06,
+      "loss": 3.6781,
+      "step": 7526
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.4868196448115676e-06,
+      "loss": 3.8305,
+      "step": 7527
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.4624218932127194e-06,
+      "loss": 3.6524,
+      "step": 7528
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.43810950217982e-06,
+      "loss": 3.9056,
+      "step": 7529
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.4138824758925822e-06,
+      "loss": 3.7015,
+      "step": 7530
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.389740818515841e-06,
+      "loss": 3.827,
+      "step": 7531
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.3656845341998865e-06,
+      "loss": 3.7998,
+      "step": 7532
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.341713627080245e-06,
+      "loss": 3.7296,
+      "step": 7533
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.3178281012777866e-06,
+      "loss": 3.8126,
+      "step": 7534
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.2940279608986714e-06,
+      "loss": 3.8943,
+      "step": 7535
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.2703132100345146e-06,
+      "loss": 3.707,
+      "step": 7536
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.2466838527620557e-06,
+      "loss": 3.808,
+      "step": 7537
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.22313989314349e-06,
+      "loss": 3.7608,
+      "step": 7538
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.1996813352263566e-06,
+      "loss": 3.8998,
+      "step": 7539
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.176308183043375e-06,
+      "loss": 3.6227,
+      "step": 7540
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.1530204406127192e-06,
+      "loss": 3.7739,
+      "step": 7541
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.1298181119377432e-06,
+      "loss": 3.7271,
+      "step": 7542
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.1067012010073114e-06,
+      "loss": 3.8976,
+      "step": 7543
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.0836697117954115e-06,
+      "loss": 3.8876,
+      "step": 7544
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.0607236482613766e-06,
+      "loss": 3.7127,
+      "step": 7545
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.0378630143499953e-06,
+      "loss": 3.7169,
+      "step": 7546
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.0150878139912906e-06,
+      "loss": 3.6807,
+      "step": 7547
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.9923980511004646e-06,
+      "loss": 3.7933,
+      "step": 7548
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.9697937295782295e-06,
+      "loss": 3.8053,
+      "step": 7549
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.9472748533105886e-06,
+      "loss": 3.7282,
+      "step": 7550
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.924841426168667e-06,
+      "loss": 3.8072,
+      "step": 7551
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.9024934520091028e-06,
+      "loss": 3.7757,
+      "step": 7552
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.8802309346737666e-06,
+      "loss": 3.9791,
+      "step": 7553
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.858053877989819e-06,
+      "loss": 3.71,
+      "step": 7554
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.835962285769822e-06,
+      "loss": 3.7832,
+      "step": 7555
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.813956161811515e-06,
+      "loss": 3.7486,
+      "step": 7556
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.7920355098979833e-06,
+      "loss": 3.9229,
+      "step": 7557
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.7702003337977123e-06,
+      "loss": 3.7772,
+      "step": 7558
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.7484506372643658e-06,
+      "loss": 3.8624,
+      "step": 7559
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.7267864240369533e-06,
+      "loss": 3.904,
+      "step": 7560
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.7052076978398842e-06,
+      "loss": 3.7532,
+      "step": 7561
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.683714462382747e-06,
+      "loss": 3.8521,
+      "step": 7562
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.6623067213605303e-06,
+      "loss": 3.7711,
+      "step": 7563
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.6409844784533456e-06,
+      "loss": 3.8908,
+      "step": 7564
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.619747737326872e-06,
+      "loss": 3.8528,
+      "step": 7565
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.5985965016318e-06,
+      "loss": 3.8272,
+      "step": 7566
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.5775307750043887e-06,
+      "loss": 3.9015,
+      "step": 7567
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.556550561066073e-06,
+      "loss": 3.7148,
+      "step": 7568
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.5356558634235806e-06,
+      "loss": 3.776,
+      "step": 7569
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.514846685668981e-06,
+      "loss": 3.9646,
+      "step": 7570
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.4941230313795247e-06,
+      "loss": 3.8598,
+      "step": 7571
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.4734849041179176e-06,
+      "loss": 3.9061,
+      "step": 7572
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.4529323074320984e-06,
+      "loss": 3.7931,
+      "step": 7573
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.432465244855242e-06,
+      "loss": 3.7988,
+      "step": 7574
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.412083719905922e-06,
+      "loss": 3.7847,
+      "step": 7575
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.391787736087947e-06,
+      "loss": 3.6977,
+      "step": 7576
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.3715772968904147e-06,
+      "loss": 3.7698,
+      "step": 7577
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.3514524057877685e-06,
+      "loss": 3.9524,
+      "step": 7578
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.331413066239685e-06,
+      "loss": 3.6945,
+      "step": 7579
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.3114592816911863e-06,
+      "loss": 3.8335,
+      "step": 7580
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.2915910555725283e-06,
+      "loss": 3.8735,
+      "step": 7581
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.2718083912992572e-06,
+      "loss": 3.798,
+      "step": 7582
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.2521112922723184e-06,
+      "loss": 3.8566,
+      "step": 7583
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.2324997618778375e-06,
+      "loss": 3.7164,
+      "step": 7584
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.212973803487284e-06,
+      "loss": 3.7564,
+      "step": 7585
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.1935334204573054e-06,
+      "loss": 3.8156,
+      "step": 7586
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.1741786161300626e-06,
+      "loss": 3.6422,
+      "step": 7587
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.1549093938327825e-06,
+      "loss": 3.7722,
+      "step": 7588
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.135725756878093e-06,
+      "loss": 3.8121,
+      "step": 7589
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.1166277085639117e-06,
+      "loss": 3.8147,
+      "step": 7590
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0976152521733905e-06,
+      "loss": 3.7791,
+      "step": 7591
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0786883909749698e-06,
+      "loss": 3.8664,
+      "step": 7592
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.059847128222381e-06,
+      "loss": 3.7666,
+      "step": 7593
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0410914671546986e-06,
+      "loss": 3.756,
+      "step": 7594
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0224214109962334e-06,
+      "loss": 3.7615,
+      "step": 7595
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0038369629565846e-06,
+      "loss": 3.8374,
+      "step": 7596
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.98533812623064e-06,
+      "loss": 3.9657,
+      "step": 7597
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.9669249039985794e-06,
+      "loss": 3.8375,
+      "step": 7598
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.9485972994257585e-06,
+      "loss": 3.8245,
+      "step": 7599
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.930355315663046e-06,
+      "loss": 3.769,
+      "step": 7600
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.9121989558463225e-06,
+      "loss": 3.7793,
+      "step": 7601
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.8941282230969247e-06,
+      "loss": 3.8006,
+      "step": 7602
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.876143120521423e-06,
+      "loss": 3.8688,
+      "step": 7603
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.8582436512116775e-06,
+      "loss": 3.8853,
+      "step": 7604
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.840429818244782e-06,
+      "loss": 3.6199,
+      "step": 7605
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.8227016246831762e-06,
+      "loss": 3.7064,
+      "step": 7606
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.805059073574533e-06,
+      "loss": 3.868,
+      "step": 7607
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.7875021679518155e-06,
+      "loss": 3.7235,
+      "step": 7608
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.7700309108332203e-06,
+      "loss": 3.7386,
+      "step": 7609
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.7526453052223445e-06,
+      "loss": 3.7063,
+      "step": 7610
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.7353453541078535e-06,
+      "loss": 3.8818,
+      "step": 7611
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.7181310604639233e-06,
+      "loss": 3.8048,
+      "step": 7612
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.7010024272498537e-06,
+      "loss": 3.8505,
+      "step": 7613
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.683959457410178e-06,
+      "loss": 3.7808,
+      "step": 7614
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.6670021538748859e-06,
+      "loss": 3.834,
+      "step": 7615
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.65013051955909e-06,
+      "loss": 3.8053,
+      "step": 7616
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.6333445573632478e-06,
+      "loss": 3.7441,
+      "step": 7617
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.6166442701730511e-06,
+      "loss": 3.7804,
+      "step": 7618
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.6000296608594256e-06,
+      "loss": 3.755,
+      "step": 7619
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5835007322786421e-06,
+      "loss": 3.65,
+      "step": 7620
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.567057487272261e-06,
+      "loss": 3.8166,
+      "step": 7621
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5506999286669654e-06,
+      "loss": 3.7207,
+      "step": 7622
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.534428059274895e-06,
+      "loss": 3.7394,
+      "step": 7623
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5182418818933675e-06,
+      "loss": 3.8267,
+      "step": 7624
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.502141399304935e-06,
+      "loss": 3.8332,
+      "step": 7625
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.4861266142775499e-06,
+      "loss": 3.9294,
+      "step": 7626
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.4701975295641768e-06,
+      "loss": 3.7985,
+      "step": 7627
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.4543541479033473e-06,
+      "loss": 3.7234,
+      "step": 7628
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.438596472018716e-06,
+      "loss": 3.7891,
+      "step": 7629
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.422924504619061e-06,
+      "loss": 3.6453,
+      "step": 7630
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.4073382483987817e-06,
+      "loss": 3.7798,
+      "step": 7631
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.39183770603718e-06,
+      "loss": 3.8554,
+      "step": 7632
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3764228801990687e-06,
+      "loss": 3.7339,
+      "step": 7633
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3610937735344386e-06,
+      "loss": 3.7066,
+      "step": 7634
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3458503886784602e-06,
+      "loss": 3.7909,
+      "step": 7635
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3306927282517034e-06,
+      "loss": 3.7306,
+      "step": 7636
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3156207948599176e-06,
+      "loss": 3.8336,
+      "step": 7637
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3006345910941963e-06,
+      "loss": 3.8931,
+      "step": 7638
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.2857341195308125e-06,
+      "loss": 3.7744,
+      "step": 7639
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.2709193827312725e-06,
+      "loss": 3.7241,
+      "step": 7640
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.2561903832424836e-06,
+      "loss": 3.7443,
+      "step": 7641
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.2415471235965315e-06,
+      "loss": 3.8938,
+      "step": 7642
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.2269896063107356e-06,
+      "loss": 3.8337,
+      "step": 7643
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.2125178338876497e-06,
+      "loss": 3.9557,
+      "step": 7644
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.1981318088152281e-06,
+      "loss": 3.731,
+      "step": 7645
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.1838315335664928e-06,
+      "loss": 3.7791,
+      "step": 7646
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.169617010599977e-06,
+      "loss": 3.784,
+      "step": 7647
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.1554882423591707e-06,
+      "loss": 3.8408,
+      "step": 7648
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.141445231273075e-06,
+      "loss": 3.8124,
+      "step": 7649
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.1274879797558146e-06,
+      "loss": 3.766,
+      "step": 7650
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.113616490206748e-06,
+      "loss": 3.932,
+      "step": 7651
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0998307650106342e-06,
+      "loss": 3.9011,
+      "step": 7652
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0861308065373554e-06,
+      "loss": 3.8283,
+      "step": 7653
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0725166171420831e-06,
+      "loss": 3.7563,
+      "step": 7654
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0589881991652784e-06,
+      "loss": 3.8447,
+      "step": 7655
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0455455549326365e-06,
+      "loss": 3.6741,
+      "step": 7656
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0321886867550868e-06,
+      "loss": 3.7776,
+      "step": 7657
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0189175969287924e-06,
+      "loss": 3.7999,
+      "step": 7658
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0057322877352615e-06,
+      "loss": 3.9561,
+      "step": 7659
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.926327614411811e-07,
+      "loss": 3.8071,
+      "step": 7660
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.796190202985832e-07,
+      "loss": 3.8215,
+      "step": 7661
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.666910665445672e-07,
+      "loss": 3.7942,
+      "step": 7662
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.538489024016328e-07,
+      "loss": 3.7136,
+      "step": 7663
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.410925300775697e-07,
+      "loss": 3.9209,
+      "step": 7664
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.284219517652349e-07,
+      "loss": 3.9044,
+      "step": 7665
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.158371696428858e-07,
+      "loss": 3.6707,
+      "step": 7666
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.033381858740697e-07,
+      "loss": 3.8244,
+      "step": 7667
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 8.909250026074012e-07,
+      "loss": 3.5668,
+      "step": 7668
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 8.785976219768954e-07,
+      "loss": 3.9063,
+      "step": 7669
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 8.663560461018016e-07,
+      "loss": 3.8861,
+      "step": 7670
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 8.542002770865476e-07,
+      "loss": 3.9254,
+      "step": 7671
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 8.421303170208505e-07,
+      "loss": 3.6429,
+      "step": 7672
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 8.301461679796618e-07,
+      "loss": 3.8213,
+      "step": 7673
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 8.182478320232223e-07,
+      "loss": 3.8696,
+      "step": 7674
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 8.064353111970068e-07,
+      "loss": 3.8045,
+      "step": 7675
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 7.947086075317245e-07,
+      "loss": 3.7476,
+      "step": 7676
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 7.830677230433181e-07,
+      "loss": 3.7091,
+      "step": 7677
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 7.715126597329647e-07,
+      "loss": 3.6954,
+      "step": 7678
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 7.600434195871864e-07,
+      "loss": 3.8402,
+      "step": 7679
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 7.486600045775727e-07,
+      "loss": 3.7934,
+      "step": 7680
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 7.373624166611137e-07,
+      "loss": 3.786,
+      "step": 7681
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 7.261506577800892e-07,
+      "loss": 3.8803,
+      "step": 7682
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 7.150247298617907e-07,
+      "loss": 3.9515,
+      "step": 7683
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 7.039846348189105e-07,
+      "loss": 3.8985,
+      "step": 7684
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 6.930303745494859e-07,
+      "loss": 3.8305,
+      "step": 7685
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 6.821619509365662e-07,
+      "loss": 3.7625,
+      "step": 7686
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 6.713793658486012e-07,
+      "loss": 3.6646,
+      "step": 7687
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 6.606826211392192e-07,
+      "loss": 3.8818,
+      "step": 7688
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 6.500717186473382e-07,
+      "loss": 3.8104,
+      "step": 7689
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 6.3954666019711e-07,
+      "loss": 3.8176,
+      "step": 7690
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 6.291074475978653e-07,
+      "loss": 3.6408,
+      "step": 7691
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 6.187540826442795e-07,
+      "loss": 3.6814,
+      "step": 7692
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 6.084865671162065e-07,
+      "loss": 3.7802,
+      "step": 7693
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 5.983049027786791e-07,
+      "loss": 3.8925,
+      "step": 7694
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 5.882090913821303e-07,
+      "loss": 3.8775,
+      "step": 7695
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.781991346621163e-07,
+      "loss": 3.8995,
+      "step": 7696
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.682750343394826e-07,
+      "loss": 3.8219,
+      "step": 7697
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.584367921202538e-07,
+      "loss": 3.6562,
+      "step": 7698
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.486844096957988e-07,
+      "loss": 3.8817,
+      "step": 7699
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.3901788874261e-07,
+      "loss": 3.7865,
+      "step": 7700
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.294372309224693e-07,
+      "loss": 3.7197,
+      "step": 7701
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.19942437882448e-07,
+      "loss": 3.7708,
+      "step": 7702
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.10533511254796e-07,
+      "loss": 3.7334,
+      "step": 7703
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.012104526569972e-07,
+      "loss": 3.9574,
+      "step": 7704
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.919732636918251e-07,
+      "loss": 3.8313,
+      "step": 7705
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.828219459472316e-07,
+      "loss": 3.9146,
+      "step": 7706
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.737565009964584e-07,
+      "loss": 3.8117,
+      "step": 7707
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.647769303979255e-07,
+      "loss": 3.8225,
+      "step": 7708
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.558832356953979e-07,
+      "loss": 3.7779,
+      "step": 7709
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.470754184177084e-07,
+      "loss": 3.9279,
+      "step": 7710
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.3835348007909004e-07,
+      "loss": 3.9457,
+      "step": 7711
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.297174221788991e-07,
+      "loss": 3.6858,
+      "step": 7712
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.2116724620183675e-07,
+      "loss": 3.7518,
+      "step": 7713
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.1270295361772737e-07,
+      "loss": 3.6912,
+      "step": 7714
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.043245458816847e-07,
+      "loss": 3.8463,
+      "step": 7715
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.960320244340565e-07,
+      "loss": 3.7594,
+      "step": 7716
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.878253907004248e-07,
+      "loss": 3.7143,
+      "step": 7717
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.7970464609160535e-07,
+      "loss": 3.8929,
+      "step": 7718
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.7166979200364826e-07,
+      "loss": 3.7786,
+      "step": 7719
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.637208298178929e-07,
+      "loss": 3.7728,
+      "step": 7720
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.558577609007463e-07,
+      "loss": 3.9144,
+      "step": 7721
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.480805866040715e-07,
+      "loss": 3.7725,
+      "step": 7722
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.403893082647991e-07,
+      "loss": 3.9101,
+      "step": 7723
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.3278392720514915e-07,
+      "loss": 3.8513,
+      "step": 7724
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.252644447325759e-07,
+      "loss": 3.7434,
+      "step": 7725
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.178308621397674e-07,
+      "loss": 3.7953,
+      "step": 7726
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.10483180704646e-07,
+      "loss": 3.8858,
+      "step": 7727
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.0322140169036784e-07,
+      "loss": 3.8532,
+      "step": 7728
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.9604552634532324e-07,
+      "loss": 3.8127,
+      "step": 7729
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.88955555903081e-07,
+      "loss": 3.7397,
+      "step": 7730
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.8195149158255497e-07,
+      "loss": 3.7431,
+      "step": 7731
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.75033334587782e-07,
+      "loss": 3.8036,
+      "step": 7732
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.6820108610808855e-07,
+      "loss": 3.9119,
+      "step": 7733
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.614547473180351e-07,
+      "loss": 3.8537,
+      "step": 7734
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.5479431937736055e-07,
+      "loss": 3.8056,
+      "step": 7735
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.482198034310934e-07,
+      "loss": 3.7682,
+      "step": 7736
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.4173120060949626e-07,
+      "loss": 3.8479,
+      "step": 7737
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.353285120279547e-07,
+      "loss": 3.7694,
+      "step": 7738
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.2901173878725478e-07,
+      "loss": 3.838,
+      "step": 7739
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.2278088197325018e-07,
+      "loss": 3.8115,
+      "step": 7740
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.1663594265708407e-07,
+      "loss": 3.8488,
+      "step": 7741
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.1057692189524468e-07,
+      "loss": 3.812,
+      "step": 7742
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0460382072928775e-07,
+      "loss": 3.8045,
+      "step": 7743
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.98716640186003e-07,
+      "loss": 3.8257,
+      "step": 7744
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.9291538127758078e-07,
+      "loss": 3.7818,
+      "step": 7745
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.8720004500122345e-07,
+      "loss": 3.8224,
+      "step": 7746
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.8157063233953387e-07,
+      "loss": 3.869,
+      "step": 7747
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.7602714426023791e-07,
+      "loss": 3.7593,
+      "step": 7748
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.7056958171635108e-07,
+      "loss": 3.7782,
+      "step": 7749
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.6519794564606727e-07,
+      "loss": 3.8135,
+      "step": 7750
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5991223697281455e-07,
+      "loss": 3.7846,
+      "step": 7751
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5471245660531042e-07,
+      "loss": 3.8014,
+      "step": 7752
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.4959860543739546e-07,
+      "loss": 3.7039,
+      "step": 7753
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.4457068434831077e-07,
+      "loss": 3.8461,
+      "step": 7754
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.396286942023095e-07,
+      "loss": 3.7029,
+      "step": 7755
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.3477263584904532e-07,
+      "loss": 3.7217,
+      "step": 7756
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.3000251012323938e-07,
+      "loss": 3.9009,
+      "step": 7757
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.2531831784506897e-07,
+      "loss": 3.716,
+      "step": 7758
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.2072005981966781e-07,
+      "loss": 3.7034,
+      "step": 7759
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.1620773683757024e-07,
+      "loss": 3.7284,
+      "step": 7760
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.1178134967454457e-07,
+      "loss": 3.9274,
+      "step": 7761
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0744089909153765e-07,
+      "loss": 3.9248,
+      "step": 7762
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0318638583467488e-07,
+      "loss": 3.6891,
+      "step": 7763
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.901781063531568e-08,
+      "loss": 3.832,
+      "step": 7764
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.493517421022002e-08,
+      "loss": 3.8876,
+      "step": 7765
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.093847726110438e-08,
+      "loss": 3.781,
+      "step": 7766
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.702772047514129e-08,
+      "loss": 3.8359,
+      "step": 7767
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.320290452462631e-08,
+      "loss": 3.7869,
+      "step": 7768
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.94640300670335e-08,
+      "loss": 3.8769,
+      "step": 7769
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.581109774512652e-08,
+      "loss": 3.7974,
+      "step": 7770
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.224410818695848e-08,
+      "loss": 3.7788,
+      "step": 7771
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.876306200565009e-08,
+      "loss": 3.7975,
+      "step": 7772
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.536795979966703e-08,
+      "loss": 3.6961,
+      "step": 7773
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 6.205880215270909e-08,
+      "loss": 3.7283,
+      "step": 7774
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.883558963359903e-08,
+      "loss": 3.7252,
+      "step": 7775
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.569832279644915e-08,
+      "loss": 3.7272,
+      "step": 7776
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.264700218066132e-08,
+      "loss": 3.7356,
+      "step": 7777
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.968162831070488e-08,
+      "loss": 3.7234,
+      "step": 7778
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.6802201696394264e-08,
+      "loss": 3.8058,
+      "step": 7779
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.400872283272239e-08,
+      "loss": 3.8445,
+      "step": 7780
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.130119219997175e-08,
+      "loss": 3.7437,
+      "step": 7781
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.8679610263547824e-08,
+      "loss": 3.7353,
+      "step": 7782
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.614397747420117e-08,
+      "loss": 3.7476,
+      "step": 7783
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.3694294267694326e-08,
+      "loss": 3.7613,
+      "step": 7784
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.133056106530141e-08,
+      "loss": 3.8099,
+      "step": 7785
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.9052778273308545e-08,
+      "loss": 3.8947,
+      "step": 7786
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.6860946283291388e-08,
+      "loss": 3.7846,
+      "step": 7787
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.4755065472115144e-08,
+      "loss": 3.8424,
+      "step": 7788
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.2735136201712524e-08,
+      "loss": 3.8266,
+      "step": 7789
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.0801158819361287e-08,
+      "loss": 3.7285,
+      "step": 7790
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.895313365757323e-08,
+      "loss": 3.881,
+      "step": 7791
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.7191061033983158e-08,
+      "loss": 3.688,
+      "step": 7792
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5514941251570934e-08,
+      "loss": 3.8459,
+      "step": 7793
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.3924774598439438e-08,
+      "loss": 3.8456,
+      "step": 7794
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.2420561348036597e-08,
+      "loss": 3.7987,
+      "step": 7795
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.100230175882233e-08,
+      "loss": 3.9569,
+      "step": 7796
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.669996074712639e-09,
+      "loss": 3.707,
+      "step": 7797
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 8.423644524691021e-09,
+      "loss": 3.7279,
+      "step": 7798
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 7.263247323086031e-09,
+      "loss": 3.7517,
+      "step": 7799
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 6.188804669349235e-09,
+      "loss": 3.6987,
+      "step": 7800
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.200316748110723e-09,
+      "loss": 3.8425,
+      "step": 7801
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.297783729456661e-09,
+      "loss": 3.8368,
+      "step": 7802
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.4812057684296962e-09,
+      "loss": 3.8356,
+      "step": 7803
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.7505830054175284e-09,
+      "loss": 3.8234,
+      "step": 7804
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.1059155660418936e-09,
+      "loss": 3.6983,
+      "step": 7805
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.547203561158561e-09,
+      "loss": 3.8074,
+      "step": 7806
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.074447086801822e-09,
+      "loss": 3.7735,
+      "step": 7807
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 6.876462241844905e-10,
+      "loss": 3.9458,
+      "step": 7808
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.8680103991994844e-10,
+      "loss": 3.6646,
+      "step": 7809
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.7191158563356623e-10,
+      "loss": 3.7613,
+      "step": 7810
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.297789824025955e-11,
+      "loss": 3.766,
+      "step": 7811
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0,
+      "loss": 3.8903,
+      "step": 7812
+    },
+    {
+      "epoch": 1.0,
+      "step": 7812,
+      "total_flos": 8.153166935752704e+17,
+      "train_loss": 3.959441412978458,
+      "train_runtime": 22134.6437,
+      "train_samples_per_second": 90.356,
+      "train_steps_per_second": 0.353
+    }
+  ],
+  "logging_steps": 1.0,
+  "max_steps": 7812,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 1,
+  "save_steps": 24000,
+  "total_flos": 8.153166935752704e+17,
+  "train_batch_size": 8,
+  "trial_name": null,
+  "trial_params": null
+}